Production Data Management of Smart Farming Based on Shili Theory

Abstract

The development of smart farming comes with a lot of data problems. Studies have shown this is due to insufficient cognition of the structural relationship between data and events. Shili Theory is an attractive concept. To embed intelligent agricultural technology in events and the natural environment, especially to unify and standardize agricultural production data, firstly, this paper has defined the concept of Shili Theory which researches the natural regularity of the event by Shili Mirrored Structure. Secondly, this paper has proposed a Shili Mirrored Structure based on the technology development path (from the human brain memory mechanism to the information storage mechanism to intelligent technology). Finally, the structure has been applied to develop an intelligent system of agricultural production data management. In rice production of Jilin Province, it forms the event chain of the whole plant 5T (seed, seeding, paddy shoot, grain, product period operation) and grain period 5T (harvesting, field stacking, drying, warehousing, storing). The system application shows that this management structure can reduce data flow, improve data utilization, and enhance the correlation between data and events. It can realize the quality improvement of the agricultural production process, especially revealing the 8.83% significant latent loss in rice harvest.

1. Introduction

In the third decade of the 21st century, the fourth industrial revolution, represented by subversive technology such as big data analysis, the Internet of Things, artificial intelligence, and intelligent equipment, emerged worldwide. This provides a new development path for breaking the dilemma of current agricultural production. By 2050, the world’s population is forecasted to grow to 9.1 billion, and the current agricultural production level will be insufficient to meet the food demand of the growing population [1,2,3]. According to traditional agricultural management methods and production modes, increasing cultivated land for mass production will lead to ecological and social unsustainability. Driven by the subversive technologies of the fourth industrial revolution, agricultural production is changing fast towards smart farming [4,5,6,7,8]. People use modern digital technologies to solve complex agro-ecosystem problems. In this context, sustainable and intelligent industrial agriculture would be achieved through real-time variable fine-grained collection, processing, and analyzing spatio-temporal data in all aspects of the agricultural industry [6]. This produces massive data at unprecedented speeds [9].

However, leveraging big data in agriculture is challenging. The agricultural system is a complex system composed of the farming economy, technology, ecology, and rural society [10]. Integration of information sources by segments of the livestock industry has been poor or non-existent [11]. The excess of data may cause vital information to be masked by noise [12]. Meanwhile, agricultural data are collected from individuals, research groups, and companies using different types of devices, which causes multisource and heterogeneity problems [13]. Faced with massive and growing agricultural data, Wu et al. [14] believe the key is to track the evolution of events and analyze their underlying events or themes. Balafoutis et al. [15] and Verdouw et al. [16] emphasized that intelligence agriculture was based not only on accurate location data but also on context information, environmental perception, and event triggers. Existing methods ignore the interaction between events, potential events, and data. To solve the problems of massive data management, multiple object recognition standards, and multisource data fusion, Li et al. [17] simplify and unify data management into a business event model. He divides business data into five categories of events: IoT events, object events, business events, inspection events, and perception events. Unlike other socio-economic systems, the biological characteristics of agricultural production determine the regularity between agricultural systems and natural environments, which leads to events that happen sequentially. Some scholars [18,19,20,21] in Western society suggest that agricultural management is required to be better embedded in the natural environment, achieving greater integration between social and biological fields [22,23,24]. Therefore, the data management of agricultural systems is increasingly emphasizing the embeddedness of intelligent agricultural technology and constructing event management based on crop-producing regularity. The mirrored structure in Shili Theory is an attractive metaphor to characterize this development.

Although predecessors have not clearly proposed the Shili structure -as will be dealt with later on, Shili Theory is fundamentally researching the basic regularity of the event. Objects have physics; events have Shili; all things have a certain regularity [25]. This is the essential discussion of the Shili Theory proposed. It studies the internal factor, environmental factors, and interaction regularity of events with systematic thinking. This regularity is reflected in the natural regularity of crop growth in agricultural systems engineering. Using Shili Theory for strategic planning of agricultural data management, people organize agronomic processes according to the natural regularity of crop growth. It is in line with the viewpoint of embedding agricultural management into the natural environment. The concept of “Shil” has been widely concerned since 1981, and scholars have tried to construct an oriental system methodology in accordance with the Chinese context from their respective research [26,27,28]. However, the Chinese academic community has had little reaction in the subsequent decades, and agricultural management research is still developing in the original direction (industrial chain research). For this situation, Song [29] mentioned in the “Shili system engineering and database technology” that nerve cells’ equivalent storage capacity in the human brain was limited, only a medium database. It limited the event processing capability of the Shili system and lacked corresponding data management and technical support. In the research on precision livestock farming and the natural environment, Blok and Gremmen [30] also emphasized the role of computer processing capabilities and sensor technology. By 2021, the fourth industrial revolution will emerge, which provides methodological and technical possibilities to conduct comprehensive integration management for the agricultural system based on Shili Theory. It promotes Shili Theory from information Shili to smart Shili, thereby promoting the construction of intelligent agriculture.

This paper aims to analyze how Shili Theory promotes smart farming, especially how to unify and standardize agricultural production data. At the same time, it studies the application mechanism of the Shili Mirrored Structure in smart farming to improve the embeddedness of technologies into events and the natural environment. More specifically, the objectives are threefold: 1. To define the concept of Shili Theory and explain its research in agricultural systems; 2. To propose a conceptual framework, i.e., Shili Mirrored Structure, based on the technology development path; 3. To apply and validate Shili Mirrored Structure to agricultural production data management system in the case study of 5T management.

2. Concepts

2.1. What Is Shili Theory?

The study of Shili Theory originates from the field of system engineering. With respect to this, it is necessary to recall the development of the agricultural system. From the viewpoint of systematology, the agricultural system is a complex system composed of the agricultural economy, technology, ecological, and rural society [10], with a multi-target structure, multi-factor correlation, and random uncertainty. It grasps the core characteristics of agricultural production when conducting comprehensive integration management for agricultural systems. Unlike other socio-economic systems, the biological characteristics of agricultural production determine the regularity between agricultural systems and natural environments. Therefore, agricultural system research should grasp the process of agricultural production and the regularity of nature. Studies have shown that the regularity disclosed by known systems can be used to understand and reveal the regularity of unknown complex systems [31,32]. In 1981, on the commonality of system engineering, Xu [25] suggested that all things had a certain regularity and objects had physics and events had Shili, first proposing the concept of Shili from the theoretical level. Qian et al. [33] affirmed the view of Xu Guozhi in “organizational management technology—system engineering” and enlisted the “Shili General Theory” and “Theoretical Shili Theory” from the similarity of complex systems [34]. According to Qian Xuesen’s technical framework, Miao [35] believed that “all events had a certain regularity … moreover all the truth, principles, regularities and methods of doing things were considered as Shil”. Thus, Shili Theory was proposed based on the introduction of the Shili concept. At present, the development regularity of agricultural production has not yet been clarified. The study of the agricultural production development process and its change regularities have not yet been established [36]. Based on Qian XueSen’s Shili study, Zhang and Qian [36] extracted Agricultural Shili Theory, researching the regularity in agricultural production. Proceeding from agricultural systems engineering, Agricultural Shili Theory is used to conduct comprehensive integration management for the regularity of agricultural production. It overcomes the drawbacks of considering the problem only from a single department, single goal, or single factor.

In essence, the evolution and development of events have their underlying regularities, leading to events that happen sequentially [37,38]. Shili Theory, which takes the basic regularity of the event as the core, associates the event according to the natural regularity of occurrence to form an event chain. Shili Mirrored Structure can conceptualize complex system management into the entitative structure and virtual structure connected by human brain memory, information storage, or intelligent technology. To better strategic planning of agricultural data management. Hence, although Shili Theory originates from systematology, agricultural systems management achieved through it must rely on information and intelligent technology. With the development of information technology, the characterization and application of Shili Mirrored Structure are also evolving.

Summing up the above analysis, the so-called Shili Theory researches the basic regularity of the event by Shili Mirrored Structure. Based on the natural characteristics of crop growth and deposit, Shili Theory shows the basic regularity of agricultural production events in smart farming. It uses the Shili Mirrored Structure to unify agricultural data and forms a networked data management mechanism, relying on the Internet of Things, big data, artificial intelligence, and other technologies. Within the perspective of holism and systematology, anything is a system [39].

At present, there is more research on event-based control in the field of automation control [40,41,42]. The basic idea is to use events as the input source to drive control and cause control behavior of the control system [43]. There are certain associations and differentiations between this and the study of Shili Theory. First, the Shili Theory is to use Shili Mirrored Structure for data management. The main purpose is to unify and standardize agricultural production data and improve data utilization. The event-triggered mechanism is used to release data through predefined status conditions. The main purpose is to reduce the packet delivery rate and improve the utilization of system and network resources [44]. Both reduce data redundancy to some extent. Second, the event of Shili Theory emphasizes the systematic properties, including real-world production events and the corresponding virtual-world full-parameter event. An event in the field of dynamic control refers to an object, an activity, or an action that causes a change in its state [45], emphasizing the verb and its properties [46]. This is closer to the key elements or parameters of the event in Shili Theory, such as event nature code, subject code, time code, etc. Third, the Shili Theory is based on the basic regularity of the event for data management. It manifests to determine the event boundary based on the natural block characteristics of the crop whole growth and deposit, building event chains. Various bases exist in event-based control studies, including less or easily-tunable trigger parameters, network bandwidth-aware, data significance, etc [47]. For example, Xianfeng Tan [43] determines the order of model expansion based on two factors: the importance of the event and the frequency of occurrence, so as to achieve the goal of real-time response to important events. Clearly, in using Shili Mirrored Structure for agricultural production management and control, event-based dynamic control to release data is necessary.

2.2. Shili Theory Structure

Since the 1960s, many scholars have discussed system engineering at different levels. The system engineering method appeared earlier, and the most significant impact is the three-dimensional structure of the Hall system project proposed by American scholar Hall. Hall [48] divided the system into three primary dimensions—logical, professional, and time by investigating system engineering projects. Since 1979, Qian and Tsien [49] have begun to introduce some new fields of system science theory from abroad [50]. At the time of discussing system engineering, Qian Xuesen and other scholars emphasized the events and the Shili as scientific concepts. They took the factory as an example to analyze three elements of people, objects, and events that constitute such systems. Further, objects were divided into substances, equipment, and wealth three subclasses. Events included two elements: task (task issued by the superior or contract booked in other units) and information (data, drawings, forms, rules, decisions, and the like) [33]. On account of this, Xu Guozhi put forward the concept of event process and formed a unified concept group of material flow, event process, and information stream. However, Miao [35] believed that the understanding of the Shili lacked sure accuracy in the past. Things were not limited to tasks and information, and everything in the project came from people, as events arose because of people. The flow of people essentially belonged to the event process, and the information stream permeated in matter flow and event process. He specifically pointed out that all Shili systems consisted of events, people, and information. In 1995, Gu and Zhu [51] proposed the WSR system methodology. From exploring regularity, they divided system practice into Wuli, Shili, and Renli. So far, there is a certain divergence in exploring the Shili, and Shili Theory has not yet formed its conceptual framework.

For the existing research of system science, this paper emphasizes the systematic properties of the Shili itself. Focusing on the basic regularity of events, Shili Theory divides the system into event type, event subject, and event entry, which take events as research objects, according to the two phases of complex system management. Firstly, people use the sensory to observe external contact and surface characteristics of “complex management”. Secondly, people use abstract thinking to understand the internal contact and essential regularity of “complex management”. The second phase needs to be completed by concept, judgment, and reasoning and takes intuitive physical complexity to form a “mirror” in a complex system space [52]. Therefore, this paper proposes a three-dimensional mirrored structure of Shili Theory (Figure 1). As shown, the solid line identifies the entitative structure, and the dashed line identifies the virtual structure. There is a corresponding mirrored relationship between the two. Founded on the research of Miao Dongsheng, the entitative structure includes event type, event subject, and event entry. Event type refers to the nature of the event and the semantic relationship between the objects. Event subject refers to all entities, such as stakeholders and related assets, during the event. Event entry refers to time, space, quantity, and other information involved in the event. Based on the Hall structure, the virtual structure contains event logic, event method, and event situation.

As yet, there is less research on the mirrored structure in the systematics field, but there is some research on the product life cycle. In 2005, Grieves proposed the Mirrored Space Model (real space, virtual space, and connection mechanism) [53], the predecessor of digital twins. In order to eliminate functional shafts and de-silo the isolated information, Grieves organized data not by its function but by the physical object with which it was associated. Unlike taking the physical object as the data core, the mirrored structure of Shili Theory emphasizes the importance of the event. It centers on the regularity of the event itself.

2.3. Shili Theory in Agricultural Management

As mentioned above, the fourth industrial revolution provided a new path for breaking agricultural production difficulties, promoting agricultural production to smart farming. Nevertheless, the application of intelligent agricultural technology also brings a lot of data problems, arising scholars’ research on the structural relationship between data and events. The agricultural system is a highly dynamic, complex system that directly acts on life [54]. Gianni believes that agriculture has been more restricted by natural conditions and objective life regularity compared to other industries. Paying attention to agriculture must value natural regularity and respect farmers so that farmers can obtain higher income through agricultural production [55]. A mere focus on technology would reinforce unequal modes of capitalist production [56]. On the one hand, to unify and standardize agricultural production data and protect the interests of all parties. On the other hand, to improve the embeddedness of intelligent agriculture technology into events and the natural environment. This paper uses Shili Theory and mirrored structure to conduct comprehensive integration management for agricultural production regularity. It is explicitly shown in the following three features:

First, this paper takes the event as an object, not targeting an object or human. This manifests in taking an agricultural production event as a management object. Objects, subjects, assets, and information are reflected in the event. It is not limited to a few interest groups but covers all subjects in the event (i.e., participants in the agricultural system), assets, and other information to ensure the comprehensiveness of information traceability. Farmers’ interests can be effectively protected. Second, this paper takes the basic regularity of the event as the core [57]. It manifests to determine the event boundary based on the natural block characteristics of the crop whole growth and deposit, building event chains. In accordance with the basic regularity of agricultural production events, agricultural data management is conducted. Third, this paper emphasizes the natural attributes. This manifests as research from the natural characteristics of crops. It is not taking a private return on investment as a primary motive but to serve broader public goods as much as possible. This furthest reduces the destructive effects of the natural environment during the agronomic process [58].

2.4. The Ideology of Shili Theory in 5T Management

5T management is the actual application of Shili Mirrored Structure in the practical application of agricultural production data management. Its development and application have the basic ideology of Shili Theory, specifically manifested as the following three principles.

The first is the natural block principle. In the growth and deposit of rice cells, organs, grains, and even plants, dry matter accumulation and quality formation are a non-regular S-type curve [59]. According to the time-sensitive point (e.g., the best harvest time), the associated time of related curves can be divided into multiple natural blocks (Figure 2), including seed period (T_I), seedling period (T_II), paddy shoot period (T_III), grain period (T_IV), and product period (T_Ⅴ). Each natural block reflects the growth characteristics of rice in different phases. The target factors throughout rice production (e.g., moisture content, dry matter, cracking rate) are selected to set controllable factors in different events. This is a plant’s physiological and physical foundation for implementing event management to agricultural data.

The second is the wet heat stress control principle. Rice grains can quickly produce micro-cracks under the temperature difference between morning and evening, moisture condensation, and rainfall. The severe situation will generate the crack, causing rice aroma loss, while the cracks’ leading cause is the hygroscopic effect of low moisture grains. According to the division of production events in the grain period (T_IV), it will cause rice cracks that delay harvesting (T₁), raining field stacking(T₂), high-temperature drying (T₃), longtime warehousing(T₄), high-temperature storing (T₅). Figure 3 shows the microstructural comparison between the newly harvested rice and rice stored for one year. The rice stored for one year has obvious micro-cracks. 5T management reasonably controls the accuracy of each event by related index. It limits the occurrence of rice grain micro-cracks and cracks, thereby maintaining a fresh and guaranteed fragrance.

The third is the taste and quality control principle. The taste and quality of rice are affected by many factors: 1. The later the harvest (T₁) time is, the lower the rice gelatinization degree will be, and the faster the taste deterioration will be. 2. The higher the drying (T₃) temperature, the more serious the germ damage and the more significant the grain quality drops. 3. The high cracking rate of grain period (T_Ⅳ) makes it easier to produce water cracks when cooking, making the rice taste bad.

Based on the above analysis, the conceptual framework is developed for managing smart farming by using Shili Theory and mirrored structure.

3. Materials and Methods

3.1. Conceptual Framework

The development of the conceptual framework is based on the paradigm of design science. With a specific prescription as a drive, it provides solutions for management issues by constructing new artifacts [60,61]. The constructed artifact in this paper is a conceptual framework of mirrored structural design and implementation in Shili Theory. The purpose is to provide intelligent solutions for agricultural data problems. This includes the management aspects of using the Shili Mirrored Structure to plan, supervise, control, retrace, evaluate, and predict, improving the utilization of agricultural data. As early as the 1980s, the study of Shili has begun. It has a far-reaching impact on China’s agriculture, industry, military, environment, and resources [62,63]. However, being subject to the limitations of the former technical level, the research on Shili Mirrored Structure is not clear. With the development of information technology, the fourth industrial revolution has emerged. Emerging technology provides methodological and technical possibilities for the application development of the Shili Mirrored Structure. As per the previous chapters have studied, the following section identifies three requirements:

• The conceptual framework organizes data around events rather than functions or physical objects, emphasizing the correlation between data and events.
• The conceptual framework takes the event’s regularity as the core, associating events with the natural regularity that occurs to form an event chain. The virtual structure corresponding to the entity structure is constructed based on critical elements.
• The conceptual framework evolves with the development of technological innovation. It is necessary to analyze Shili Mirrored Structure under different stages (human brain memory mechanism, information storage mechanism, intelligent technology).

3.1.1. Shili Mirrored Structure based on the Human Brain Memory Mechanism

The virtual structure is both ancient and novel. Said that it is ancient because, as a human, people have already owned a digital bit-based virtual space for a long time. This is as long as humans have self-awareness (human brain memory mechanism) [64]. Before computer technology did not appear, farmers worked on crop production based on their experience and cognition. They conduct agricultural management according to the natural regularity of crops, which is the embodiment of the human brain memory mechanism. In this context, a three-dimensional mirrored structure of Shili Theory is manifested as the real world and ideological world connected through the human brain memory mechanism. All stakeholders can participate autonomously. As shown in Figure 4, a single event meta is used as an expansion object. People determine the event nature of the real world based on their cognition and abstract it into the logical dimension of event logic in the ideological world through human brain memory. Similarly, people clarify the relevant subjects and assets of the event and abstract them into the professional dimension of the event management method. People classify the time, space, quantity, and other information involved in the event and abstract them into the time dimension of the event situation. On these grounds, people have achieved effective management of events through the human brain memory mechanism.

It can be seen that without computer and artificial intelligence technology, people want to realize the “mirroring” of the physical complexity of the real world in the system space. This requires converting the theoretical logic and discourse system of complex problems, using complexity thinking to cognize, analyze, and solve problems. Whereas this connection also has a substantial defect. First, the insufficiency of fidelity and the long-lastingness of the human brain will let us forget a lot of expected memories [64]. It is estimated that the total number of nerve cells in the human brain is about 1010. As for how much equivalent storage capacity of each nerve cell, there is no reliable experimental data. According to the McCulloch-Pitts neural network model [65], each neuron can only remember a binary bit of information. Even if every brain cell can store 100 binary information, the human brain is only equivalent to a medium database [29]. Second, human brain memory cannot be directly shared [66]. The information exchange network is not smooth. Hence, the emergence of computer technology promotes the development of Mirrored Structure to a certain extent.

3.1.2. Shili Mirrored Structure based on the Information Storage Mechanism

In the 1980s, with the development of computer technology, Song [29] proposed a Shili System Engineering. He believed that the Shili could be regarded as a system that played an increasing role in the various activities of human society, which was inseparable from the emergence and development of modern computer technology. Before 1960, computer technology was still not enough to provide adequate speed and storage space. The ideology and theory of system engineering could not become a realistic project and could not have such a significant social effect. So to speak, modern computer technology and database technology are the premises of the socialization of Shili System Engineering.

Based on the elements of the entitative and virtual structure (three-dimensional mirrored structure), this paper refers to the previously studied patent “One Event Contact Traceable Coding Method”. The mirrored structure utilizes the event evolutionary graph, database technology, and computer technology, converting the event meta of the real world to the shared, massive-stored, and callable data meta. Among them (Figure 5), the event type, corresponding to the event logic, can be identified as an event nature code consisting of related strings. This identifies a series of events developed following basic regularity. The event subject, corresponding to the event method, can be identified as an event subject code and an event asset code. This reflects the degree of participation and asset ownership of all relevant people during the event. The event entry, corresponding to the event situation, can be identified as the event location code, time code, and quantity code. This reflects the operating specifications and crop data in the event. On these grounds, people have achieved effective control of single event meta through the human brain memory mechanism.

At the same time, the evolution and development of the event have its own basic regularity, which causes the event in order. The accuracy of abstracting the event meta into the data meta is closely related to the scene of the event. If there is no context scenario, a single event may become too abstract, and people cannot understand it [38]. This needs to analyze event regularity based on past research and expert opinions. A crucial aspect of Shili Theory is that following the regularity of events to form an event chain, it organizes event meta and data meta and constructs an event code corresponding to the event sequence (Figure 6).

As discussed above, the information storage mechanism implements the effective storage and sharing of event management. However, solving complex system problems only relies on data and models, which is not enough [67,68]. First, human brain thinking consists of logical thinking, imaginal thinking, and creative thinking. The single information storage mechanism is good at the precise processing of information and large-scale computing data in logical thinking. Nevertheless, imaginal thinking needs to be improved, and there is still a blankness in creating thinking [69,70]. This limits the diversity of information sources. Only supporting structured data causes information sources to be incomprehensively unable to form series-wound event codes. Second, the technical cost, at that time, limits the using ability of the marginalized group, lacking the most basic digital tools to input information. Third, the single information storage mechanism is difficult to provide simple and directly useful information. This will exacerbate the inequality of data sharing, only convenient for those who can more easily analyze data. Hence, the intelligent technology of the fourth industrial revolution promoted the further development of the Shili Mirrored Structure.

3.1.3. Shili Mirrored Structure Based on the Intelligent Technology

The fourth industrial revolution has emerged. Artificial intelligence, the Internet of Things, big data, and intelligent equipment provide a material foundation for applying Shili Mirrored Structure. Expert’s knowledge, administrator’s opinions, and farmer’s experience offer a cognitive basis. Based on the meta-synthesis methodology of Qian Xuesen [71,72], this paper constructs Shili Mirrored Structure, see “Results and discussion”. More open and transparent governance frameworks are used to unify and standardize agricultural production data, reduce data flow, and improve data utilization. This structure can better address farmers’ concerns over who has access to farm data, who derives the benefits of data sharing as well as privacy concerns [73].

3.2. Case Study

At present, the study of Shili Theory and mirrored structure has not been clear. A case study as a mainstream research method that solves complex problems in design science is one of the best ways to analyze the problem mechanism and promote deduction and induction [74,75]. Therefore, this paper uses a case study method to assess the applicability of the proposed conceptual framework in the background of smart farming.

The case study was the smart farming project that our topic group continued to carry out in the three provinces of Northeast China from 2019 to 2021, including 5T (Time) management, data management system development, and latent loss measurement [76]. Preliminary research and application covered one administrative demonstration area, two demonstration farms, and ten companies of different scales. Among them, we chose two representative rice production as case studies to illustrate the application of the concept of Shili Theory, particularly through Shili Mirrored Structure, to manage smart farming.

5T (Times) management, proceeding with the natural block characteristics of the rice whole growth and deposit, conducts event management for agricultural systems. It divides rice production into five natural blocks: seed period (T_Ⅰ), seedling period (T_II), paddy shoot period (T_III), grain period (T_IV), and product period (T_Ⅴ), which is whole plant 5T [77]. The seed period (T_Ⅰ) includes four sub-natural blocks: pre-sowing preparation, soaking seed, accelerating germination, and sowing. The seeding period (T_II) includes three sub-natural blocks: pre-seedling preparation, pesticide treatment, and sprout management. The paddy shoot period (T_III) includes three sub-natural blocks: land preparation, transplanting, and control of field growth (topdressing, irrigating, pest control, disease control, and weed control). Based on the growth and deposit of grains, the grain period (T_IV) includes five sub-natural blocks: harvesting (T₁), field stacking(T₂), drying (T₃), warehousing(T₄), storing (T₅), which is grain period 5T. The product period (T_Ⅴ) includes three sub-natural blocks: processing, marketing, and consumption. Management of each sub-natural and natural block is a complete “event”. The management of rice production can be regarded as a series of producing events based on natural regularity.

In 2019, in the smart farm project, our team conducted 5T management of rice operations. At the same time, after comprehensive and sophisticated measurements, we built a “5T Post-harvest management technique code for high-quality paddy” [78], which had been included in the local standards of Jilin Province. In 2020, relying on the Industrial Alliance of Jilin University, our team developed an agricultural production data management system [76] which was gradually applied in Northeast China. In 2021, our team disclosed the “One event contact traceable coding method” based on the Internet of Things and inventory identification code.

Unlike previous industrial and agricultural management, 5T management is the event management of rice growth regularity as the core. The purpose is to improve data utilization, enhance fresh quality and reduce full chain losses. This is a new exploration to protect farmers’ interests and improve the embeddedness of intelligent agricultural technology in the natural environment and event.

4. Results and Discussion

4.1. Shili Mirrored Structure

According to the previous research and expert opinion to analyze event regularity, Shili Mirrored Structure (see Figure 7) divides the entitative structure into a series of events. It uses Internet of Things technology like various sensors, radio frequency identification, and spatial information equipment to collect and transmit information. At the same time, widespread digital tools (e.g., smartphones) allow open entry and support diverse data. In data release, an event triggering mechanism is used, taking the key elements of the event (e.g., event nature code) as the input source for drive control, thus reducing the data packet delivery rate [44]. Based on event division, Shili Mirrored Structure intelligently handles, stores, and analyzes large numbers of high-dimensional, heterogeneous, and multisource data in the virtual structure. It divides the data into root position (the event information of entering the event chain for the first time), epistatic position (the previous event information connected to the current event), current position (the current event information), hypogynous position (next event information connected to the current event). Thus, Shili Mirrored Structure identifies the event order and forms an event chain to lay the foundation for information traceability. Refer to the primary control model [16,79,80], discriminator measures, mines, and reasons information based on related operation standards, management specifications, or technique codes. It realizes digital monitoring, giving evaluation opinions and predictive signals. Then it issues deviations to the decision-maker. In order to clarify the data contribution and operational requirement of each subject, the decision support system conducts peer information sharing. The decision-maker selects the appropriate intervention to eliminate signal disturbances. The effector performs intervention and intelligent control of operation management following the input signal.

4.2. Application of Shili Mirrored Structure in Smart Farming

4.2.1. Whole Plant 5T

The 5T management of the Shili Mirrored Structure is shown in Figure 8. From the natural block characteristics of the growth and deposit, the rice production management includes five events: seed period (T_I) operation, seedling period (T_II) operation, paddy shoot period (T_III) operation, grain period (T_IV) operation, and product period (T_V) operation. Each event contains several sub-events.

Information is collected using a combination of satellite remote sensing, unmanned aerial vehicles, and proximal sensing, as well as manual entry while docking data from IoT hardware devices and network systems accessed by multiple subsystems. For example, for crop growth at planting, based on a dense network of weather and climate data, drones equipped with remote sensing equipment (multispectral, hyperspectral, or thermal sensing systems) can effectively scan the farm ground, evaluate crop growth through a 3D reconstruction, identify dry parts of farmland and estimate yields. When considering visible spectral bands, specific colors can be accentuated to acquire data about greenness (the case of green) or soil segmentation (the case of red and green) [81]. Through data processing and analysis, different data can be sorted according to “nature code + subject code + asset code + location code + time code + quantity code”, forming an event code. According to the basic regularity of the event, Shili Mirrored Structure connects the root event code, the epistatic event code, the current event code, and the hypogynous event code. Then it forms an event connect code and constructs a networked data management mechanism of the mirrored event information. It provides a distributed computing framework based on the Big Data architecture through a cloud computing center, using cluster resources to achieve distributed parallel execution of computing tasks [76]. Regarding the 5T management technique code, the discriminator analyzes and reasons mirrored event information, including crop production and variate application. Therefore, it evaluates rice quality, gives the predictive analysis and the deviation signal, and avoids excessive fertilizers to maintain soil nutrients. Under the support of the agricultural production data management system, the decision-maker selects the appropriate intervention, forming a decision-making plan for production requirements and variate applications. The effector uses automatic navigation control based on the input signal and mirrored event information, which can reduce environmental footprints [82]. It takes intelligent drive-power of agricultural machinery to reduce energy consumption and conducts the intelligent operation to improve agricultural intensification. Using agricultural robots and autonomous systems to perform various field tasks, such as sowing, pruning, phenotyping, targeted fertilization, harvesting, and sorting in automated or near-automated modes [83].

People use Shili Mirrored Structure to manage rice operations and data. On the one hand, this paper uses events to simplify the representation of agricultural production data and perceives event information through the Internet of Things. On the other hand, it follows the regularity of rice production to form the sequence of events. It uses a combination of event chains and management indicators to organize and manage data. From the above, each event consists of several sub-events. Elaborating and studying all sub-events is beyond the scope of this paper. Therefore, we select grain period 5T as an illustrative example in the next section. The aim is to describe the application of the Shili Mirrored Structure more exhaustively.

4.2.2. Grain Period 5T

The case “grain period 5T” is shown in Figure 9. From the growing and depositing regularity of rice grains, the management of the grain period (T_IV) includes five events: harvesting (T₁), field stacking (T₂), drying (T₃), warehousing (T₄), and storing (T₅).

Shili Mirrored Structure achieves perception and transmission of grain period (T_IV) events through the Internet of Things. The environmental sensors can collect temperature, humidity, light, wind direction, wind speed, and other data, which provides a basis for variate applications, quality traceability, and grade evaluation. The crop sensor collects data such as grain cultivar, maturity, and moisture content and gives reference to the operation time of rice. For example, 95–97% of grains transferred to yellow is the most appropriate harvesting date. Agricultural machinery sensor collects data related to tractors, harvesters, and dryers, contributing to management decisions and fuel savings. The pest sensor collects disease and pest data for easy warehousing, ventilation, and fumigation. A global navigation satellite system collects field area, location, and other data, laying the foundation for route planning. At the same time, there are still discrepancies in technology application among enterprises and farmers of different scales. The mobile phone APP allows farmers to enter various operating information in simple manners and with widespread digital tools. So that farmers are better embedded in information exchange networks with other groups (peers, professionals, and service providers), improving the openness and equality of technology access. For example, farmers can record the harvesting time and weather temperature (which can also be uploaded by sensors) on the app. Then the discriminator gives feedback.

The information collected is converted into the data package of mirrored events through intelligent processing, storage, and analysis. For example, machine vision is used to mine the rich information of maize images for the determination of high-moisture maize during harvesting (T₁) [84]. From the perspective of multi-factor coupling, the mutual information entropy is used as the coupling degree as an evaluation index to establish a high moisture maize image detection model. For grain bin detection during storing (T₅), the 1H proton density distribution data collected by the MRI analyzer is converted to actual moisture content data using image processing techniques [85]. As shown in Figure 10, following a certain encoding method, Shili Mirrored Structure forms the event contact code of rice production based on an event chain. On the one hand, this can unify and simplify the presentation of agricultural production data, reducing data flow. On the other hand, it increases the transparency of the process and makes it difficult to change the information. Further, it constructs the networked data management mechanism, realizing the quality traceability of rice operations.

Referring to the “5T Post-Harvest Management Technique Code for High-Quality Paddy” (

Table 1. 5T Post-harvest management technique code for high-quality paddy (Jilin DB22/T 3113-2020) [78].

Event	Necessary Indicators	Grade and Parameters 1			Sufficient Indicators	Grade and Parameters
		A	B	C		A	B	C
Harvesting (T1)	harvesting time offset/d	±2	±3.5	±5	loss ratio/%	≤2.0	≤2.3	≤2.7
	harvesting moisture content/%	24	24	—
Field stacking (T2)	mechanical harvesting time/h	≤6	≤8	≤12	stacking temperature/°C	≤20	≤25	—
	semi-mechanical harvesting time/h	≤12	≤16	≤24	loss ratio/%	≤0.1	≤0.15	≤0.2
					mildew rate/%	≤0.5	≤1.0	≤1.5
Drying (T3)	drying accumulated temperature/(°C·h)	≥300	≥280	≥270	final moisture content/%	15	14.5	14.5
	drying rate/(%·h−1)	0.5	0.7	0.8	grain temperature/°C	≤30	≤35	≤40
					loss ratio/%	≤0.1	≤0.15	≤0.2
Warehousing (T4)	warehousing time/d	≤1	≤1.5	≤2	impurity rate/%	≤0.5	≤0.8	≤1.0
	warehousing moisture content/%	15	14.5	14.5	loss ratio/%	≤0.06	≤0.08	≤0.1
Storing (T5)	storing average temperature/°C	15	20	25	final moisture content/%	≤15	≤14.5	≤14.5
	annual accumulated temperature/(°C·d)	≤8395	≤10,220	≤12,045	mildew rate/%	≤0.5	≤1.0	≤1.5
					fatty acid value/(mg·100g−1)	≤15.0	≤20.0	≤25.0

¹ Grade A is the super grade of high-quality paddy, grade B is the medium grade of high-quality paddy, and grade C is the common grade of high-quality paddy.

), we set the necessary indicators and sufficient indicators in different events of the grain period (T_IV). Based on the natural block principle, the wet heat stress control, and the taste and quality control, this technique code defines management requirements primarily related to time, loss, and accumulated temperature. Then the discriminator compares mirrored event information with technique code and other bases to carry out operation information management, dynamic simulation, comprehensive evaluation, and grain situation monitoring. For example, historical grain temperature data is called up and pre-processed to generate a temperature field cloud of each plane of the grain pile. The RGB color distribution of the temperature field clouds is used to calculate the clouds’ similarity and detect anomalies based on thresholds, thus enabling the determination of grain conditions, including dynamic analysis, static analysis, and multi-layer grain temperature analysis [86].

The “Agricultural Production Data Management System” (Figure 11 and Figure 12) can carry out full-scale traceability and comprehensive analysis of rice production events. It reflects the labor payment and operational needs of stakeholders in each event. This makes people selectively share data and disclose information in the system to better address farmers’ concerns about privacy issues and data sharing inequality. “Agricultural Production Data Management System” builds a Petri net traceability model based on event contact code and uses Markov chains to analyze the reachability of the model. In data forgery prevention, various smart contracts are designed based on blockchain technology, and smart contracts are used to build a bimodal data storage model and a data security management model. The subject competence evaluation model is constructed using the combination evaluation method based on maximizing deviations, combining the integrated index method and the principal component analysis method. The process operational evaluation model is constructed using fuzzy integrated evaluation and GA-BP neural network. After integrating the empowerment using game theoretical ideas, the product quality evaluation model is then constructed using GRA-TOPSISI. In grain situation monitoring, it uses a dynamic grain storage supervision method based on RGB color characteristics of the temperature field cloud map [87]. The amount of information generated annually after the system scale application includes 10 GB of operational production information and 1T of attachment information. It satisfies a concurrent access number of 500 and a maximum access number of 100,000. The system is developed and operated using a web server, a database server, and an application server. The configuration specification of the database server is Intel Xeon (Ice Lake) Platinum 8369B 3.5 GHz|24 cores|48 GB; the system is Windows Server 2019 Data Center Edition 64-bit Chinese version and above; bandwidth is 100M and above. The configuration specification of an application server is Intel Xeon (Ice Lake) Platinum 8369B 3.5 GHz|16 cores|32 GB; the system is Windows Server 2019 Data Center Edition 64-bit Chinese version and above; bandwidth is 100M and above. The system has undergone unit testing, code review, integration testing, functional testing, performance testing, and user testing, with an overall completion time of less than 0.5 h, data configuration time of less than 60 s, and parameter standardization of 1.3 s [88].

Further, the decision-maker programs post-harvest operation and agricultural machinery, generating the output signal. Delivering simple, actionable information to farmers on mobile terminals allows them to directly benefit from data sharing rather than just serving those who can easily access or analyze data. For example, for delayed harvesting problems, after hyperspectral images are collected by drones or manually recorded by farmers, the mobile terminal will remind for the delayed operation (if any) according to the harvesting time offset in the technique code (Figure 13). Targeting the characteristics of large hysteresis, non-linearity, and strong coupling in the post-harvest drying of rice, a multi-hidden layer BP neural network model is established using deep learning techniques. For optimal intelligent control of drying, precise regulation signals for local segmentation are generated and self-optimized with the continuous increase of drying data [89].

The effector receives signals based on mirrored event information. In harvesting (T₁), the GPS-guided harvester harvests crops with high precision and automatically measures yields, which reduces resource energy consumption and increases production intensity. In field stacking (T₂) and warehousing (T₄), it plans navigation paths, reducing environmental footprints. In drying (T₃), the dryer with intelligent control uses machine commands based on Shili Mirrored Structure and controller to simultaneously regulate the waiting time (discharge speed) and the humidity of the hot air in each drying section, improving rice quality. In storing (T₅), the effector performs environmental intelligence regulation [90]. A novel screw-drive granary robot receives signals for the automatic leveling of loose grain surfaces to reduce manual risk [91,92].

4.2.3. Application Situation and Efficiency

Relying on the Jilin Rice Industry Alliance, the Agricultural Production Data Management System is installed and running in batches. It is expected to launch large-scale applications in the fall of 2022. As shown in Figure 14, the current coverage includes 11,973 hm² of cultivated lands, 51 harvesters, 201,487 m² stacking fields, 41 dryers, 191,697 m² warehouses, and 619,563 t storehouses. These form a batch of event information, initially realizing data management of rice production (Figure 15).

Meanwhile, in the rice data of six cultivars of Jilin Province collected, the traditional operation method had a visible loss, namely mechanical harvest loss of 3.83% in harvesting (T₁), transportation loss of 1.85% in field stacking (T₂), loss of drying process of 0.5% in drying (T₃), mildew loss of 1% in warehousing (T₄), storing loss of 2% in storing (T₅), and over-processing loss of 3% in product period (T_V) (Figure 16). In addition, delayed harvesting also resulted in a latent loss of 7.16% (

Table 2. Latent loss rate of six rice varieties on different days after heading [76].

Days after Heading/d	Yield/(kg·hm−2)						Loss Rate/%						Average Yield/(kg·hm−2)	Latent Loss Rate/%
	Songjing16	Jiho-ng6-300N	Jiho-ng6-260N	Xin-feng6	Lon-gdun1614	Dao-huaxiang2	Son-gjing16	Jiho-ng6300N	Jiho-ng6260N	Xin-feng6	Lon-gdun1614	Dao-huaxiang2
55	7546	6604	6039	6680	6634	6238	0	0	0	0	0	0	6623	0
60	7456	6382	6029	6500	6523	6126	1.20	3.36	0.16	2.70	1.67	1.79	6503	1.81
65	7365	6160	6019	6319	6413	6015	2.39	6.72	0.32	5.41	3.34	3.58	6382	3.63
70	7275	5939	6010	6138	6302	5903	3.59	10.08	0.49	8.11	5.00	5.37	6261	5.44
75	7203	5733	6005	5957	6191	5792	4.55	13.19	0.56	10.82	6.67	7.16	6147	7.16

). We used the nuclear magnetic resonance technology to test synchronously and found that this mainly included dry matter loss of 3.5346% [93], naturally over-drying loss, and shattering loss. According to this, 5T management adjusts operation time, which can directly retrieve visible and latent losses of 8.83%. With an annual output of 6.67 million tons of rice, Jilin Province can save 588961 t of rice in the fields every year in the future. Taking into account the climate and rice cultivation, with an annual production of 49.6 million tons of rice, north of the country and the Yellow River Basin can save 4379680t of rice in the fields every year in the future. With 57789 km² of existing rice arable land, China is equivalent to an additional 5103 km² of arable land.

In summary, based on Shili Theory and Shili Mirrored Structure, we carry out 5T management of high-quality rice operations by using the agricultural production data management system. This enhances the correlation between data and events. It can realize the impairment lessen and quality improvement of the agricultural production process, especially revealing the 8.83% significant latent loss in rice harvest.

5. Conclusions

The fourth industrial revolution provides a new path for breaking agricultural production difficulties, promoting agricultural production to smart farming. Nevertheless, the application of intelligent agricultural technology also brings a lot of data problems. Therefore, this paper has analyzed how the Shili Theory can unify and standardize agricultural production data. More specifically, it addressed the three objectives mentioned in the introduction.

First, this paper has defined the concept of Shili Theory which researches the basic regularity of the event by Shili Mirrored Structure. Based on the natural characteristics of crop growth and deposit, Shili Theory shows the basic regularity of agricultural production events in smart farming. It uses the Shili Mirrored Structure to manage agricultural production data, relying on the Internet of Things, big data, artificial intelligence, and other technologies. Combined with literature analysis, this paper expounds on the research of Shili Theory in agricultural systems. It shows three features: taking an agricultural production event as a management object, determining the event regularity based on the natural block characteristics of the crop, and emphasizing the natural attributes.

Second, the paper has proposed a conceptual framework. Based on the possibility of technology development, it proposes a Shili Mirrored Structure from the human brain memory mechanism to the information storage mechanism to intelligent technology. This structure determines root, epistatic, current, and hypogynous events around event regularity. It constructs a virtual structure corresponding to the entity structure, providing simple and directly useful information with an open, transparent, and accessible structure. Shili Mirrored Structure forms a networked data management mechanism with “event” as the core, reducing data flow and improving data utilization.

Third, the framework has been applied and verified in the Agricultural Production Data Management System of the 5T management case. According to the natural block characteristics of rice whole growth and deposit, the management process is several production events based on natural regularity. On the one hand, it utilizes the subversive technologies of the Fourth Industrial Revolution to perceive event information. On the other hand, the combination of an event chain and indicator system is used to manage data, improving technology’s embeddedness in events and the natural environment. The case-specific structure focuses on the application of the Shili Mirrored Structure in rice whole plant 5T and grain period 5T. The current coverage includes 11,973 hm² of cultivated lands, 51 harvesters, 201,487 m² stacking fields, 41 dryers, 191,697 m² warehouses, and 619,563 t storehouses. Rationally regulating the operation time can directly retrieve visible and latent loss of 8.83%.

In summary, Shili Theory and Shili Mirrored Structure offer possible solutions for data management and the application of intelligent agricultural technology. However, deploying new technologies for smart farming in real-time scenarios with limitations such as internet connectivity, security, cost, and high computing resources remains a challenge globally. Climate change may abruptly alter the seasonal life cycle events of crops. Therefore, using Shili Mirrored Structure to combine all technologies for crop field growth, clarifying data analysis, managing needs, and implementing intelligent control to support the complex event that is modern cultivation may be a future research direction. Furthermore, with regard to the organizational governance, data ethics, and ownership issues involved in smart farming, we would like to encourage researchers in these disciplines also to undertake research on Shili Theory, as this may be key to the practical application of Shili Mirrored Structure on a large scale in agriculture.