Design of Dynamic Deep Sowing System for Peanut Planter with Double-Loop Feedback Fuzzy PID Control

Abstract

To enhance peanut sowing depth consistency, an active depth adjustment planter was designed. This study employs inclination and pressure sensors for ridge surface detection, coupled with a hydraulic cylinder and profiling mechanism to dynamically adjust furrow depth according to ground variations. A mathematical model integrating detection, adjustment, and execution processes was established. The control system adopts an improved DLF-Fuzzy PID (double-loop feedback fuzzy PID) control strategy, with co-simulation in MATLAB/AMESIM for performance comparison. The results demonstrate the improved algorithm’s superiority in sowing depth accuracy. Field experiments evaluated three operational parameters (vehicle speed, pressure, and sowing depth) with the qualification rate as the metric. At 50 mm sowing depth and 3 km/h speed, the system achieved a 94.6% dynamic qualification rate and 2.38% maximum depth variation coefficient. Compared with existing methods, this approach enhances sowing depth control effectiveness by 6.05% and reduces variation by 2.85%.

1. Introduction

Peanuts, known for their high nutritional and economic value, are a crucial oil crop in China [1,2]. With the current saturation in cultivation area, the focus has shifted to enhancing yield within the existing acreage. The uniformity of sowing depth significantly impacts both the yield and quality of peanuts [3]. Consistent sowing depth alleviates seedling emergence difficulties, boosts germination rates, and increases yield, thereby serving as a vital strategy for economic growth [4,5,6]. However, at present, there are fewer studies on peanut sowing depth control systems in China, and most peanut planters do not have a sowing depth adjustment function. The development of a peanut sowing depth control system is of necessity and practical significance.

Currently, two main types of control systems on peanut sowing depth control systems are existing in China. One is the mechanical sowing depth adjustment device developed by Chang Xueliang et al. The consistency of sowing depth is achieved by pushing the soil at the top of the ridge surface flat [7]. Geng Duanyang [8] developed a ridge profiling sowing depth control device, which adopts a similar mechanical structure. The other is the sowing depth control system based on fuzzy control developed by Wang Linsheng et al. This system detects the height of the ridge surface through sensors and further regulates the sowing depth using an electric actuator [9]. These two approaches improved the consistency of sowing depth of peanut but still need further improvement compared to other major crops. Research by domestic and international scholars highlights two primary phases for ensuring sowing depth consistency: measurement and execution. Depending on whether or not the measuring device is in contact with the ridge surface, the measurement method can be classified as contact or non-contact measurement. Non-contact methods, as utilized by Søren Kirkegaard Nielsen [10,11,12] and Han Bao [13], involve technologies like ultrasonic or laser sensors for ground height measurement. For instance, Sajad Kiani [14] adjusted sowing depth based on trench depth measurements using ultrasonic sensors. Contact methods, adopted by Jensen Lynndavid [15], Weatherly E.T. [16], Huang Dongyan [17], and Fu Weiqiang [18], typically use angle, displacement, or pressure sensors. Li Yuhuan [19], for example, calculated ground height based on pressure sensor readings from compaction wheels. The execution phase involves researchers like Fu Weiqiang [18] and Xie Yangmin [20,21,22] employing electro-hydraulic or electric systems to control microcontrollers or PLCs (Programmable Logic Controller). Zhao Jinhui [23] utilized displacement sensors to detect movement and adjusted the sowing depth via a hydraulic system, achieving a stability coefficient over 90%. Control strategies, such as PID algorithms and fuzzy PID controls, were applied by Wen Liping [24], Xia Junfang [25], and Qi Wenchao [26] for depth regulation. Wang Can [27] integrated a multi-channel neural network intelligent algorithm with PLC to adjust sowing depth in garlic planters. From the above description, it can be seen that compared with the existing peanut sowing depth control system, the control system for other crops is more deeply researched and has higher control accuracy. However, they are not directly usable for peanut planting due to the difference in sowing depth and plant spacing. Therefore, it is necessary to develop a sowing depth control system tailored to the characteristics of peanuts.

Based on the ridging characteristics of peanut cultivation, this study proposes an active sowing depth control system integrating tilt/pressure sensor detection, electro-hydraulic actuators, and a feedforward-improved double-loop feedback fuzzy PID algorithm [28]. An integrated mathematical model encompassing detection, adjustment, and execution processes was established to address the insufficient adjustment accuracy of traditional mechanical devices and poor adaptability of existing control strategies. The system features a terrain-adaptive profiling mechanism and implements the double-closed-loop control of pressure and actuator height through the enhanced control strategy, with parameter optimization achieved via MATLAB-R2020b/AMESIM-2020.2 co-simulation. Compared to existing methods, the proposed system demonstrates superior performance under operational conditions of 50 mm sowing depth and 3 km/h speed, achieving a 94.6% dynamic qualification rate (a 6.05% improvement over comparable systems) and a maximum depth variation coefficient of 2.38% (a 2.85% reduction). This study realizes the precise control of peanut sowing depth, offering both theoretical references and technical foundations for improving sowing depth consistency in peanut cultivation.

2. Materials and Methods

2.1. Mechanical Structure and Working Principle

2.1.1. The Mechanical Structure of the Sowing Depth Control System

The sowing depth control device of the peanut planter consists of three main components: the detection unit, the execution unit, and the control unit. The detection unit includes a profiling wheel, a profiling wheel frame, a pressure sensor, and an inclination sensor. The execution unit comprises a parallel four-bar linkage, a hydraulic cylinder, a electro-hydraulic proportional valve, a pressure sensor, and a furrow opener. The control unit is composed of a control box. The overall structure of the sowing depth control device is illustrated in Figure 1.

2.1.2. Working Principle of the Sowing Depth Control System

During operation, the system is started and initialized. Control parameters are input via the HMI (Human–Machine Interface). The inclination sensor measures angular changes in the profiling wheel frame. The pressure sensor detects pressure between the profiling wheel and its frame, enabling ridge height detection. Data are sent to the PLC in the control box. The PLC calculates the height difference from the previous unit value. It determines if a step change or feedforward system intervention is needed. The PLC computes and outputs a PWM (Pulse Width Modulation) waveform current value. This controls the electro-hydraulic proportional valve, adjusting the hydraulic cylinder to move the parallel four-bar linkage. Simultaneously, the hydraulic cylinder’s pressure sensor measures pressure and sends data to the PLC. The laser sensor also transmits data to the PLC, providing a reference for the next sowing point. The system records all data and calculates furrow opener performance using predefined formulas. The results are displayed in real time at the HMI. The working principle is shown in Figure 2.

2.2. Key Structures and Working Mechanism

2.2.1. The Design of the Ridge Surface Profiling Mechanism

A working schematic of the detection device is shown in Figure 3. The profiling wheel contacts the ground. When ground elevation changes, the annular pressure sensor captures pressure values. The inclination sensor detects angular changes in the profiling wheel-connecting frame.

In Figure 3, H₁ is the vertical distance between the transverse bar of the detection device and the center of the profiling wheel; L₁ is the center distance between the center of the profiling wheel and the center of the upper center of the inclination sensor; L_T is the vertical distance between the center of the upper center of the inclination sensor and the spring; and H_y is the vertical distance between the center of the profiling wheel with radius R and the contact surface. C_h is the distance between the contact surface of the profiling wheel and the vertical ridge surface of the center line of the profiling wheel, H_L is the distance between the ridge surface, the angle between the profiling wheel and the transverse bar is α, and the angle between the vertical and vertical directions of the profiling wheel is γ. F_DS is the force of the ground on the profiling wheel; G is the gravity of the profiling wheel and the bracket; and F_n is the pressure applied to the detection device at moment n.

The actual measurement of the imitation wheel beam frame is 300 mm from the ground, L₁ is 300 mm, R is 60 mm, and L_T is 200 mm. To detect the height of the monopoly surface H_L, substituting Equations (1) and (2) into Equation (3) is carried out to obtain Equation (4):

$$H_1=L_{1}sin\alpha$$

(1)

$$H_y+C_h={\displaystyle \frac{R}{cos\gamma }}$$

(2)

$$H_L=300-(H_1+H_y+C_h)$$

(3)

$$H_L=300-L_{1}sin\alpha -{\displaystyle \frac{R}{cos\gamma }}$$

(4)

Further, the following was obtained to measure the height of the ridge at that point during the operation of the planter:

$$\begin{array}{r}{H}_{L_n}=300-L_{1}sin{\alpha }_n-{\displaystyle \frac{R}{cos{\gamma }_n}}\\ (n=1,2,3\dots \dots )\end{array}$$

(5)

where α is measured and collected by the inclination sensor and γ is calculated by Equation (9); the relationship between α and γ is

$$\Delta H_{L\mathrm{n}}=H_{L\mathrm{n}}-H_{L\mathrm{n}-1}=(L_{1}sin{\alpha }_{n-1}+{\displaystyle \frac{R}{cos{\gamma }_{n-1}}})-(L_{1}sin{\alpha }_n+{\displaystyle \frac{R}{cos{\gamma }_n}})$$

(6)

We reduce Equation (6) to Equation (7).

$$\begin{array}{r}\Delta H_{Ln}=L_1(sin{\alpha }_{n-1}-sin{\alpha }_n)+{\displaystyle \frac{R\left(cos{\gamma }_n-cos{\gamma }_{n-1}\right)}{cos{\gamma }_{n-1}cos{\gamma }_n}}\\ (n=1,2,3\dots \dots )\end{array}$$

(7)

The value of the initial state is calculated through Equation (8).

$$\left\{\begin{array}{l}{\alpha }_0=\arcsin {\displaystyle \frac{H_1}{L_1}}\\ {\gamma }_0=0\\ H_{L_1}=300-L_{1}sin{\alpha }_0-{\displaystyle \frac{R}{cos{\gamma }_0}}\end{array}\right.$$

(8)

The relationship between pressure and angle in the detection device is discerned via Equations (9) and (10).

$$F_n=G\cos \alpha -{\displaystyle \frac{L_1{F}_{DS}\sin \gamma }{L_T}}=k\Delta x_n=k{L}_T\left(tan{\alpha }_n-tan{\alpha }_{n-1}\right)$$

(9)

$$\begin{array}{r}\Delta F_n=F_n-F_{n-1}=k\left(\Delta x_n-\Delta x_{n-1}\right)=k{L}_T(tan{\alpha }_{n-2}-2tan{\alpha }_{n-1})\\ \left(n=1,2,3\dots \dots \right)\end{array}$$

(10)

The value of the change in the displacement of the profiling device in the vertical direction is

$$\Delta H_N={\displaystyle \frac{\Delta x_n-\Delta x_{n-1}}{\cos {\alpha }_n}}={\displaystyle \frac{\Delta F_n}{k\cos {\alpha }_n}}$$

(11)

In the formula, the following applies:

ΔH_N—The value of the change in the displacement of the profiling device in the vertical direction, m.

The initial values of the variables in Equations (9) and (10) are calculated by Equation (12). F_DS (the ground reaction force) is determined through calibration, involving the vertical damping coefficient and motion damping characteristics of the profiling wheel. Based on rheological theory, the soil is modeled as a nonlinear elastic material. Under the assumption that soil deformation dominates during wheel operation (negligible wheel structural deformation), F_DS can be expressed as

$$\left\{\begin{array}{l}{F}_{n0}=k\Delta x_0=G\cos {\alpha }_0\\ F_{DS}=K_{1}H+C\dot{H}\end{array}\right.$$

(12)

In the formula, the following applies:

K₁—The resistance coefficient of the ditching machine in the vertical direction, N/m.

C—The damping of the trenching machine’s movement in the vertical direction, N/(m/s).

H—The depth of the trenching machine compared to the ridge surface, m.

$\dot{H}$—The first-order time derivative of H, m/s.

Equation (5) is the relationship between the angle measured and collected by the inclination sensor and the height of the ridge surface. Equation (9) is the relationship between the pressure measured and collected by the pressure sensor and the angle measured and collected by the inclination sensor. The definitions of the variables in this chapter are shown in

Table 1. Parameter definition table.

Parameters	Connotation	Unit
H1	the vertical distance between the transverse bar of the detection device and the center of the profiling wheel	m
L1	the center distance between the center of the profiling wheel and the center of the upper center of the inclination sensor	m
R	the radius of the profiling wheel	m
LT	the vertical distance between the center of the upper center of the inclination sensor and the spring	m
Fn	the pressure applied to the detection device at moment n	N
ΔFn	the change value of Fn in the adjacent detection point	N
G	the gravity of the profiling wheel and the bracket	N
FDS	the force of the ground on the profiling wheel	N
Δx0	the initial change value of the spring	m
Δxn	the change value of the spring at moment n	m
k	the elastic coefficient of the spring	-
Hy	the vertical distance between the center of the profiling wheel with radius R and the contact surface	m
Ch	the distance between the contact surface of the profiling wheel and the vertical ridge surface of the center line of the profiling wheel	m
HL	the distance between the ridge surface	m
ΔHLn	the change value of HL in the adjacent detection point	m
ΔHN	the value of the change in the displacement of the profiling device in the vertical direction	m
α	the angle between the profiling wheel and the transverse bar	°
γ	the angle between the vertical and vertical directions of the profiling wheel	°
K1	the resistance coefficient of the ditching machine in the vertical direction	N/m
C	the damping of the trenching machine’s movement in the vertical direction	N/(m/s)
H	the depth of the trenching machine compared to the ridge surface	m
$\dot{H}$	The first-order time derivative of H	m/s

.

2.2.2. Design of Sowing Depth Adjustment Mechanism

Due to variations in peanut rotary tillage equipment or field soil types, the ridge surface may become uneven after ridging, causing ground fluctuations. The furrow opener requires floating adjustment to adapt to these changes. By adjusting the hydraulic cylinder and the position of the parallel four-bar linkage, the system accommodates these variations. The schematic diagram is shown in Figure 4.

In Figure 4, F_YK is the downforce of the hydraulic cylinder on the parallel four-bar linkage rod, N; F_yk is the hydraulic cylinder in the vertical direction component of N; G is the gravity of the floating part, N; φ is the swinging angle of the four-bar connecting rod, °; and θ is the angle of swing of the hydraulic cylinder, °.

The downward pressure of the hydraulic cylinder on the parallel four-bar linkage is in the vertical direction:

$$\left\{\begin{array}{l}{F}_{yk}=F_K=F_{YK}sin\mathsf{\theta }\\ \tan \mathsf{\theta }={\displaystyle \frac{l_1+l_2\sin \mathsf{\phi }}{l_2\cos \mathsf{\phi }}}\end{array}\right.$$

(13)

$$F_D=F_{yk}+G$$

(14)

In the formula, the following applies:

F_K—The force detected by the sensor.

F_D—The downforce of the execution adjustment device on the ground.

The downward pressure of the execution adjustment device on the sensor is

$$F_{yk}=F_{YK}sin\left(\arctan {\displaystyle \frac{l_1+l_2\sin \mathsf{\phi }}{l_2\cos \mathsf{\phi }}}\right)$$

(15)

Because the execution adjustment device acts on the trenching machine with all the underground pressure, the following applies:

$$F_D=F_{YK}sin\left(\arctan {\displaystyle \frac{l_1+l_2\sin \mathsf{\phi }}{l_2\cos \mathsf{\phi }}}\right)+G$$

(16)

The downforce of the execution adjustment device on the ground at moment n can be shown as

$$F_{Dn}=F_{Kn}+G=F_{YKn}sin\left(\arctan {\displaystyle \frac{l_1+l_2\sin {\mathsf{\phi }}_n}{l_2\cos {\mathsf{\phi }}_n}}\right)+G$$

(17)

In the formula, the following applies:

F_Dn—The downforce of the execution adjustment device on the ground at moment n.

F_Kn—The force detected by the sensor at moment n.

Then, we can obtain the value of change in pressure at two neighboring points as

$$\Delta F_{Dn}=\Delta F_{Kn}=F_{Kn}-F_{Kn-1}=F_{YKn}sin{\mathsf{\theta }}_n-F_{YKn-1}sin{\mathsf{\theta }}_{n-1}$$

(18)

By structural design, the initial state of the four-link is horizontal, and the initial value of the above equation can be expressed as

$$\left\{\begin{array}{l}{\mathsf{\phi }}_0=0°\\ {\mathsf{\theta }}_0=\arctan \left(l_1/l_2\right)\\ F_{\mathrm{K}0}=G\\ F_{\mathrm{D}0}=0\end{array}\right.$$

(19)

The value of the change in the displacement of the trenching device in the vertical direction is

$$\Delta H_{Dn}=l_1\left(\sin {\mathsf{\phi }}_n-\sin {\mathsf{\phi }}_{n-1}\right)$$

(20)

In the formula, the following applies:

ΔH_Dn—the value of the change in the displacement of the trenching device in the vertical direction.

The reaction force of the ground on the executive adjusting device comes from the deformation resistance and soil deformation friction when the two are in contact; according to the rheology theory, the soil is approximated to be a nonlinear elastic material, assuming that the deformation variable of the executive adjusting device and the ground is mainly the soil in the process of the operation, and then the ground’s reaction force on the executive adjusting device is

$$F_D=K_{1}H+C\dot{H}$$

(21)

F_D is also obtained by calibration, and the parameters required by its calibration formula are shown in Equation (12). Equation (21) shows the relationship between the pressure between the actuating regulator and the ground, which can be measured by a pressure sensor. The above parameters are defined as shown in

Table 2. Parameter definition table.

Parameters	Connotation	Unit
FYK	the downforce of the hydraulic cylinder on the parallel four-bar linkage rod	N
Fyk	the hydraulic cylinder in the vertical direction component of N	N
G	the gravity of the floating part below the sensor	N
l1	the length of the vertical connecting rod	m
l2	the length of the horizontal connecting rod	m
φ	the swinging angle of the four-bar connecting rod	°
θ	the angle of swing of the hydraulic cylinder	°
FD	the downforce of the execution adjustment device on the ground	N
FDn	the downforce of the execution adjustment device on the ground at moment n	N
FK	the force detected by the sensor	N
FKn	the force detected by the sensor at moment n	N
ΔFDn	the value of the change in pressure at two neighboring points on the ground at moment n	N
ΔFKn	the value of the change in pressure at two neighboring points by the sensor at moment n	N
ΔHDn	the value of the change in the displacement of the trenching device in the vertical direction	m

.

2.3. Hardware Design of Sowing Depth Control System

During field operation, the peanut planter encounters soil agglomeration due to residual crop debris, resulting in uneven ridge surfaces post ridging. This study employs a contact measurement method to precisely detect ridge surface irregularities.

The sowing depth control system hardware consists of three units: detection, execution, and control. The detection unit includes a profiling wheel, profiling wheel frame, pressure sensor, and inclination sensor to monitor ridge height changes in real time. The execution unit comprises a parallel four-bar linkage, hydraulic cylinder, electro-hydraulic proportional valve, and furrow opener to adjust sowing depth based on control signals. The control unit, centered on a PLC, integrates an HMI interface and laser sensor for data processing, parameter input, and system status display. A power module ensures the stable operation of all components.

The pressure sensor and inclination sensor communicate with the PLC via RS485, transmitting ground height data in real time. The PLC calculates height differences and outputs PWM signals to control the electro-hydraulic proportional valve. This drives the hydraulic cylinder to adjust the parallel four-bar linkage, changing the furrow opener depth. The HMI connects to the PLC via Ethernet for parameter setting and status monitoring. The laser sensor measures ground height and provides the difference between the execution and detection units. The hardware connection diagram is shown in Figure 5.

The detection unit is located at the front of the planter. The pressure sensor and inclination sensor are mounted on the profiling wheel frame, with the profiling wheel at the bottom of the frame. The PLC and HMI are installed in an easily accessible position, and the power module is placed near the PLC. The execution unit is positioned in the middle of the planter. The electro-hydraulic proportional valve, hydraulic cylinder, and parallel four-bar linkage are located at the front of the seed box. The furrow opener is placed at the bottom of the execution unit. The hardware parameters of the control system are listed in

Table 3. Hardware parameters of the control system.

Component	Model	Key Parameters	Function
Pressure Sensor	DVLF-102	Range: 0–500 N, Accuracy: 2.0 ± 10%	Measures the pressure of the profiling wheel and the furrow opener
Inclination Sensor	LVT518T	Range: ±90°, Accuracy: ±0.1°	Detects profiling wheel frame angle
Hydraulic Cylinder	ROX28 × 125	Stroke: 60 mm, Force: 1000 N	Drives parallel four-bar linkage
Electro-Hydraulic Proportional Valve	EFBG-03-H	Response Time: <50 ms, Control Accuracy: ±0.1%	Controls hydraulic cylinder movement
PLC	Siemens S7-1214C	Input: 16 Channels, Output: 12 Channels, Communication: RS485	Data processing and control
HMI	Siemens KTP-400	Size: 4 inches, Resolution: 800 × 480	Parameter input and status display
Laser Sensor	BL-400NZ	Measurement Range: 200–260 mm, Accuracy: ±0.8 mm	Measures ground height for feedback

.

2.4. Control Strategy of Sowing Depth Control System

The software of the peanut planter’s sowing depth control system includes data acquisition, processing, control algorithm, and human–machine interaction modules. The data acquisition module reads data from pressure sensors, inclination sensors, and laser sensors via RS485 and analog interfaces. The data processing module filters, calibrates, and calculates height differences. The control algorithm module employs a DLF-Fuzzy PID (double-loop feedback fuzzy PID) strategy to dynamically adjust PID parameters and output PWM signals to control the electro-hydraulic proportional valve. The execution control module drives the hydraulic cylinder to adjust the furrow opener depth. The human–machine interaction module enables parameter setting, status display, and fault alarm through the HMI. All modules work together to ensure precise sowing depth control. The system is programmed using Portal v16 software, with a workflow including initialization, main loop (data acquisition, processing, and control execution), and fault handling (anomaly detection and alarm).

The DLF-Fuzzy PID control strategy uses an outer loop to control height difference and an inner loop to regulate the hydraulic cylinder position. When the height difference from the target depth is significant, a feedforward system is introduced to enhance stability and precision. The response speed of the execution unit directly affects sowing depth consistency during ridge surface fluctuations. A mathematical model is established, with target depth as the factor and execution unit response time and downward pressure as variables, providing theoretical support for the DLF-Fuzzy PID control strategy.

2.4.1. Mathematical Model of DLF-Fuzzy PID Control Strategy

• Mathematical Model Without Step Changes

The actuator uses a fuzzy controller algorithm to obtain a stable pressure output to ensure a consistent target depth if the change in depth between the target depth and the previous target point does not exceed 10 mm. Through Equations (11) and (20), the following can be obtained.

$$\left\{\begin{array}{l}\Delta H_{Dn}=\Delta H_N\\ k{L}_T(tan{\alpha }_{n-2}-2tan{\alpha }_{n-1})=F_{YKn}sin{\mathsf{\theta }}_n-F_{YKn-1}sin{\mathsf{\theta }}_{n-1}\end{array}\right.$$

(22)

2.

Mathematical Model with Step Changes

When the depth change between the target depth and the previous target point is more than 10 mm, a feedforward system is introduced to input a reserve amount of pressure, and the PID controller is used to adjust the algorithm in order to obtain a stable pressure output and to ensure that the target depth is consistent. Through Equations (11) and (20), the following can be obtained.

$$\left\{\begin{array}{l}\Delta H_{Dn}=\Delta H_N+5\\ k{L}_T(tan{\alpha }_{n-2}-2tan{\alpha }_{n-1})=F_{YKn}sin{\mathsf{\theta }}_n-F_{YKn-1}sin{\mathsf{\theta }}_{n-1}+F_b\end{array}\right.$$

(23)

In the formula, the following applies:

F_b—the step pressure input when ΔH_Dn is equal to (ΔH_N + 5 mm).

Based on the change in H_L of the monopoly surface, the value of the amount of change between that point and the previous point is calculated, and the condition for whether a step occurs is set to a difference of more than 10 mm between the two points;

When 0 < ΔH_Ln < 5 mm, i.e., the variation between two points does not exceed 5 mm, PWMi is adaptively adjusted by fuzzy PID;

When 5 < ΔH_Ln < 10 mm, when the change between the two points exceeds 5 mm but does not exceed 10 mm, the feedforward system PWMi is introduced to be adjusted adaptively by fuzzy PID;

When ΔH_Ln > 10 mm, the amount of change between the two points exceeds 10 mm, and it undergoes a step change and is introduced into the feedforward system; PWMi is adjusted using a PID controller.

2.4.2. Design of DLF-Fuzzy PID Control Strategy

DLF-Fuzzy PID combines the advantages of fuzzy PID [29,30] control and conventional PID control, ensuring real-time performance and accuracy through double-loop feedback. When a step change in target depth occurs, the system automatically switches to conventional PID control to avoid overshoot caused by fuzzy PID control. During stable target depth conditions, fuzzy PID control is employed to enhance dynamic response and steady-state precision. Additionally, feedforward control is introduced during step changes to further reduce response time and steady-state error. The depth control program diagram is shown in Figure 6.

DLF-Fuzzy PID monitors and adjusts system states in real time through inner and outer loops. The inner loop ensures rapid response, while the outer loop achieves the precise tracking of the target depth. The pressure of the proportional pressure-reducing valve is regulated by outputting PWM signals of varying frequencies, enabling precise pressure control and ensuring sowing depth stability. By intelligently switching control modes and incorporating feedforward control, DLF-Fuzzy PID significantly improves the system’s dynamic performance and steady-state accuracy, making it suitable for high-precision sowing operations.

3.

Fuzzy PID algorithm

(1) Design of fuzzy PID control system

The pressure value at the target depth is compared with the feedback value from the pressure sensor, and the error e and the error change rate e_c are used as input quantities. The PID parameters are adjusted through the fuzzy PID controller outputs Kp, Ki, and Kd to correct the PWM generation frequency, and at the same time, the ground undulation perturbation information is transferred into the feedforward controller to be analyzed and processed, and the output PWM control signals are used to control the opening size of the proportional valve and the lifting and lowering of the hydraulic cylinder. The current loop is used as the inner loop, with the pressure loop as the output, and the current information feedback and proportional valve information is used as the input to offset the perturbation and ensure the stability of the dynamic performance of the hydraulic cylinder. As inputs are the difference e between the target pressure of the hydraulic cylinder of the actuator and the pressure feedback and the rate of change in the difference e_c, e and e_c are exact quantities, which are fuzzified to obtain the fuzzy quantities E and Ec. The three parametric quantities are ΔKp, ΔKi, and ΔKd, which represent the correction parameters of the controller. The three correction parameters are automatically optimized and adjusted in real time to achieve the self-tuning of the parameters. The depth of peanut planting in Shandong is generally between 4 and 6 cm, and the height variation range of monopoly height is 12–16 cm from Equation (5), and the pressure variation range is 0–800 N from Equation (10), Equation (22), and Equation (23), and the fuzzy theoretical domains of the input variables e and e_c are [−6, 6] through several tests and parameter adjustments.

For the output variables ΔKp, ΔKi and ΔKd, the fuzzy theoretical domains are [8.5, 9.1], [0.2, 0.8] and [0.09, 0.15]. The fuzzy subsets of fuzzy linguistic variables are defined as FA, FB, FC, DL, ZE, ZF, and ZG, meaning Negative Big, Negative Medium, Negative Small, Zero, Positive Small, Positive Medium, and Positive Big. The parameter settings of the fuzzy PID algorithm are shown in

Table 4. Parameter settings of fuzzy PID algorithm.

Input and output variables	e	ec	ΔKp	ΔKi	ΔKd
Fuzzy language variable	E	EC	ΔKp	ΔKi	ΔKd
Fuzzy subset	[FA FB FC DL ZE ZF ZG]
Fuzzy domain	[−6, 6]	[−6, 6]	[8.5, 9.1]	[0.2, 0.8]	[0.09, 0.15]

.

(2) Determination of affiliation function

In this study, the triangular affiliation function is selected for input variables and the Gaussian-type affiliation function is selected for output variables. The fuzzy controller is of the Mamdani type. Its membership function is shown in Figure 7.

(3) Fuzzy control rule setting

The fuzzy rule-based in-depth control system is a set of control rules accumulated by the practical experience and expert knowledge of engineers and technicians. The fuzzy design rules in this paper are jointly analyzed by programmers and field operators and formulated on the basis of professional technology and operational experience, as shown in

Table 5. Control parameter self-tuning rule table.

Kp/Ki/Kd	ec
e	FA	FB	FC	DL	ZE	ZF	ZG
$FA$	$ZG/FA/ZE$	$ZG/FA/FC$	$ZF/FB/FA$	$ZF/FB/FA$	$ZE/FC/FA$	$DL/DL/FB$	$DL/DL/ZE$
$FB$	$ZG/FA/ZE$	$ZG/FA/FC$	$ZF/FB/FA$	$ZE/FC/FB$	$ZE/FC/FB$	$DL/DL/FC$	$FC/DL/DL$
$FC$	$ZF/FA/DL$	$ZF/FB/FC$	$ZF/FC/FB$	$ZE/FC/FB$	$DL/DL/FC$	$FC/ZE/FC$	$FC/ZE/DL$
$DL$	$ZF/FB/DL$	$ZF/FB/FC$	$ZE/FC/FC$	$DL/DL/FC$	$FC/ZE/FC$	$FB/ZF/FC$	$FB/ZF/DL$
$ZE$	$ZE/FB/DL$	$ZE/FC/DL$	$DL/DL/DL$	$FC/ZE/DL$	$FC/ZE/DL$	$FB/ZF/DL$	$FB/ZG/DL$
$ZF$	$ZE/DL/ZG$	$DL/DL/FC$	$FC/ZE/ZE$	$FB/ZE/ZE$	$FB/ZF/ZE$	$FB/ZG/ZE$	$FA/ZG/ZG$
$ZG$	$DL/DL/ZG$	$DL/DL/ZF$	$FB/ZE/ZF$	$FB/ZF/ZF$	$FB/ZF/ZE$	$FA/ZG/ZE$	$FA/ZG/ZG$

.

The parameters (Kp, Ki, and Kd) of the PID controller are adaptively tuned online through fuzzy control rules based on real-time measurements of pressure error (e) and the error rate (e_c). The control strategy operates in three phases: When |e| is large, Kp is maximized with minimized Kd and zero Ki to enhance response speed while preventing derivative-induced overshoot. For intermediate |e| and |e_c| values, all parameters maintain moderate magnitudes. When |e| becomes small, Kp and Ki are increased to improve steady-state accuracy, with Kd appropriately adjusted to suppress oscillations. The adjustment algorithm formulas for Kp, Kd, and Ki are as follows:

$$\left\{\begin{array}{l}Kp=K{p}^{\prime }+\left\{\left.E,E_C\right\}Kp=K{p}^{\prime }+\Delta Kp\right.\\ Ki=K{i}^{\prime }+\left\{\left.E,E_C\right\}Ki=K{i}^{\prime }+\Delta Ki\right.\\ Kd=K{d}^{\prime }+\left\{\left.E,E_C\right\}Kd=K{d}^{\prime }+\Delta Kd\right.\end{array}\right.$$

(24)

In the formula, Kp, Ki, and Kd are PID control parameters, with their initial values ${K_p}^{\prime }$, ${K_i}^{\prime }$, and ${K_d}^{\prime }$ derived from conventional methods. During operation, the system continuously monitors the output, calculates the error e and its rate of change e_c, and fuzzifies them into linguistic variables E and Ec. Based on the fuzzy adjustment matrix, Kp, Ki, and Kd are dynamically adjusted to achieve adaptive fuzzy PID control.

(4) Defuzzification methods

The output of the fuzzy controller in this control system is a fuzzy dataset after inference. To obtain a precise numerical value, defuzzification is required. To ensure real-time performance, this study adopts the centroid method via the offline lookup of the fuzzy controller’s control rule table as the output strategy. The centroid of the fuzzy set is selected as the crisp output, defined by the following function:

$$d{f}_g\left(u\right)={\displaystyle \frac{{\int }_a^{b}u\cdot {\mu }_c(u)du}{{\int }_a^b{\mu }_c(u)du}}$$

(25)

In the formula, df_g(u) is the clear value and μ_c(u) is the closeness function of u.

4.

PID algorithm with introduction of feedforward term

When the ridge surface detection device detects a step change in target depth, the feedforward term is introduced to reduce system response time. The deviation e between the target pressure and the pressure sensor value, calculated based on depth theory, serves as the input. The relationship between input e and output u is defined as follows:

$$u(t)=\Delta K_{p}e(t)+\Delta K_i{\displaystyle {\int }_0^{t}e(t)dt+}\Delta K_d{\displaystyle \frac{de(t)}{dt}}$$

(26)

In the formula, Kp is the proportional coefficient of the system; Ki is the integral coefficient; and Kd is the differential coefficient. The PID parameters are adjusted through trial and error. Finally, the proportional coefficient Kp is determined to be 1.8, the integral coefficient Ki is 1.6, and the differential coefficient Kd is 0.1. A flowchart of the DLF-Fuzzy PID control process is shown in Figure 8.

3. Results

3.1. Simulation Verification and Analysis Based on MATLAB and AMESIM

To validate the algorithm’s advantages, simulation models of double-loop feedback fuzzy PID (DLF-Fuzzy PID), traditional PID, and fuzzy PID control were developed in Matlab-R2020b/Simulink, with a hydraulic simulation model created in Amesim-2020.2 for co-simulation (Figure 9).

The simulation used a sinusoidal signal to mimic ground ridges and a step signal to simulate ground disturbances, replicating peanut planter operations on gentle ridge slopes. Key parameters included a control system sampling time of 0.025 s, proportional solenoid valve current range of 100–2600 mA, hydraulic cylinder total length of 300 mm, piston rod stroke of 60 mm, maximum oil supply pressure of 5 Mpa, and hydraulic cylinder inner and outer diameters of 28 mm and 45 mm, respectively. Partial electro-hydraulic system parameters are listed in

Table 6. Partial electro-hydraulic system parameter settings.

Parameters	Connotation	Value	Unit
m	Armature mass	0.5	kg
K	Electromagnetic force conversion coefficient	160.5	m/s×A
Ky	Force conversion coefficient	470	N×m−1
E	Armature assembly spring stiffness	1 × 104	N×m−1
R	Internal resistance of proportional valve electromagnet	25	Ω
L	Proportional valve electromagnet inductance	6.6 × 10−2	H
Kq	Proportional valve flow gain is sufficient	2.4	-
Kce	Total flow pressure coefficient	4.5 × 10−12	-
M	The load is converted to the equivalent mass on the piston	150	kg
β	Effective bulk modulus of elasticity port	1 × 109	Pa
Vt	Total volume of hydraulic cylinder	6 × 10−4	m3
Ap	Equivalent area of load flow	3.75 × 10−3	m2

.

A co-simulation was conducted 8000 times, with Figure 10a,b presenting partial magnified views of input simulation signals from representative cases. Figure 10a illustrates scenarios with significant differentials between two measurement points on the ridge surface, while Figure 10b demonstrates minimal differentials corresponding to smoother ridge profiles. To validate control performance under both flat and undulating ridge conditions, Figure 10c,d comparatively present results from conventional PID, Fuzzy PID, and the proposed double-loop feedback Fuzzy PID control methods. All three control strategies exhibit adaptive parameter adjustment capabilities in response to ridge profile variations.

During sudden ridge deformation, all strategies demonstrate comparable response times with significant parameter adjustments yet exhibit substantial accuracy discrepancies. Under stable ridge conditions, improved tracking accuracy correlates with prolonged response times across the control schemes. The parameter Δt denotes the deviation between the algorithm’s response time and the prescribed standard duration, reflecting system responsiveness. Δd represents the discrepancy between actual actuator adjustment depth and theoretical requirements, indicating control accuracy and tracking performance. Under operational parameters of 3 km/h sowing speed and 50 mm target depth, the enhanced DLF Fuzzy PID algorithm demonstrates marked improvements in both response speed and control accuracy compared to conventional PID and standard Fuzzy PID implementations. It enhances dynamic performance and stability, making it more suitable for high-precision sowing operations.

3.2. Field Trials

To verify the performance of the peanut planter with a dynamic sowing depth control system, field tests were conducted using the 2MB-1/2 peanut planter manufactured by Shandong Yuanquan Machinery Co., Ltd. (Yishui County, Linyi City, China). The tests used Luhua peanut seeds and were carried out in May 2023 in Zhujiazhuang Village, northwest of Yishui County, Linyi City, Shandong Province. The test plot measured 100 m in length and 40 m in width, with soil primarily consisting of cinnamon soil (brown earth) characterized by a loam or light loam texture, which had undergone rotary tillage treatment. The field trial is shown in Figure 11.

In order to make the comparison test more accurate, the traditional sowing control method was adopted on the peanut planter side, and the rest of the sowing devices were simplified to avoid the deviation of results caused by human factors. On the basis of GB-T6973-2005 Test Methods for Single Seeder (Precision Seeder) [31], the qualification rate of sowing depth and coefficient of variation in sowing depth were tested under the influence of different speeds.

3.2.1. Test Methods

The test parameters were set according to Shandong peanut sowing standards, with a typical sowing depth of 40–60 mm and a working speed of 3–5 km/h. Based on the mechanical plant spacing of the peanut planter, three sowing depths (40 mm, 50 mm, and 60 mm) and three speeds (3 km/h, 4 km/h, and 5 km/h) were tested. To eliminate acceleration and deceleration effects, measurements were taken in the middle area, 20 m from each end of the plot. The ground surface height was calibrated, and the soil was scraped to expose peanut seeds. The distance between the seeds and the ground surface was measured. Six groups of measurements were conducted, with 100 seeds per group. The test data are shown in Figure 12.

3.2.2. Calculation of Test Results

The experimental data of the two different methods were analyzed according to the formula of each calculated parameter, and the results are shown in Table 7.

$$\overline{h}={\displaystyle \frac{{\displaystyle \sum h_i}}{N}}$$

(27)

$$\eta ={\displaystyle \frac{n}{N}}\times 100\%$$

(28)

$$S_h=\sqrt{{\displaystyle \frac{{\displaystyle \sum {\left(h_i-h\right)}^2}}{N}}}$$

(29)

$$V_h={\displaystyle \frac{S_h}{\overline{h}}}\times 100\%$$

(30)

In the formula, H—the sowing depth pass rate, %; n—the number of seeding depth passes, piece; N—the total number of sowing depth measurement points, piece; h_i—seeding depth measurement, mm; S_h—the standard deviation of sowing depth, mm; V_h—the coefficient of variation of sowing depth, %; and $\overline{h}$—mean value of sowing depth, mm.

Table 7. Comparison of results of each test’s data.

Measuring Parameters	Operating Speed/(km/h)			Sowing Depth (mm)			Adjustment Mode
Working speed/km/h	3	4	5	3	3	3	-
Sowing depth/mm	50	50	50	40	50	60	-
Average sowing depth/mm	50.4	47.6	44.04	41.27	50.4	44.32	Dynamic
	46.02	52.9	41.02	46.32	46.02	44.97	Mechanical
Sowing depth pass rate/%	94.6	90.4	86.19	94.03	94.6	92.35	Dynamic
	89.21	83.7	75.78	89.67	89.21	87.52	Mechanical
Standard deviation of sowing depth/mm	2.38	3.74	6.19	2.36	2.38	4.67	Dynamic
	4.53	5.67	7.95	5.02	4.53	5.39	Mechanical
Sowing depth coefficient of variation	3.99	5.76	6.94	5.32	3.99	5.84	Dynamic
	7.94	10.95	14.28	7.06	7.94	12.69	Mechanical

Table 7. Comparison of results of each test’s data.

Measuring Parameters	Operating Speed/(km/h)			Sowing Depth (mm)			Adjustment Mode
Working speed/km/h	3	4	5	3	3	3	-
Sowing depth/mm	50	50	50	40	50	60	-
Average sowing depth/mm	50.4	47.6	44.04	41.27	50.4	44.32	Dynamic
	46.02	52.9	41.02	46.32	46.02	44.97	Mechanical
Sowing depth pass rate/%	94.6	90.4	86.19	94.03	94.6	92.35	Dynamic
	89.21	83.7	75.78	89.67	89.21	87.52	Mechanical
Standard deviation of sowing depth/mm	2.38	3.74	6.19	2.36	2.38	4.67	Dynamic
	4.53	5.67	7.95	5.02	4.53	5.39	Mechanical
Sowing depth coefficient of variation	3.99	5.76	6.94	5.32	3.99	5.84	Dynamic
	7.94	10.95	14.28	7.06	7.94	12.69	Mechanical

3.2.3. Analysis of Test Results

A comprehensive analysis of the 3 km/h and seeding depth at 50 mm, compared with the setting depth, dynamic depth average difference of 0.4 mm, and the traditional average depth of 3.98 mm, was conducted. The qualified rate of dynamic sowing depth was 94.6%, while that of traditional sowing depth was 89.21%. The standard deviation of dynamic sowing depth was 2.38 mm and that of traditional sowing depth was 4.53 mm. At 4 km/h and 5 km/h, the mean sowing depth, pass rate, standard deviation and coefficient of variation decreased when the sowing depth was 50 mm. At 3 km/h, when the sowing depth is 40 mm and 60 mm, and when the sowing depth is 50 mm, the coefficients reach the best results. When the speed rises from 3 km/h to 5 km/h, the performance of the dynamically adjusted peanut planter is better than that of the traditional peanut planter. The complex shows the better performance of the stable and consistent adjustment of the sowing depth of the hydraulic adjustment mode, achieving the purpose of the design.

4. Conclusions

(1)

Compared to traditional peanut planters, this study developed a dynamic sowing depth control system based on an existing planter. It uses hydraulic control to adjust downward pressure and includes a ridge height detection device.

(2)

A dynamic control mathematical model was established, incorporating angle–pressure–depth dynamics. A double-loop feedback fuzzy PID (DLF-Fuzzy PID) control strategy was designed. Co-simulation using Matlab-R2020b and Amesim-2020.2, along with a downward pressure control platform, validated the algorithm. It demonstrated significant advantages over traditional methods.

(3)

Field tests showed that the dynamic sowing depth adjustment outperformed traditional methods at speeds of 3–5 km/h. The dynamic system exhibited stable and reliable control performance. At 5 km/h, the sowing depth qualification rate exceeded 85%, laying the foundation for precision peanut planting.