Development and Performance Evaluation of a Pressure-Adjustable Waterjet Stubble-Cutting Device with Thickness Detection for No-Till Sowing

Abstract

No-till maize (Zea mays L.) sowing is often affected by stubble. The high-pressure waterjet at a constant rate is powerless to precision applications of stubble cutting and causes water waste. In this study, a pressure-adjustable stubble-cutting device with a stubble-thickness detection device was designed. Through experiments, the quantitative analysis of the moisture content and electrical conductivity (EC) of the field soil and stubble during the spring sowing period was conducted, and the effect of soil moisture content (SMC), soil compaction (SC) and machine forward speed (V) on the stubble-thickness detection error (STDE) was explored. On this basis, the optimal parameters of the device were analyzed and evaluated, and a verification experiment was applied. The results showed that STDE decreased with the increase of SMC and SC and increased with the increase of V. The response time of the pressure adjustment system is 0.12 s. The stubble-cutting device with thickness detection for two-level pressure regulation reduced the water consumption (WC) by 13.22% under the condition that the stubble-cutting rate (SCR) remained unchanged. The WC increased with the increase of waterjet pressure (P) and decreased with the increase of V. The SCR increased first and then decreased with the increase of cutting angle (α). The response surface optimization analysis showed that when P was 26 MPa, α 90.45° and V was 3.36 km/h, the performance was optimal with a 3.03% STDE, a 95.49% stubble SCR and a 49.98 L/ha WC. The measured value of the field verification experiment had a 4.11% relative error existing in STDE, a 4.06% relative error existing in the SCR and a 1.81% relative error existing in WC compared with the predicted value of the regression model. In contrast to the constant rate waterjet cutting device, the application of this device can save WC by 13.22%. This study can provide a reference for the application of waterjet technology and conductivity detection technology in the agricultural field.

1. Introduction

Conservation tillage is a fundamental property of sustainable agricultural development and soil conservation, which has been widely used in Northeast China since the 1950s [1]. The remaining stubble, which covers the ground surface, can protect the soil from wind erosion, increase soil organic matter content and improve soil pore structure [2] but also hinder sowing [3]. Therefore, a good performance of stubble cutting is essential for no-till sowing machines.

The stubble-cutting device used by sowing machines can typically be classified into two categories: active and passive, which depends on its working form. The active stubble-cutting device cuts stubble by spinning metal blades with driving power. The biomimetic design was applied by Zhao et al. [4], Li et al. [5] and Chang et al. [6] to increase the effectiveness of stubble cutting and decrease resistance. This kind of device has good stubble-cutting performance but also costs energy and disturbs the soil. According to Jia et al. [7] and Qin et al. [8], the passive stubble-cutting device primarily employed a disc knife (corrugated, notched, flat, etc.) and a stubble-shifting wheel to cut the stubble and shift it toward both sides. The passive stubble-cutting device typically requires a heavy counterweight, which compacts the soil and hinders plant growth [9,10].

Compared with the above soil-contact stubble-cutting device, the high-pressure waterjet is a non-contact cutting technology that uses water as a cutting medium. In recent years, it has been widely used in industry and agriculture, such as precision manufacturing [11,12,13], mining [14,15,16], agricultural material processing [17,18,19,20,21,22,23,24,25] and so on. Cao et al. [26] studied the interaction and erosion mechanism of waterjet on the surface of A-grade marine steel under 20 MPa water pressure. It can be seen that waterjet cutting has greatly improved the cutting ability, whether used as a primary cutting method supplemented by abrasives [13,27,28] or as an auxiliary method [14,15,29,30].

Researchers have conducted studies on factors such as pressure, nozzle diameter, movement speed, cutting angle and target distance on cutting effectiveness [13]. Cui et al. [31] summarized the application of waterjet cutting in agricultural engineering, analyzed how to improve the cutting ability and conducted the problems and solutions of waterjet cutting for agricultural material. Hu et al. [3] designed an ultra-high-pressure waterjet-assisted furrow opener for no-till seeders and conducted the optimal operating parameters for three factors: water pressure, impact angle and machine forward speed. Desbiolles et al. [32] developed a liquid coulter for cutting crop residues and found that the best cutting performance occurred when wet straw was compressed. Perotti et al. [33] evaluated the cutting ability of an abrasive waterjet on wheat straw by comparing three factors: wheat straw cover thickness, water pressure and nozzle diameter. However, efficient utilization of water resources is necessary, and the adjustment of the water pressure toward the thickness of the agricultural material in the field requires further study.

At present, the stubble coverage area detection methods mainly use manual rope pulling and machine vision. The manual rope-pulling method has low efficiency and high labor cost. In recent years, machine vision has been successfully applied to stubble coverage area detection, making up for the efficiency shortcomings of manual rope pulling. Liu [34], Memon [35], Riegler Nurse [36] and Laamrani [37] estimated the stubble coverage area using different machine vision algorithms based on the stubble coverage images collected in the field. However, these estimation methods for stubble coverage have timeliness and other limitations, and there are few studies focused on the thickness detection of stubble.

Electrical conductivity (EC) can reflect soil physical parameters such as moisture content and compaction. Researchers have conducted studies on soil EC and EC detection devices. Liang et al. [38] designed a vehicle-mounted field soil EC detection system based on the four-electrode method. The system, whose response time was about 540 ms, mainly consisted of a constant current signal circuit, signal processing circuit, Arduino controller and vehicle-mounted sensors. The experiment results verified the effects of tractor vibration, sensor insertion depth, movement speed and soil compaction on the detection accuracy of the system. Naderi-Boldaji et al. [39] developed a combined horizontal penetrometer composed of tine, EC sensors (Veris^® 3100) and load cell housing. The experiment showed that the apparent conductivity of soil at a depth of 0–0.3 m was significantly positively related to the penetration resistance measured by the combined horizontal penetrometer. Beatriz Batista de Melo et al. [40] found that the main characteristics of soil compaction were the increase in soil bulk density and the decrease in porosity. Therefore, soil with high compaction had higher electrical conductivity and could be used to estimate the soil moisture content. However, the EC association between stubble and soil during spring sowing in Northeast China requires further study.

Based on the aforementioned problems, this research created a pressure-adjustable waterjet-cutting device according to stubble thickness. It could achieve high-performance stubble cutting as well as effective water utilization. This research can serve as a guide for the development and use of waterjet technology in agriculture.

2. Materials and Methods

2.1. Design and Workflow of the System

As shown in Figure 1a,b, the structure of the stubble-thickness detection device in this study adopted a similar structure to the depth-limiting wheel. When working, the wheel pressed on the stubble and ground, rotating passively. Therefore, the measured stubble thickness was in a compact state. There were two EC sensors embedded in the center of the wheel. The six-wire signal slip ring was applied to transmit the signal collected by the sensors. The angle sensor rotated synchronously with the wheel to record the rotation angle, fixed above the six-wire slip ring through a synchronous pulley of the same size. Inside the controller box, there was a 12 V power supply, Arduino Uno and shield v5 expansion board (which can provide more IIC module interfaces, SD card module communication interfaces and 32 servo controller interfaces), LCD display screen (equipped with IIC modules for easier brightness adjustment) and SD memory card module.

The waterjet stubble-cutting device in Figure 1c applied a plunger pump powered by a gasoline engine and a manual gear throttle to fix the speed, providing stable pressure. While working, the water was pressurized by the plunger pump through a filter and sprayed out through a 0.5 mm gemstone nozzle to achieve the stubble-cutting effect. The pipeline was also equipped with a pressure monitoring valve. The pressure was controlled by a proportional solenoid valve that communicates with Arduino. The technical details of instruments mentioned above are listed in

Table 1. Instrument technical details.

Instrument	Type	Manufacturer	Range	Accuracy
EC sensor	PR-3000-ECH	Jinan Renzhi Measurement and Control Technology Co., Ltd., Jinan, Shandong Province, China	0–10,000 μs/cm	1 μs/cm
Angle sensor	GT-A-V-2236	Taizhou Quantum Electronic Technology Co., Ltd., Taizhou, Zhejiang Province, China	0–360°	0.09° (4096 bits)
Six-wire signal slip ring	MT0733-S06-VC	Moflon Technology Co., Ltd., Shenzhen, China
Water pressure gauge	GX390 Honda gasoline engine matching pressure gauge	Honda Power (China) Co., Ltd., Chongqing, China	0–40 Mpa	1 Mpa
Proportional solenoid valve	CND-10 L-400 P DN10H	ChuanNai Valve (Shanghai) Co., Ltd., Shanghai, China	5–400 bar	1 bar
0.5 mm nozzle	Gem nozzle 6.99-10.5-0.508	Shenzhen Speed Industrial Materials Co., Ltd., Shenzhen, China

.

Figure 2 shows the system workflow as follows: (1) Arduino Uno received signals from angle sensors and EC sensors and worked according to the thickness detection principle set by the pre-program. If ${EC}_{T_{i+1}}-{EC}_{T_i}\ge {EC}_{T_X}$, the angle at the time of T_i+1 would be recorded. (2) Arduino Uno processed and converted the data from angle sensors and EC sensors according to Equation (2) in Section 2.2 to obtain the stubble thickness $S_{T_{i+1}}$ at the time of T_i+1. Then, the stubble thickness $S_{T_{i+1}}$ data was stored on the SD card and displayed on the LCD display screen. (3) The stubble thickness was compared with the S_X. If the current stubble thickness was less than S_X, Arduino Uno would send a signal to keep the proportional solenoid valve at the low-level water pressure (22 MPa). If the current stubble thickness was greater than S_X, Arduino Uno would send a signal to adjust the proportional solenoid valve to the high-level water pressure (26 MPa).

2.2. Working Principle and Calibration of the Stubble-Thickness Detection

The working principle of the stubble thickness detection device is shown in Figure 3. When working, the horizontal axis in the forward direction needs to be set to 0° of the angle sensor in advance. When the probe of the EC sensor penetrated the soil from the stubble, the measured soil and stubble EC value would have a significant change. The EC value difference between T_i+1 and T_i was greater than ${EC}_{T_X}$, which meant the probe was exactly right at their junction. Then, the rotation angle at T_i was recorded according to Pythagorean’s theorem as in Equation (1):

$$\frac{OB}{OC}=\mathit{cos}\left(90°-\alpha \right)$$

(1)

Then, the stubble thickness AB:

$$AB=OC·sin\alpha -OA$$

(2)

The stubble-thickness detection device was calibrated in the soil bin (Figure 4) of the Agricultural Machinery Laboratory of Jilin Agricultural University. The soil bin (Figure 4) was 3.5 m long and 0.8 m wide. The slide rail was driven by a motor and controlled by an external frequency converter. The sliding speed was 1.5–12 km/h, which could meet the demand of the speed factor. The stubble thickness with similar physical characteristics was 0–10 cm manually laid, with an increase of 1 cm each time. The EC detection probe was smeared with colored dye to mark the original position in the soil bin. The data measured by the stubble-thickness detection device was stored in the SD card for subsequent sorting and calculation. The actual stubble thickness was manually measured using a ruler to measure the distance from the soil surface to the outer surface of the wheel, which was applied to press on the original position marked by the colored dye, shown in Figure 1a. The relationship between the actual stubble thickness and the device-detected and device-converted stubble thickness were figured out with three replications (Figure 5).

2.3. Water Pressure Adjustment System Response Time

The stubble thickness detection device (Figure 1a) was placed in the soil bin (Figure 4) for the experiment. Stubble 5 cm and 10 cm thick with a length of 50 cm were artificially put on the soil in the bin as a test for pressure adjustment. The water pressure test system [41,42,43] is shown in Figure 6a. The 0.5 mm nozzle of the waterjet system was put directly above the center of the metal plate, with a 10 mm target distance between the nozzle and the plate. The GJBLS-I force sensor (Bengbu High Precision Sensing System Engineering Co., Ltd., Bengbu, Anhui Province, China) was placed in the center of two metal plates (10 cm × 10 cm × 0.5 cm), and the metal plate was placed firmly and horizontally. The high- and low-pressure waterjet impact force measured by the pressure sensor was recorded by the software on the computer at the mean time with a sampling frequency of 200 Hz. The stubble thickness detection device was rolled over the 1 cm and 10 cm areas in the soil bin continuously, and the response times with three replications for the high- and low-level pressure conversions of the pressure adjustment system were calculated. Figure 6b shows that the pressure adjustment system had a response time of 0.12 s when switching the pressure between high and low levels.

2.4. Experiment Design

Firstly, the EC of the field soil and stubble was measured before the spring sowing in 2021 in order to figure out the EC range of the field soil and stubble. Then, the soil bin experiment was carried out, and the field state of the soil and stubble was reconstructed by sprinkling water and roller compaction. The soil moisture content (SMC), soil compaction (SC) and machine forward speed (V) were used as the experiment factors. The stubble-thickness detection error (STDE) was used as the indicator to evaluate the performance of the stubble-thickness detection device. Finally, a compared experiment with and without a stubble thickness detection device to adjust the pressure was conducted to evaluate the stubble-cutting performance and water consumption. A field experiment was carried out under the optimized parameters, with the waterjet pressure (P), cutting angle (α) and machine forward speed (V) as experiment factors and stubble-cutting rate (SCR) and water consumption (WC) as evaluation indicators to verify the effectiveness of the waterjet stubble-cutting device with a stubble thickness detection device to adjust pressure.

The experiment data were analyzed by Design-Expert V12. The data graphs were drawn by OriginPro 2022.

2.4.1. EC Measurement Experiment of Soil and Stubble in the Field

Field soil and stubble EC measurement experiments were carried out in the experiment base of Jilin Agricultural University (125.41° E, 43.83° N) in 2021. The soil type of the field was chernozem, and the previous crop was maize (Zea mays L.) (Jinongyu 898), planted by no-till sowing and harvested by a maize harvester in 2020 autumn and maize stubble returned to the field. As shown in Figure 7, three areas of 25 m × 100 m were randomly selected in the field; 200 experiment points for soil and stubble were randomly selected, and the average value after obtaining the measurement values of each area was calculated to eliminate random errors between the areas. The original positions EC and MC were detected by VMS-3002-HHT-N01 (Shandong Vemsee Technology Co., Ltd., Jinan, Shandong Province, China) hand-held multi-function detector (made-in temperature compensation sensor with a 0–50 °C compensation range). The soil measurement depth was between 1 cm and 5 cm, and the stubble measurement depth was subject to the penetration of the compacted stubble without exposing the probe. The EC and MC of soil and stubble of each experiment point were measured three times to obtain the average value.

2.4.2. Stubble-Thickness Detection Performance Experiment

The Box–Behnken method was adopted, and SMC, SC and V were used as the experiment factors. STDE was used as the evaluation.

Table 2. Box–Behnken factor level for STDE.

Factor	Unit	Level
		−1	0	1
SMC	%	10	20	30
SC	MPa	0.6	1.2	1.8
V	km/h	3	4	5

shows the range of each experiment factor. A total of 17 combination experiments were conducted with three replications. The results were evaluated by analysis of variance (ANOVA, 95% confidence interval).

The calculation Equation (3) of STDE is as follows:

$$STDE=\frac{\left|D_D-D_O\right|}{D_O}\times 100\%$$

(3)

where STDE is the stubble-thickness detection error. D_D is the thickness detection value of the stubble thickness detection device. D_O is the stubble thickness value measured manually in the same position.

2.4.3. Waterjet Stubble-Cutting Performance Experiment

Firstly, a single-factor experiment was designed to evaluate the stubble-cutting performance and water consumption with and without a stubble thickness detection device to adjust the pressure. Under the conditions of the cutting angle of 90° and forward speed of 3 km/h, the only difference between the two groups was that one group worked with a constant water pressure of 26 Mpa, and the other group worked with a variable pressure, which was equipped with a stubble thickness detection device to adjust the pressure of two levels, 22 Mpa and 26 Mpa.

Then, a Box–Behnken design method was applied to establish the response surfaces for two responses to analyze the relationships between operation parameters (water jet pressure, cutting angle and forward speed) and responses (CR, WC). Optimization analysis was used to determine the recommended operation parameters.

Table 3. Box–Behnken factor level for SCR and WC.

Factor	Unit	Level
		−1	0	1
P	Mpa	18	22	26
α	°	75	90	105
V	km/h	3	4	5

shows the range of each experiment factor. A total of 17 combination experiments were conducted with three replications. The results were evaluated by analysis of variance (ANOVA, 95% confidence interval).

The SCR referred to the ratio of the number of stubble cut off completely after the machine passed to the total number of stubble on the machine’s route. As shown in Figure 7, before the machine was driven into the detection area, 10 m was reserved at the front and back as the preparation zone for the tractor to start, accelerate, turn and stop, which was to ensure that the forward speed in the detection area was uniform. The row space in the zone was 5 m different from the conventional 30–50 cm row spacing, which prevented the tractor from mixing the stubble. In the 80 m detection zone, ten detection areas of 1 m were randomly selected to collect all the stubble on the route cut by the device. The stubble was divided into two categories according to the cutting degree. The number of types of stubble in every detection row was counted with three replications, and the SCR was calculated according to Equation (4):

$$SCR=\frac{S_1+{\omega S}_2}{Sn}\times 100\%$$

(4)

where SCR is the stubble-cutting rate. S₁ is the number of completely cut stubble. S₂ is the number of incompletely cut stubble. ω is 50%, which is the coefficient of incompletely cut stubble. Sn is the total number of collected stubble.

The statistics of WC (L/ha) were obtained through the water level gauge in the water tank after every experiment. The difference between the water level gauge before and after the experiment was the water consumption under that experiment condition. The work diagram of the machine working in the field is shown in Figure 8.

3. Results and Discussion

3.1. EC and MC Measurement Results

Figure 9 shows the measurement results of the EC and MC of the field soil and stubble before spring sowing. It could be clearly seen that during this period, the EC and MC of soil and stubble had a significant difference. The average MC of the soil was 15 ± 2.48%, and the average MC of the stubble was 4.71 ± 2.4%. The maximum value of soil MC was 20.59%, and the minimum value was 8.42%. The maximum value of stubble MC was 12.39%, and the minimum value was 0.04%. The average EC of the soil was 444.20 ± 32.34 μs/cm, and the average EC of the stubble was 72.26 ± 25.54 μs/cm. The maximum value of soil EC was 528.28 μs/cm, and the minimum value was 361.21 μs/cm. The maximum value of stubble EC was 137.19 μs/cm, and the minimum value was 6.04 μs/cm.

Soil EC was affected by SMC [39], SC [40] and organic matter content [44]. The stubble returning to the field also contributed to storing soil moisture [45]. On the other hand, the MC of stubble covering the field was naturally relatively low even in the spring sowing period after being exposed to wind and sun.

3.2. Effect of Experiment Factors on STDE of Stubble Thickness

Since the stubble thickness was converted from the angle value determined by detecting the difference in EC between the soil and stubble, the STDE affected by the EC could be equivalent to the STDE affected by the stubble thickness.

Table 4. Box–Behnken experiment matrix and results for STDE.

Standard	Factors			Response
	SMC/(%)	SC/(MPa)	V/(km/h)	STDE (%)
1	10	0.6	4	30.25
2	30	0.6	4	10.86
3	10	1.8	4	10.24
4	30	1.8	4	9.44
5	10	1.2	3	10.66
6	30	1.2	3	3.73
7	10	1.2	5	28.71
8	30	1.2	5	12.66
9	20	0.6	3	7.61
10	20	1.8	3	6.11
11	20	0.6	5	35.63
12	20	1.8	5	14.29
13	20	1.2	4	18.68
14	20	1.2	4	20.74
15	20	1.2	4	18.39
16	20	1.2	4	16.81
17	20	1.2	4	20.44

demonstrates the result of SMC, SC and V on STDE. The main effects and interactions were derived from ANOVA results (

Table 5. ANOVA for response surface reduced quadratic model for STDE.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	1246.82/1243.24	9/8	138.51/155.40	43.65/48.19	<0.0001 ***
SMC	232.96	1	232.96	73.39/72.33	<0.0001 ***
SC	244.98	1	244.98	77.18/75.96	<0.0001 ***
V	498.96	1	498.96	157.20/154.72	<0.0001 ***
SMC × SC	86.40	1	86.40	27.22/26.79	0.0012 **/0.0008 **
SMC × V	20.79	1	20.79	6.55/6.45	0.0376 */0.0348 *
SC × V	98.41	1	98.41	31.00/30.51	0.0008 **/0.0006 **
SMC2	35.22/36.51	1	35.22/36.51	11.10/11.32	0.0126 **/0.0099 **
SC2	3.58	1	3.58	1.13	0.3234
V2	20.01/20.96	1	20.0120.96	6.30	0.0404 */0.0342 *
Residual	22.22/25.80	7/8	3.173.22
Lack of fit	11.85/15.43	3/4	3.95/3.86	1.52/1.49	0.3380/0.3549
Pure error	10.37	4	2.59
Cor total	1269.04	16

Note: Value before and after the slash is ANOVA for response surface reduced quadratic model for STDE before and after eliminating the insignificant terms. *** shows significance at p < 0.001; ** shows significance at p < 0.01; * shows significance at p < 0.05.

and

Table 6. R-Squared values of STDE statistics analysis.

Std. Dev.	1.78/1.80	R2	0.9825/0.9797
Mean	16.19	Adj R2	0.9600/0.9593
C.V.%	11.00/11.09	Pre R2	0.8379/0.8717
		Adeq Precision	22.9614/24.7547

Note: Value before and after the slash is R-Squared values of STDE statistics analysis before and after eliminating the insignificant terms.

), and the regression model and optimized parameters were obtained. The regression model of STDE was extremely significant (p < 0.001), and the lack of fit of the model was not significant (p > 0.05), indicating that the fitted regression equation was highly reliable and could accurately reflect the relationship among SMC, SC, V and STDE.

As shown in Table 5, all experiment factors, SMC, SC, V and the interaction of SC and V, were significantly (p < 0.001) related to STDE. The interaction of SMC and SC, the interaction of SMC and V and the second-order terms of SMC were significantly (p < 0.01) related to STDE. The interaction of SMC and V and the second-order terms of V were significantly related to STDE (p < 0.05). Table 6 shows that a “Pre R²” of 0.8717 was in reasonable agreement with an “Adj R²” of 0.9593. After removing insignificant terms, the final regression model was as given in Equation (5):

$$\begin{array}{c}STDE=-77.8521+0.6192\times SMC+8.3521\times SC+40.2038\times V+0.7746\times SMC\times SC\\ -0.228\times SMC\times V-8.2667\times SC\times V_1-0.0294\times {SMC}^2-2.2283\times V^2\end{array}$$

(5)

Figure 10 illustrates that STDE decreased with the increase of SMC and SC and increased with the increase of V. The reasons were analyzed as follows: when other factors remained unchanged, the increase of SMC promoted the ion migration rate in the soil, and the EC detection was more accurate, resulting in the reduction of STDE. However, the influence was limited. When SMC reached 30% [46], the continuous increase of SMC caused dilution and had a negative impact on EC detection. Therefore, the range of SMC in this research was set up at 10–30%. When other factors remained unchanged, the increasing SC represented the increasing density of the soil so that the sensor probe was more fully in contact with the soil and the STDE was reduced. When other factors remained unchanged, the increasing V resulted in insufficient contact between the sensor probe and the soil, which in turn led to an increase in STDE. Therefore, the stubble thickness detection device referred to a working condition with a low speed and a sufficient gap between MC of soil and stubble, and SMC should not exceed 30%.

3.3. Effect of Experiment Factors on SCR and WC

Table 7. Effect of stubble thickness detection device.

Stubble-Thickness Detection Device	CR (%)	WC (L/ha)	Block Degree
Not equipped	95.66 ± 5.4	53.46 ± 2.7	3 light blocks
Equipped	94.27 ± 3.9	46.39 ± 1.9	5 light blocks

Note: ± means standard deviation from the original data.

shows the compared experiment with and without a stubble thickness detection device to adjust the pressure. Under the condition that SCR remained unchanged, the WC equipped with a stubble thickness detection device to adjust the pressure was reduced by 13.22%. However, it was undeniable that compared with the no-blocking effect of the ultra-high-pressure waterjet [32], the waterjet pressure of 22–26 MPa had certain limitations. But it could still meet the requirements, and the slight blocking only appeared with the strongest stubble root parts.

Table 8. Box–Behnken experiment matrix and results for SCR and WC.

Standard	Factors			Response
	P/(MPa)	α/(°)	V/(km/h)	CR (%)	WC (L/ha)
1	18	75	4	71.67	32.91
2	26	75	4	82.50	47.45
3	18	105	4	85.57	33.19
4	26	105	4	86.37	48.66
5	18	90	3	88.75	43.56
6	26	90	3	96.75	51.66
7	18	90	5	84.67	28.35
8	26	90	5	90.33	45.35
9	22	75	3	80.75	46.75
10	22	105	3	85.10	47.11
11	22	75	5	72.45	37.31
12	22	105	5	83.20	38.96
13	22	90	4	90.15	42.65
14	22	90	4	90.77	41.99
15	22	90	4	91.66	42.33
16	22	90	4	92.25	43.63
17	22	90	4	90.37	44.35

shows the effect of P, α and V on SCR and WC. The main effects and interactions were derived from ANOVA results (

Table 9. ANOVA for response surface-reduced quadratic model for SCR and WC.

Source	SCR		WC
	F-Value	p-Value	F-Value	p-Value
Model	120.04/150.96	<0.0001 ***	53.75/105.41	<0.0001 ***
P	122.19/119.76	<0.0001 ***	302.39/267.27	<0.0001 ***
α	206.42/202.31	<0.0001 ***	1.22	0.3059
V	81.86/80.23	<0.0001 ***	152.29/134.61	<0.0001 ***
P × α	38.44/37.67	0.0004 **/0.0002 **	0.1722	0.6906
P × V	2.09	0.1913	15.77/13.94	0.0054 **/0.0029 **
α × V	15.65/15.34	0.0055 **/0.0035 **	0.3314	0.5829
P2	0.0908	0.7720	6.29/5.81	0.0405 */0.0329 *
α2	596.96/585.93	<0.0001 ***	3.82	0.0915
V2	6.88/6.68	0.0343 */0.0295 *	1.25	0.3008
Lack of fit	0.5976/0.7199	0.6495/0.6424	1.74	0.2965/0.3115

Note: Value before and after the slash is ANOVA for response surface-reduced quadratic model for SCR and WC before and after eliminating the insignificant terms. *** shows significance at p < 0.001; ** shows significance at p < 0.01; * shows significance at p < 0.05.

and

Table 10. R-Squared values of SCR and WC statistics analysis.

SCR				WC
Std. Dev.	0.8089/0.8171	R2	0.9936/0.9916	Std. Dev.	1.12/1.19	R2	0.9857/0.9723
Mean	86.08	Adj R2	0.9853/0.9850	Mean	42.13	Adj R2	0.9674/0.9631
C.V.%	0.9397/0.9492	Pre R2	0.9612/0.9698	C.V.%	2.66/2.83	Pre R2	0.8610/0.9086
		Adeq Precision	39.8288/43.0403			Adeq Precision	27.4097/36.4431

Note: Value before and after the slash is R-Squared values of SCR and WC statistics analysis before and after eliminating the insignificant terms.

). The regression models and optimized parameters were obtained. The regression model of SCR was extremely significant (p < 0.001), and the lack of fit of the model was not significant (p > 0.05). The regression model of WC was extremely significant (p < 0.001), and the lack of fit of the model was not significant (p > 0.05), indicating that the fitted regression equation had high reliability and could accurately reflect the relationship between P, α, V and SCR, WC.

Table 9 demonstrates that P, α, V and the second-order term of α were significantly (p < 0.001) related to SCR. The interaction term of P and α and the interaction term of α and V were significantly (p < 0.01) related to SCR. The second-order term of V was significantly (p < 0.05) related to SCR. P and V were significantly (p < 0.001) related to WC. The interaction of P and V was significantly (p < 0.01) related to WC. The second-order term of P was significantly (p < 0.05) related to WC. Table 10 shows that the “Pre R²” of 0.9698 and the “Adj R²” of 0.9850 related to SCR, and the “Pre R²” of 0.9086 and the “Adj R²” of 0.9631 related to WC were reasonable. After removing insignificant terms, the final regression models were as given in Equations (6) and (7):

$$\begin{array}{c}SCR=-347.88687+4.55156\times P+8.46667\times \alpha -3.9675\times V-0.041792\times P\times \alpha +0.106667\times \alpha \times V\\ -0.042778\times {\alpha }^2-1.0275\times V^2\end{array}$$

(6)

$$WC=31.19219+3.33458\times P-17.12625\times V+0.55625\times P\times V-0.087214\times P^2$$

(7)

Figure 11a shows the effect of P and α on SCR when V was 4 km/h. SCR increased with the increase of P and increased and then decreased with the increase of α. When P was 18 MPa, as α increased from 75° to 105°, SCR increased from 71.67% to the highest point and then decreased to 85.57%. When P was 26 MPa, as α increased from 75° to 105°, SCR increased from 82.5% to the highest point and then decreased to 86.37%.

Figure 11b illustrates the effect of α and V on SCR when P was 22 MPa. When α was 75°, as V increased from 3 km/h to 5 km/h, SCR decreased from 80.75% to 72.45%. When α was 105°, as V increased from 3 km/h to 5 km/h, SCR decreased from 85.1% to 83.2%.

Figure 11c demonstrated the effect of P and V on WC when α was 90°. WC increased with the increase of P and decreased with the increase of V. When P was 18 MPa, as V increased from 3 km/h to 5 km/h, WC decreased from 43.56 L/ha to 28.35 L/ha. When P was 26 MPa, as V increased from 3 km/h to 5 km/h, WC decreased from 51.66 L/ha to 45.35 L/ha. The change in WC caused by the increase of P was greater than that caused by the increase of V.

The higher P produced a faster waterjet stream, which meant higher momentum and more destruction. The waterjet with higher impact energy was able to cut stubble deeper, with the benefit of increasing SCR. Increasing P also improved the cutting efficiency. Conversely, a faster V caused less exposure time, which decreased the impact energy of the waterjet on stubble. Aiming at increasing the efficiency of agricultural production, which meant a faster V, demanding a higher P to compensate for the less exposure time caused by the faster V. As for WC, it had nothing to do with α; it was only related to P and V. Under the condition that V remained unchanged, the greater the P, the greater the WC. Under the condition that P remained unchanged, the faster the V, the smaller the WC.

Compared with the research by Hu et al. [3], this research found that when P and V remained constant, the optimal α did not always appear in the vertical direction of 90°. As shown in Figure 7, it could be seen that the optimal α was in the range of 90–95°. The reason might be caused by the difference in P. The equipment designed by Hu [3] adopted the range of pressure factor to be 240–280 MPa, which had stronger kinetic energy in the vertical direction and was not easily affected by the horizontal movement of the machine. However, the P range in this study was 18–26 MPa. When the cutting angle was 90–95°, the horizontal movement of the machine helped correct the horizontal force component of the waterjet [46], which was more effective for stubble cutting.

3.4. Optimization Analysis and Verification Experiment

In order to obtain the optimal parameters for the stubble-cutting device, the Design Expert V12 was used to optimize the solution of the constrained objective according to the above experiment results and the regression equations. In order to ensure the minimum STDE, the SCR over 95% and the minimum WC, the objective equations and constraints were set as follows:

$$\left\{\begin{array}{c}STDE\left({SMC}_i,{SC}_i,V_i\right)=min\\ s.t.\left\{\begin{array}{c}10\le {SMC}_i\le 30\\ 0.6\le {SC}_i\le 1.8\\ 3\le V_i\le 5\end{array}\right.\end{array}\right.$$

(8)

$$\left\{\begin{array}{c}SCR\left(P_i{,\alpha }_i,V_i\right)\ge 95\%\\ WC\left(P_i{,\alpha }_i,V_i\right)=min\\ s.t.\left\{\begin{array}{c}18\le P_i\le 26\\ 75\le {\alpha }_i\le 105\\ 3\le V_i\le 5\end{array}\right.\end{array}\right.$$

(9)

The optimal parameters for the STDE were: SMC was 29.91%, SC was 0.84 MPa and V was 3.01 km/h. Under these conditions, the STDE was 3.023%. The optimal parameters for SCR and WC were P, 26 MPa; α, 90.45°; and V, 3.36 km/h. Under these conditions, SCR was 95.49% and WC was 49.98 L/ha.

The two factors (SMC and SC) in the STDE experiment were not controllable in the field and were not applied in the field verification experiment. The forward speed was set the same as the SCR and WC experiment. According to the optimization analysis results, P was set to 26 MPa, α was set to 90.45° and the forward speed was set to 3.36 km/h in the field verification experiment.

Table 11. Results of the verification experiment.

Parameter	Evaluation Indexes
	STDE/%	CR/%	WC/L/ha
Predictive value	3.03	95.49	49.98
Measured value	3.16 ± 0.76	91.76 ± 2.47	49.09 ± 1.89
Relative error	4.11	4.06	1.81

Note: ± means standard deviation from the original data.

shows the results of the verification experiment, and the work performance and optimal parameters are shown in Figure 12. The relative error of STDE between the measured value of the verification experiment and the predicted value of the regression model was 4.27%. The relative error of SCR was 4.07%. The relative error of WC was 1.82%, which was basically consistent with the theoretical optimization results, indicating that the experimental regression model was reasonable.

4. Conclusions

In this study, a pressure-adjustable waterjet stubble-cutting device with a stubble-thickness-detection device was designed, and a prototype was made for the field experiment. Through statistical experiments, the effectiveness of the stubble thickness detection device was confirmed. The parameters of the stubble-cutting device were further optimized by using the response surface method. The main conclusions were as follows.

SMC, SC and V all had a significant impact on STDE of the stubble-thickness detection device. STDE could be smaller under the conditions of higher SMC and SC and lower V. Under the condition that SCR remained basically unchanged, WC equipped with the stubble thickness detection device to adjust pressure was reduced by 13.22%. P, α and V had a significant impact on SCR. P and V had a significant impact on WC. The optimization analysis results showed that when P was 26 MPa, α was 90.45° and V was 3.36 km/h; the optimal result occurred with a 95.49% SCR and a 49.98 L/ha WC. The field verification experiment confirmed that the relative error of STDE between the measured value in the verification experiment and the predicted value of the regression model was 4.27%. The relative error of SCR was 4.07%, and the relative error of WC was 1.82%, which verified the accuracy of the optimization. The research on pressure-adjustable waterjet stubble-cutting devices with a stubble thickness detection device to adjust pressure can provide basic data and reference for the application of waterjet technology and EC detection technology in agricultural material processing.

Waterjet cutting technology has natural advantages in processing agricultural materials, such as fast cutting, high efficiency and no thermal effect. The cutting medium is water, not causing pollution to agricultural materials. The design of adjustable pressure based on the stubble allows for efficient utilization of water resources for large-scale cutting, which is of great significance in areas short of water resources. Future studies are needed to explore the possibility of improving the cutting efficiency and analyzing the disturbance of soil caused by waterjet cutting stubble. For example, studying parameter calibration of stubble and soil under low moisture conditions to better visualize virtual simulations. Another option is establishing a model for the forward speed of waterjet pressure machines and linearly controlling the water pressure through thickness and speed; that is, linearly controlling the waterjet pressure based on the stubble thickness on the soil surface and the machine forward speed.