Design and Exploitation of a Dual-Channel Direct Injection System

Abstract

Soybean–maize intercropping is a traditional yet high-yield cultivation model that faces technical challenges in weed management due to the different herbicide requirements of soybean and maize. This study presents the design and experiments of the innovative dual-herbicide direct injection system, which can simultaneously deliver glyphosate and fomesafen through real-time concentration modulation. The system operates by measuring the relationship between the mixing ratio and the conductivity value, mathematical model, and control algorithm. Experimental validation demonstrated that the correlation coefficient of herbicide mixing ratios and measured conductivity values across pressure ranges of 0.1–0.3 MPa are greater than 0.98, which means that measuring the mixing ratio using conductivity is reliable. Optimal operational performance was achieved at 0.2 MPa spraying pressure, characterized by superior mixing uniformity (CV < 5%) and system stability. This technological advancement provides a practical solution for precision agrochemical application in complex cropping models, with potential applications extending to other crop combinations requiring differential herbicide treatments.

1. Introduction

Soybean (Glycine max (L.) Merr.), as an important food and widely used raw material, is one of the most important crops worldwide [1,2,3,4]. Maize (Zea mays L.) is an important cereal crop worldwide, serving as a vital food, feed, and industrial raw material [5,6,7,8,9]. Soybean–maize intercropping is a traditional planting method that can improve the utilization of resources, such as sunlight and land, enhance soil fertility and stability while increasing yield and contributing to sustainable production [10,11,12,13,14,15,16,17,18].

Weed management in the field is an important part of agricultural production [19]. Weeds can compete with crops for sunlight, water, and nutrients, leading to a loss of 23–64% of crop yield [20,21]. Therefore, herbicides are crucial for controlling weeds [22]. However, maize belongs to the monocotyledon group and soybean belongs to magnoliopsida; in areas where herbicide-resistant genetically modified (GM) soybeans have not been promoted, applying both monocotyledon herbicide and magnoliopsida herbicide can cause injury to soybean or maize while trying to ensure weed control efficiency [23]. Therefore, a system is needed that can simultaneously spray two different herbicides.

In order to adapt to working conditions, several spray systems have been developed that can be adjusted with the working speed to ensure a fixed pesticide application rate. From the perspective of application rate control, these spray systems can be divided into three categories. The first method is total flow-based control of the tank mix. The herbicide is mixed with water in the water tank. Therefore, the concentration of herbicide solution sprayed during work is constant. The second method is chemical flow-based control, and the third method is the combination of chemical and carrier-flow control [24].

The operation of a total flow-based control system is based on changing the nozzle flow rate in proportion to the forward speed, which can maintain a constant acceleration rate. The adjustment of flow rate is achieved by adjusting the nozzle pressure. The approach to chemical flow-based control application is the use of a direct injection system. In a direct injection system, herbicides and water are stored separately. At work, herbicides are added to the carrier at a certain rate and mixed evenly with water before spraying. This increases the flexibility of the mixing system, such as the need to apply higher concentrations of herbicides or different types of herbicides.

In a mixing system, pesticides are concentrated and water mixed in pipelines. The external energy sources of the injection-type mixing system mainly include compressed air and variable displacement pumps [25]. If classified according to the position of the injection port, it can be divided into an injection nozzle-type mixing system and intermediate injection total flow-type mixing system [26,27].

Utilizing dual tanks for pre-diluted herbicide storage on sprayers entails substantial retrofitting costs and operational complexity, which could also bring potential environmental pollution and health risks [28]. To address this limitation, this study applied a direct injection system where two concentrated herbicides interacted with water through controlled contact, ultimately achieving dual-herbicide application through a single sprayer.

Direct-injection technology has the advantages of improving work and operational efficiency, avoiding environmental pollution, avoiding contact between operators and pesticides, and separating water and pesticide storage [29]. It can indirectly reduce the use of pesticides, which is necessary [30]. Direct injection technology includes two parts: pesticide-concentration detection technology and liquid injection technology. Botta et al. (2019) [31] used Raman scattering technology to detect paraquat and methylene blue herbicides. The experimental results showed that the detection limit of methylene blue was 27 ppm, and the detection limit of paraquat was 2.7 ppb. Chen et al. (2022) [32] constructed a ratio fluorescence sensing system based on N-CQDs and nanoclusters, detecting paraquat and thiuram. The experimental results showed that the recovery rate varied from 82% to 114%, and the relative error varied from −14% to 18%. Chauhan and Upadhyay (2021) [33] developed a conductivity detection system to detect arsenic content in soil. The experimental results showed that the system exhibits a linear relationship with different concentrations of arsenic (0.1 ppm~20 ppm), with a detection limit of 5 ppb.

In this study, a kind of dual-herbicide direct injection system was designed to meet the special requirements of a weeding operation in an intercropping plant model. This system can control the mixing ratio of herbicide by real-time conductivity value measurement. At the same time, extensive tests were conducted on soybean–corn intercropping fields to ensure the uniformity and stability of herbicide mixing, thereby verifying the performance of the dual-herbicide direct injection system. Two kinds of herbicide were used in this study: glyphosate for weeding a soybean field, and fomesafen for a maize field.

2. Materials and Methods

2.1. System Design

The direct injection system for a soybean–maize intercropping sprayer consists of the power module (24V, Camel Group Co., Ltd., Hubei, China), machine interface, Programmable Logic Controller (PLC) (CP1H, OMRON Corporation, kyoto, Japan), OLE for Process Control (OPC) server (KEPserverEX6, Dingchen Technelogy, Beijing, China), transmitter and actuator. The block diagram of a direct injection system is shown in Figure 1.

The transmitter consists of two conductivity transmitters, two flow transmitters and two pressure transmitters, which are used to measure herbicide parameters. The PLC is mainly responsible for processing the herbicide parameters measured by the transmitters and transmitting the data to the machine interface and Matlab/Simulink (Matlab 2019a, The MathWorks, Natick, MA, USA) through Ethernet interface. Since the PLC and Matlab/Simulink cannot communicate between each other directly, OPC protocol is used as a bridge to communicate. Therefore, the KEPserver EX6 is used as an OPC server in this direct injection system. The upper computer is mainly responsible for processing the signals transmitted by the PLC through the OPC protocol and transmitting the three Proportional Integral Derivative (PID) parameters processed by the fuzzy control algorithm to the PLC through the OPC protocol. After the PID algorithm is tuned by the PLC, the control commands are transmitted to pulse generation boards one and two. Pulse generation boards one and two convert analog signals into pulse signals and transmit them to driver one and driver two. Then, driver one and driver two control the operation of peristaltic pump one and peristaltic pump two. The human–computer interaction interface can achieve real-time display of detected parameters of glyphosate solution and fomesafen solution, and can control the start and stop of peristaltic pump one and peristaltic pump two. It can also communicate with PLC in real time. The executing mechanism consists of peristaltic pump one and peristaltic pump two. Peristaltic pump one mainly controls the extraction amount of glyphosate solution, achieving a precise mixing ratio of glyphosate solution. Peristaltic pump two mainly controls the extraction volume of the original solution of fomesafen, achieving a precise mixing ratio of fomesafen solution.

2.2. Program Design for a Direct Injection System

PID control is one of the most commonly used control methods in industrial control. However, different operating conditions correspond to different PID parameters, and if the operating conditions change, the original parameters may no longer be applicable. The various methods for PID controller parameter calibration are relatively complex. Fuzzy PID control is an effective control method to solve the problem of traditional PIDs not being able to correct parameters in real time. Compared to PID control, the overshoot is smaller, and the system is more stable when using fuzzy PID control [34]. The transmitter transmits real-time data to the fuzzy controller, which then fuzzifies and infers the data to obtain a fuzzy output. The fuzzy output is then deblurred to obtain the optimal parameter combination for the controller, thus forming the fuzzy PID algorithm. The structure diagram of the fuzzy PID algorithm is shown in Figure 2.

The fuzzy PID controller takes e and ec as its input values, which can complete the task of tuning the three parameters of PID with errors and error change rates under different operating conditions, enabling the controlled object to have better system performance.

This study uses two sets of two-dimensional fuzzy controllers for the direct injection system, with three fuzzy controllers in each set to adjust the proportional band δ, integral time T_I, and derivative time T_D of the PID for maize and soybean in real-time. The input variables of the fuzzy controller are the deviation e and deviation change rate ec between the measured value of the conductivity transmitter and the conductivity value corresponding to the set herbicide mixing ratio, and the output variables are the PID parameter correction value Δδ, ΔT_I, and ΔT_D. The PID parameters after fuzzy reasoning are shown in Equations (1)–(3).

$$\delta ={\delta }^{\prime }+\Delta \delta$$

(1)

$$T_I={T_I}^{\prime }+\Delta T_I$$

(2)

$$T_D={T_D}^{\prime }+\Delta T_D$$

(3)

where δ′, T_I′, and T_D′ are initial parameters of the PID determined by the stable boundary method.

The fuzzy sets of input and output variables selected by the fuzzy controller are all 7, namely negative large, negative medium, negative small, zero, positive small, median, and positive large, that is, {NB,NM,NS,ZO,PS,PM,PB}. The membership function of each fuzzy set adopts a triangular membership function. After analysis, the selected method for resolving ambiguity is the maximum membership degree method. The selected input–output fuzzy domain is [−6, 6], and then the scaling factor and quantization factor are solved according to the actual control requirements. The specific quantification factors and scaling factors for solving are given in the system simulation model.

2.3. Mixing Ratio Calibration Test Design

In order to verify the correlation between conductivity values and herbicides mixing ratios, herbicides with different mixing ratios were tested as the basis for detecting herbicides mixing ratios in the direct injection system for soybean-maize intercropping.

The conductivity value can reflect the electrolyte content in the solution, and the concentration of electrolyte solution can be detected by a conductivity transmitter. Conductive ability of a solution can be represented by conductance G and conductivity k. The herbicide conductance G_X is the reciprocal of resistance R_X, and the conductivity k is the reciprocal of the resistivity ρ. R_X conforms to the law of resistance, which can be calculated by Equation(4), and the conductivity k can be derived by Equation (5).

$$R_X=\rho \cdot {\displaystyle \frac{L}{A}}$$

(4)

$$k={\displaystyle \frac{l}{\rho }}={\displaystyle \frac{L}{A}}\cdot {\displaystyle \frac{l}{R_X}}={\displaystyle \frac{L}{A}}\cdot G_X=K\cdot G_X$$

(5)

where

L = Distance between the two plates of the conductive electrode (cm);

A = Area of the electrode plate (cm²);

K = Electrode constant.

In this study, glyphosate solution with a concentration of 30% (KESAI AGROCHEM HOLDINGS, Shandong, China) and fomesafen solution with a concentration of 250 g/L (Qingdao Fengbang Agrochemical, Qingdao, China) were used as experimental subjects. The calibration test is shown in Figure 3.

In order to reduce the impact of measurement errors in conductivity transmitters and herbicide uniformity errors, herbicide samples with each mixing ratio were tested 10 times, and the average value was taken as the result. Then, the standard deviation and coefficient of variation of the data at each mixing ratio were calculated. There was a sufficient gap between 10 tests to ensure that each test was independent of each other.

To verify the accuracy of the conductivity transmitter for herbicide mixing ratio detection, an accuracy analysis experiment was designed.

Three sets of glyphosate test solutions with mixing ratios of 1.3:100, 1.8:100, and 2.3:100, as well as three sets of fomesafen test solutions with mixing ratios of 0.07:100, 0.12:100, and 0.23:100, were prepared and tested using conductivity transmitters. The conductivity values corresponding to different mixing ratios were obtained, and the mixing ratios obtained from the test calculations were compared with the actual mixing ratios to verify the reliability and accuracy of the conductivity transmitter in detecting mixing ratios.

2.4. Direct Injection System Test Design

The direct injection system test can be divided into the direct injection system uniformity test, direct injection system stability test, and direct injection system field performance evaluation test.

To test the uniformity and stability of the direct injection system when it runs on the soybean–maize intercropping sprayer, this study introduces spatial and temporal coefficients of variation.

The spatial coefficient of variation is intended to describe the degree of difference in the mixture ratio at different spatial positions but at the same time period, which specifically refers to the degree of difference in the mixture ratio of different nozzles in the same time period. The spatial coefficient of variation is intended to describe the degree of difference in the mixture ratio at the same spatial position and continuous time periods. Which specifically refers to the degree of difference in the mixing ratio at different time periods under the same nozzle.

When testing the uniformity of mixed herbicide ratios, the spatial coefficient of variation is used as the evaluation criterion. The soybean–maize intercropping sprayer with direct injection system is shown in Figure 4. The spray machine was developed by the authors based on the spray bar sprayer machine. In the hoods of the spray machine, there are 2, 3, 2, 3 and 2 sprayers from left to right, totaling 12 sprayers. As weed management is mainly carried out at the early growth stage of soybean and corn, soybean is about 20 cm high, and corn is about 40 cm high. The nozzles and hoods of the spray are designed according to the height of the crop. The spray uses a Lechler 90° sector nozzle, and the model is IDK-90-03 (Lechler, Changzhou, China).

The above experiment was conducted after the system mixing ratio stabilizes, approximately 168 s after the system runs.

The herbicide mixture ratio used to calculate the coefficient of variation refers to the average mixture ratio during the time period, rather than the instantaneous mixture ratio.

The test was conducted under the set spray pressure (0.2 MPa, 0.3 MPa, 0.4 MPa) and mixture ratio (glyphosate solution: 1.3:100, 1.5:100, 1.7:100; fomesafen solution: 0.15:100, 0.18:100, 0.21:100). There were 9 tests for each herbicide in the uniformity test, and 9 tests for each herbicide in the stability test.

In the uniformity test, beakers were used to collect herbicide under each nozzle, and the numbers were 1~12 in turn. After the direct injection system stabilized, sampling lasted 10 s each time, and each nozzle was sampled three times. The conductivity value of medicine solution was measured by a conductivity sensor, and the average of the sampling data recorded from each nozzle.

In the stability test, beakers were used to collect herbicide under each nozzle, and the numbers were 1~12 in turn. After the direct injection system stabilized, sampling would last for 10 s each time, and each nozzle was sampled ten times. The conductivity value of medicine solution was measured by a conductivity sensor, and the average of the sampling data recorded from each nozzle after 10 s of system operation.

In the field performance evaluation test, the deposited pesticide concentration on plant leaves was measured to evaluate the uniformity and stability of the pesticide mixing system under field operations, which can better align with practical conditions. The field test setup is shown in Figure 5. Each field performance evaluation test was repeated 4 times on different random plots, and the experimental results were averaged.

To mitigate potential confounding effects caused by weed leaf morphology and surface interactions, a nylon mesh deployment methodology was adopted in this study to collect deposited pesticide samples from weed leaves, thereby isolating the target analytes from biological matrix interferences.

Two spray paths each for glyphosate solution (1.5:100 ratio) and fomesafen solution (0.18:100 ratio) were randomly established with 50 m length. Square nylon meshes with a side length of 2 inches spaced at 5 m intervals were deployed along the paths. Under pressure of 0.2 MPa, the spraying test commenced after system stabilization. After the test, nylon meshes were sealed and stored. In the laboratory, the nylon meshes were placed into 100 mL clean water and agitated sufficiently with a glass rod. Herbicide conductivity was measured using a conductivity transmitter to derive field-applied concentrations.

3. Results

3.1. Mixing Ratio Calibration Test

The conductivity measurements of glyphosate solution at different mixing ratios is shown in

Table 1. The conductivity of glyphosate solution at different mixing ratios.

Mixing Ratio, ${100}^{-1}$	Average Conductivity Value, $\mathbf{\mu }\mathbf{s}\mathbf{·}\mathbf{c}{\mathbf{m}}^{-1}$	Standard Deviation of Conductivity	Coefficient of Variation of Conductivity, %
0.0	317.3	3.77	1.19
0.5	2659.6	21.74	0.82
1.0	5006.8	20.52	0.41
1.5	6909.4	14.99	0.22
2.0	8813.9	10.79	0.12
2.5	10,613.6	8.98	0.08
3.0	12,423.8	10.08	0.08
3.5	14,488.0	10.83	0.07
4.0	16,555.2	29.92	0.18
4.5	18,255.4	26.81	0.15
5.0	19,955.5	25.18	0.13

.

Regression analysis of the experimental data was performed to obtain the standard working curve of glyphosate mixing ratio C₁ and conductivity value σ₁, as shown in Figure 6.

The conductivity measurements of fomesafen solution at different mixing ratios is shown in

Table 2. The conductivity of fomesafen solution at different mixing ratios.

Mixing Ratio, ${100}^{-1}$	Average Conductivity Value, $\mathbf{\mu }\mathbf{s}\mathbf{·}\mathbf{c}{\mathbf{m}}^{-1}$	Standard Deviation of Conductivity	Coefficient of Variation of Conductivity, %
0.0	316.1	7.52	2.38
0.5	346.1	2.21	0.63
1.0	376.4	3.10	0.82
1.5	395.0	2.79	0.71
2.0	415.5	1.96	0.47
2.5	437.3	3.63	0.83
3.0	460.0	3.32	0.72
3.5	484.3	2.83	0.58
4.0	505.0	2.57	0.51
4.5	530.6	2.58	0.49
5.0	555.4	3.10	0.56

.

Regression analysis of the experimental data can obtain the standard working curve of fomesafen mixing ratio C and conductivity value σ, as shown in Figure 7.

Through experiments, the relationship curves between the conductivity and mixing ratio of glyphosate solution as well as fomesafen solution were obtained, and the coefficients of determination R-square were all greater than 0.9, indicating that the conductivity emitter is feasible for detecting the mixing ratio of these two solutions.

The results of the accuracy test are shown in

Table 3. Results of the accuracy test.

Sample Type	Actual Mixing Ratio, ${100}^{-1}$	Calculated Value		Absolute Error of Mixing Ratio, ${100}^{-1}$	Relative Error of Mixing Ratio, %
		Average Mixing Ratio, ${100}^{-1}$	Coefficient of Variation of Mixed Herbicide Ratio, %
Glyphosate solution	1.3	1.326	3.38	0.026	2.00
	1.8	1.788	2.64	−0.012	−0.67
	2.3	2.373	4.05	0.073	3.17
Fomesafen solution	0.07	0.071	9.96	0.001	1.43
	0.12	0.119	6.07	−0.001	−0.83
	0.23	0.231	2.93	0.001	0.43

.

According to Table 3, the coefficient of variation of the glyphosate solution mixing ratio is within 4.054%, and the coefficient of variation of the fomesafen solution mixing ratio is within 9.961%, which basically meets the requirements of the direct injection system. The relative error of the experiment is within ±3.5%, which meets the error requirements for mixed drug ratio detection. This can prove the accuracy and reliability of the conductivity transmitter for detecting two types of herbicides.

3.2. Direct Injection System Uniformity Test

In the uniformity test, the results of the glyphosate solution are shown in Figure 8.

Based on the results, the spatial coefficients of variation for the uniformity test of the glyphosate solution mixture were obtained, as shown in

Table 4. Spatial variation coefficients of uniformity test of the glyphosate solution mixture.

Actual Mixing Ratio, ${100}^{-1}$	Pressure, MPa	Average Mixing Ratio, ${100}^{-1}$	Mixing Ratio Standard Deviation, ${100}^{-1}$	Spatial Coefficient of Variation, %
1.3	0.2	1.32	0.042	3.21
	0.3	1.31	0.042	3.26
	0.4	1.28	0.049	3.82
1.5	0.2	1.50	0.024	1.57
	0.3	1.51	0.029	1.91
	0.4	1.49	0.040	2.69
1.7	0.2	1.71	0.023	1.33
	0.3	1.70	0.032	1.90
	0.4	1.71	0.042	2.47

.

In the uniformity test, the results of the fomesafen solution are shown in Figure 9.

Based on the results, the spatial coefficients of variation for the uniformity test of the fomesafen solution mixture were obtained, as shown in

Table 5. Spatial variation coefficients of uniformity test of the fomesafen solution mixture.

Actual Mixing Ratio, ${100}^{-1}$	Pressure, MPa	Average Mixing Ratio, ${100}^{-1}$	Mixing Ratio Standard Deviation, ${100}^{-1}$	Spatial Coefficient of Variation, %
0.15	0.2	0.150	0.0025	1.64
	0.3	0.151	0.0037	2.45
	0.4	0.150	0.0049	3.24
0.18	0.2	0.180	0.0026	1.47
	0.3	0.180	0.0029	1.63
	0.4	0.180	0.0050	2.77
0.21	0.2	0.210	0.0022	1.03
	0.3	0.209	0.0035	1.68
	0.4	0.210	0.0050	2.38

.

Figure 8 and Figure 9, at different mixing ratios, show that the variation degree of glyphosate and fomesafen solution mixing ratios in each nozzle with a pressure of 0.2 MPa is the smallest, and the variation degree increases with the increase in pressure. Table 4 and Table 5 show that the mixing ratio standard deviation and spatial coefficient of variation in different mixing ratios increase with the increase in pressure. This means that when the spray pressure is 0.2 MPa, the uniformity of this direct injection system is the best.

In Table 4 and Table 5, the maximum spatial coefficient of variation is 3.82%, indicating that the herbicide mixed homogeneously.

3.3. Direct Injection System Stability Test

From the stability test, the results of the glyphosate solution are shown in Figure 10.

Based on the results, the time coefficients of variation for the uniformity test of the glyphosate solution mixture were obtained, as shown in

Table 6. Time variation coefficients of uniformity test of the glyphosate solution mixture.

Actual Mixing Ratio, ${100}^{-1}$	Pressure, MPa	Average Mixing Ratio, ${100}^{-1}$	Mixing Ratio Standard Deviation, ${100}^{-1}$	Time Coefficient of Variation, %
1.3	0.2	1.30	0.029	2.24
	0.3	1.31	0.044	3.39
	0.4	1.31	0.063	4.83
1.5	0.2	1.51	0.029	1.94
	0.3	1.51	0.041	2.69
	0.4	1.51	0.062	4.12
1.7	0.2	1.71	0.027	1.56
	0.3	1.71	0.043	2.49
	0.4	1.72	0.059	3.41

.

From the stability test, the results of the fomesafen solution are shown in Figure 11.

Based on the results, the time coefficients of variation for the uniformity test of the fomesafen solution mixture were obtained, as shown in

Table 7. Time variation coefficients of uniformity test of the fomesafen solution mixture.

Mixing Ratio, ${100}^{-1}$	Pressure, MPa	Average Mixing Ratio, ${100}^{-1}$	Mixing Ratio Standard Deviation, ${100}^{-1}$	Time Coefficient of Variation, %
0.15	0.2	0.151	0.0029	1.94
	0.3	0.151	0.0041	2.69
	0.4	0.151	0.0062	4.12
0.18	0.2	0.179	0.0029	1.63
	0.3	0.181	0.0043	2.39
	0.4	0.182	0.0061	3.36
0.21	0.2	0.209	0.0029	1.40
	0.3	0.211	0.0045	2.15
	0.4	0.211	0.0062	2.95

.

Figure 10 and Figure 11, in different mixing ratios, show that the variation degree of glyphosate and fomesafen solution mixing ratio in each nozzle with pressure of 0.2 MPa is the smallest, and the variation degree increases with the increase in pressure. Table 6 and Table 7 show that the mixing ratio standard deviation and time coefficient of variation in different mixing ratios increase with the increase in pressure. This means that when the spray pressure is 0.2 MPa, the uniformity of this direct injection system is the best.

In Table 6 and Table 7, the maximum time coefficient of variation is 4.83%, indicating that the herbicide mixed uniformly.

3.4. Direct Injection System Field Performance Evaluation Test

The results of the evaluation test are presented in Figure 12.

Based on the results, the coefficients of variation for the field test of the herbicide mixture were obtained, as shown in

Table 8. The results of the field test.

Sample Type	Actual Mixing Ratio, ${100}^{-1}$	Average Mixing Ratio, ${100}^{-1}$	Mixing Ratio Standard Deviation, ${100}^{-1}$	Mixing Ratio Coefficient of Variation, %
Glyphosate solution	1.5	1.47	0.072	4.88
		1.47	0.056	3.83
		1.51	0.069	4.58
Fomesafen solution	0.18	1.77	0.0075	4.24
		1.77	0.0070	3.94
		1.78	0.0085	4.75

.

In Table 8, the maximum mixing ratio coefficient of variation is 4.88%, indicating that the herbicide mixing system is stable and is actually working.

4. Discussion

The different pesticide requirements in the soybean–maize intercropping system has brought different challenges to traditional spraying methods. The aim of this study was to develop a dual-herbicide direct injection system specifically tailored for a soybean–maize intercropping model. By measuring the real-time conductivity of herbicides to determine the real-time mixing ratio, and using fuzzy PID control to achieve real-time adjustment, the precise control of herbicide concentration is ensured. Experiments were carried out to verify the uniformity and stability of the direct injection dual-channel online mixing system, and on-site tests were carried out using our team’s existing self-propelled spray bar sprayer as a test platform to verify the operating performance of the developed system. The following are the main findings of the study.

(1) The conductivity of herbicides is highly linearly correlated with the mixing ratio, which can be used for real-time measurement in operations.

(2) On the basis of the traditional single-channel pesticide mixing system, the dual-herbicide direct injection system is extended to pesticide spraying applications with hoods. We designed a control system and principle for a direct injection system using the fuzzy PID control algorithm. After determining the corresponding relationship between the mixture ratio and the conductivity, we conducted a spray consistency test to verify the feasibility of the direct injection system.

5. Conclusions

To address the weed control problem in soybean–maize intercropping, a direct injection system was designed through research on the functional relationship between the mixing ratio and conductivity value, system structure and mathematical model, control algorithm, hardware design and selection, and software design. This system can apply pesticides to two crops simultaneously, thereby improving the efficiency and convenience of weed control operations, while reducing pesticide waste and pesticide exposure.

The coefficient of variation of the herbicide mixing ratio was within 10%, and the relative error of the mixing ratio calibration test was within ±3.5%, indicating a highly linear correlation between the herbicide mixing ratio and conductivity, and also demonstrating the feasibility of using conductivity to characterize the mixing ratio. At the same time, uniformity and stability tests were conducted to verify the performance of the system. The results showed that the coefficient of variation of the mixing ratio in the tests was less than 5%, indicating that the mixing was uniform, and the system was stable. Finally, field performance evaluation tests were conducted, and the maximum mixing ratio coefficient of variation was 4.88%, indicating that the herbicide mixing system is stable in practical operation.