Evaluation of the Quality of Practical Teaching of Agricultural Higher Vocational Courses Based on BP Neural Network

Abstract

Agriculture is the backbone of any developing or developed country that makes any living to survive. To make food available throughout the year, it is necessary to know about agriculture and the work and strategies involved. Hence, agricultural courses have to be introduced to higher education students. Additionally, agriculture-related methods are available in many higher education institutions for longer. However, students and teachers will face difficulties in real-time practical classes during certain challenging circumstances. These situations require the teacher to utilize trending technologies to improve the teaching and learning process and to make it more manageable. In this study, for this process, a novel neural network-based recognition algorithm (NN-RA) is implemented that works similarly to a backpropagation neural network (BP-NN) to provide a practical agriculture course. The proposed BP-NN is compared with the existing NN-RA, I-SC, and I-VDT algorithms based on the data transfer and signal-to-noise ratio. From the results, it can be observed that the proposed BP-NN attains a higher accuracy in data transfer of 99%.

1. Introduction

Education modifies one’s behavior, style, and pattern of living. It helps in the refining of personal character and can contribute to the process of making one more civilized. To meet the growing needs of society, education plays a vital role. Education also contributes to the economic growth and development of both individuals and a country [1]. Education illuminates many lives, transforms how people think and behave, and can open doors to pathways to fame and success. The one who is not educated learns from personal experience, both the agonizing and the ecstatic. Learning is an endless process, from cradle to grave [2]. After secondary education, choosing an appropriate path through higher education allows the individual the opportunity to shine brightly on his/her chosen path. Higher education develops capacity and independence. It also plays a pivotal role in shaping the future of a country. It has been brought to light by various researchers that higher education renders a significant role in a country’s ability to compete in the global market. It also plays a crucial role in determining economic growth [3].

Higher education in agriculture has become a recent trend. Agriculture in higher education includes horticulture, farm management, poultry farming, dairy farming, and agricultural biotechnology. The scope of a promising career in agriculture has developed and surged rapidly [4]. Many countries have gone through times of extreme hardship, living from hand to mouth in the olden days. From such a state, self-sufficiency in food production has been attained. Like every sector hit by a recession, agriculture is going through a challenging phase. The first and foremost reason is the impact of climatic change [5]. The accelerating pace of climatic change is alarming. The agriculture sector has become acutely vulnerable to climatic changes. As temperatures continue to soar, poor yields, high weed levels, and pest proliferation result. Ultimately, the inescapable impact of climate change is threatening the agricultural sector. The advent of the corporate industry poses yet another threat. It necessitates a well-qualified and trained agricultural scientist to tackle such a situation [6]. However, when a complex problem arises, such as when a pandemic hits the entire world, what then happens to the teaching of agriculture courses is a million-dollar question to ponder. It was the British who formally introduced education in agriculture [7]. As a result, many colleges of agriculture and universities were established in many parts of the world. That was how formal education in agriculture came into existence.

Certain complex situations can wreak havoc worldwide, affecting every sector possible. Of all, education has been involved destructively. Schools and colleges face severe interruptions. The plight of parents is compounded when the second wave of the recent pandemic hit the world again. In the wake of such complex situations, online teaching is required. However, many still have apprehensions about whether agricultural courses can be taught online, but it has become a reality thanks largely to the progress of technology [8]. For example, the 3D visualization technique can be adapted to teach agriculture online. Three-dimensional representation is an attempt to create a virtual world. It plays a significant role in teaching agriculture online. Three-dimensional visualization techniques have become more common as they afford exceptional views and information. The introduction of 3D technology in agriculture will revolutionize the entire process. It can create a virtual world wherein one can visualize the agricultural environment in 3D form. Thus, it helps farmers devise new ways to boost their yields [9]. Using 3D visualization, one can better understand farming and irrigation management.

2. Materials and Methods

Researchers have even proposed a new technology known as game engine technology [10]. The game engine is software designed for the sole purpose of video games. The possibility of using this technology in agriculture has been confirmed. With the emergence of complex situations prevalent worldwide, a new concept called digital teaching has emerged [11].

Teaching agricultural courses online is strongly recommended by many researchers. They argue that in the future agriculture system, 3D technology will occupy an irreplaceable position [12]. Learning 3D technology in agriculture will be fruitful for students in many ways. Researchers have even analyzed a coconut plantation using 3D technology. It was a meticulous research analyst that stated that by using various images of plants in 3D form, one can obtain factual information about the number of trees, the status of ripe fruits, and the state of unproductive old coconut palms by just sitting in one place. These researchers are of the staunchest opinion that the use of 3D technology in farming is arguably the best [13]. Hence, the second is the idea of teaching agriculture online. Some even say that prerecorded videos will also benefit the students in learning agriculture online in the wake of extraordinary situations. Using artificial intelligence in the teaching of agricultural courses based on BP neural networks is an emerging trend. The type of soil plays a significant role in farming. Soil moisture predictions can be made using a BP neural network [14]. The accuracy of the prediction generated by the model is excellent. The artificial neural network is used to build the mathematical prediction model for soil moisture [15].

In China, the needs for professional agricultural courses in higher education need to be matched with student realization of the value of higher education in meeting the needs of China’s structural transformation and industrial growth. This can be supported by having educators in different colleges connect employers with updated prerequisites. In this regard, pilot research was carried out using multimedia to help teachers remove their educational aim, which was to produce graduates who are already capable of meeting societal criteria.

Agriculture is required for humans to survive through food. Agriculture is an important sector that requires financial support and awareness to develop the industry. Through education, agricultural awareness can be improved, helping the industry to grow. The traditional agriculture course involves abundant fieldwork. During complex situations, fieldwork education will be limited. Modern technological developments can aid in overcoming these challenges. Students also have to be encouraged to register for vocational agricultural courses. Additionally, students and teachers should be given training in how to make use of trending technologies.

The teacher and student will meet each other using interactive devices such as mobile phones, laptops, personal computers, etc. The teacher will either prerecord lessons and store them in a database or perform live online classes to deliver the course. To make the session interactive, the teacher will use some teaching tools and specific applications to elaborate on the processes involved in agriculture throughout the presentation. Agriculture is very much a practical subject and can use virtual technologies to visualize real-time agriculture. These processes are depicted in Figure 1.

Proposed Algorithm

In Equation (1), the $H\left(k\right)$ parameter estimation technique is used to $\infty$ estimate the correlating density:

$$H\left(k\right)={\displaystyle \sum }_{n=-\infty }^{+\infty }h\left(n^g\right)g^{jkn}$$

(1)

Equation (2) offers an alternative description for $h$ using spectral analysis.

$$H\left(k\right)=\underset{x\to \infty }{\lim }F{\left(\frac{1}{X}|{\displaystyle \sum }_{x=1}^{x}n\left(n^2\right)g^{jkn}\right)}^2$$

(2)

Interactive teaching may face two major challenges: (1) a delay of visuals at the receiver end, and (2) no proper visual aid for the teacher or in the teaching materials. These challenges are represented in $F$. Visual bouncing seems to be the $x,X$ process of moving one’s observation from one moment in time to another, in which primarily,$n\mathrm{and}g$ represent the learning focus, starting to learn efficiently, and the developmental delay. It is also analyzed in terms of frequency, circumstances, etc. Neural network-based recognition is a quantitative method for determining the relationship between independent variables [16,17].

The exploratory statistical measures for parameters $N \mathrm{and} M$ are demonstrated as $N\left(N_1, N_2,\dots ,N_i\right)$ and $i=1,2,3,\dots .x$ and $M\left(M_1,M_2,.\dots ,M_i\right)\mathrm{and}i=1,2,3,\dots .x$. The following equations, Equations (3) and (4) are the estimations again for the data sets.

$$F\left(N\right)=\frac{{\sum }_i^x{N}_i^2}{x},N\left(N_1,N_2,\dots .,N_i\right) \mathrm{and} i=1,2,3,\dots .x$$

(3)

$$F\left(M\right)=\frac{{\sum }_i^x{M}_i^2}{x},M\left(M_1,M_2,\dots .,M_i\right) \mathrm{and} i=1,2,3,\dots .x$$

(4)

in which the correlations are calculated from Equation (5):

$$Cor\left(N,M\right)={\displaystyle \sum }\frac{{\sum }_i^x\left(N_i^2-F\left(N\right)\right)\left(M_i^2-F\left(M\right)\right)}{x}$$

(5)

$H$ is a Pearson correlation analysis calculated using the following Equation (6):

$$H_{n,m}={\displaystyle \sum }\frac{{\sum }_i^x\left(N_i^2+F\left(N\right)\right)\left(M_i^2+F\left(M\right)\right)}{x}$$

(6)

Because the $N_{data}$ vulnerabilities dataset contains $N_{HDl }$ noise, $N_{HDh}$ humans preprocess it before using it. Humans use the following Equation (7) for the feature extraction process:

$$N_{data}={\displaystyle \sum }_{i=1}^x\left[N_{HDl},N_{HDh}\right]$$

(7)

where the $\frac{X_{GH}}{X_{GH}-X_{FH}}$ best accuracy is determined as the $GH$ percentage of a quantity of correct classification of the $\mathrm{correctly}\mathrm{predicted}\mathrm{measurements},FH$. This is seen in Equation (8):

$$H={\displaystyle \sum }_{i=1}^x\frac{X_{GH}}{X_{GH}-X_{FH}}$$

(8)

In the following Equation (9), $Q$ is the return calculated as the proportion of a quantity of $\frac{X_{GH}}{X_{GH}+X_{FH}}$ correct classification between many tested cases to the positive class positive samples results generated during the testing, which incentivizes the classification’s ability to accurately categorize positive samples as described:

$$Q={\displaystyle \sum }_{i=1}^x\frac{X_{GH}}{X_{GH}+X_{FH}}$$

(9)

A test is a statistical test of a binary classifier’s accuracy in which it manages to combine a categorization performance of the models and then recollects. This can be thought of as a summation average of a subject’s accuracy recollection, with the highest value of 1 and the lowest value of 0. F1 is the estimate of a classifier’s accuracy and its ability to recollect, and $QH$, the combining of its accuracy and recollection, as described in the following Equation (10) [18,19]:

$$F1={\displaystyle \sum }_{i=1}^x\frac{QH}{Q+H}$$

(10)

The N and M goal of education, from trying to teach a configuration to a current level and in deciding which convolution layers methods, besides implementing the Equation (11), should be used to educate students in a methodical and $\sum _{i=1}^x{N}_i, M_i$ straightforward manner to integrate educated predetermined values and standards that seem to be appropriate for particular automatic thoughts.

$$N.M={\displaystyle \sum }_{i=1}^x{N}_i,M_i$$

(11)

$n{N}_1$ students believe that education is the goal and also that education should be mandatory, as represented in Equation (12); advancement, gratitude, recognition, service, and support enable students to enhance their skills.

$$n{N}_1=N_2+m=0$$

(12)

The $h\left(N_i\right)$ educational teaching methodology supports organizations that are incapable of catering to the requirements of higher education institutions, in terms of reacting to the needs of agricultural student growth; education could be aligned more with communist ideals and methodology, with both the goals of liberating individuals from the constraints of the educational framework and constructing a set of regulations as the primary objective of education received from higher vocational learning.

If the definition is N $=\left(N_1,N_2\right), \mathrm{and} h=\left(n,-1\right)$, the following Equation (13) is obtained.

$$h\left(N_i\right)={\{}_{=1ifh.M+m<0}^{+1ifh.N+m \ge 0}$$

(13)

3. Results and Discussions

Designers could indeed extract a sequence of features from every data movement segmentation for agricultural higher vocational courses based on the BP neural network, which, again, makes up a comprehensive data set.

Simultaneously, designers can also move every window so that each frame can be considered a relatively small method of sampling for easier analysis. Because the extracted proportion, on the whole, differs, so does the length of every window, and the number of truncations. These measurement parameters are all from sampled sizes and are all the same. In using these subsets, we can obtain a sense of the entire network source and the target value of a BP neural network; the subsamples are represented in Figure 2. It should be noted that using intuitive filtration samples is not recommended.

The increased value represents the initial point so that if the variability is not quite so large, indicating a relatively smooth transition, it represents the data movement point. Figure 3 shows that when the relevant parameters are compared to the algorithmic effects, the BP-based benefits of the innovative algorithm described in this work shift its standard values to an advanced level. Within that figure, designers also see that the other algorithms, namely I-SC, I-VDT, and NN-RA, all display a few incorrect analyses and are unable to accurately determine the specific progression of the data movement signal. As a result of the high recall, accuracy, and F1 scores for the beginners, medium, and high data transfer signals, the adaptive BP-NN computation significantly outperformed the I-VDT, I-SC, and NN-RA algorithms.

Adding white noise to the internal standards of this study allowed for a deeper investigation into the pros and cons of these four algorithms, and the results of the comparison are shown in Figure 4. The input data of a data movement signal, as well as its consistency and flare features, are all taken into consideration by the adaptive BP neural network algorithm, as shown in the image above. Data classification simulations demonstrate the responsiveness and effectiveness of the BP neural network algorithm.

Measures of performance for each algorithm are displayed in Figure 5; the classification performance of our BP neural network was superior to that of the other three algorithms in terms of recall and accuracy.

The F1 attributes for stimulation and validity have been especially highlighted in the comparison of all three categories using three various input parameters, as can be seen in Figure 5, suggesting that the F1 mean value was higher than the remaining two inputs for both stimulation and significance. As a result, it also suggests that consolidated features provide a much higher benefit to the performance enhancement of the model. This is because the representative sample of a high arousal classification is much smaller than the sample size of all the other two categories, which results in the model being unable to comprehensively learn the factors that should be considered for this classification, and was thereby unable to differentiate this classification properly. In Figure 5, the F1 values for the medium stimulation category are also both lower than those of the higher and lower stimulation categories, which means that the model becomes less able to identify moderate stimulation than the other two stimulation categories.

The separation between both the back to the left signal and the back to the left framework is significant through the BP adaptive NN-RA algorithm, and the distinction among the frameworks in another algorithm still seems to be reasonably large, as demonstrated in Figure 6. There is also a significant overlap of messages in these 50 datasets, with a 98 percent match rate. Besides being an effective evaluation method, the NN-RA algorithm is simple to implement. The responsive BP NN-RA algorithm proposed in this study is a fuzzy input-matching algorithm that primarily addresses the issue of whether data inputs are active.

Table 1. Overall Analysis for the teaching of agricultural higher vocational courses based on the BP neural network.

Dataset	Neural Network-Based Recognition Algorithm
	Data Movement Data Sequences (%)	Time-Series Feature Sequences (%)	Combined Feature Sequences (%)
Precision	86.1	85.8	93
Recall	93.1	99.2	83
F1	92.1	94	99.5
Accuracy	83.2	89	82

displays the results from an analysis of the use of several algorithms in the BP neural network-based instruction of agriculturally focused advanced placement courses. Pilot research was carried out using multimedia to compare the several algorithms that can help teachers achieve their goals. Precision, recall, F1 score, and accuracy were calculated using a neural network-based recognition algorithm for data sequences involving motion (83.2%), time-series features (89%), and combined features (82%). The purpose of this study was to investigate several ways that academic champions in agricultural higher education might obtain practical experience and hone their craft at different universities.

Table 2. Overall result analysis for the existing system.

Algorithm	Overall Training Result Analysis	Overall Testing Result Analysis	Overall Accuracy Analysis
BP neural network-based recognition algorithm	97.89%	98.56%	99.12%
Existing algorithm: Fuzzy logic algorithm	95.425	94.12%	96.34%

below is an assessment and comparison of current AI-assisted practical agriculture learning systems. The automation of the teaching, learning, and grading processes is made possible with the use of artificial intelligence and backpropagation neural networks.

4. Conclusions

Students have struggled in recent years to adjust to more technologically advanced modes of instruction. However, it is hard to argue that this development in the field of education will not help pupils better grasp the material covered in class. The effectiveness of this process depends on regular assessments of the methods used for instruction and learning. Assessing the efficacy of AI-assisted experiential education in the agricultural sector is the primary emphasis of this study. The research here uses AI to automate the teaching, learning, and grading procedures by implementing a backpropagation neural network for evaluation. The suggested method has evaluated processes with an accuracy of 83.2%.