An Intermittent Fertilization Control System for Fruit Tree Crown Detection

Abstract

In light of the current dearth of strip intermittent fertilization devices in standardized orchards, this study presents the design of an intermittent fertilization control system based on fruit tree crown detection, developed with the objective of meeting the agronomic requirements of strip furrow fertilization in standardized orchards. The initial stage of the process entails the design of the essential components of the fertilization apparatus, followed by the construction of the intermittent fertilization control system. The ultrasonic sensor was employed as the fruit tree crown detection module, and a mathematical model of fertilization speed was constructed to achieve uniform intermittent fertilization. Furthermore, in order to enhance the responsiveness and reliability of the fertilization servo motor, MATLAB Simulink was employed to assess the dynamic performance of the system under disparate control strategies. Ultimately, to validate the simulation outcomes, a field trial was conducted to assess the precision and uniformity of intermittent fertilization. The results demonstrate that the dynamic performance of the system under the fuzzy PID control strategy is optimal, and the coefficient of variation of the fertilizer uniformity of the intermittent fertilization device is less than 7%. The mean effective fertilization rate exceeded 85%, with the primary indices satisfying the agronomic criteria.

1. Introduction

The fruit industry constitutes a significant component of the agricultural sector. China boasts the largest total orchard area and total fruit output in the world on an annual basis, which plays a pivotal role in the provision of the global fruit market [1,2]. According to the most recent statistical data, the total area of orchards in the country is projected to reach 1,301,000 hectares (equivalent to 3,237,000 acres) in 2022, with an estimated yield of 313 million tons of fruit [3]. The application of orchard fertilization technology represents a critical determinant of fruit yield and quality. When employed in a scientifically and rationally sound manner, such technology can optimize the utilization efficiency of fertilizer. Concurrently, it can efficaciously diminish the probability of contamination resulting from the imprudent application of chemical fertilizers, thereby ensuring the sustainable development and ecological equilibrium of orchards [4]. The predominant fertilization techniques currently employed in orchards are universal application, scattered application beneath the canopy, annular furrow application, strip furrow application, radial furrow application, and hole application [5]. Among these methods, strip ditching fertilization is of particular significance in orchard fertilization due to its prolonged effectiveness, enhancement of soil quality, high operational adaptability, and extensive applicability.

The distance between plants of the same species in standardized orchards is typically between 1.5 and 4 m [6]. The traditional continuous strip ditching fertilization method is associated with two key issues: the application of a relatively large amount of fertilizer, and a low utilization rate of the applied fertilizer. The application of intermittent fertilization represents an effective means of addressing the issue of fertilizer wastage.

In the study of intermittent fertilization, the precise control of fertilization amount is of paramount importance. Additionally, the accuracy of fertilization location is a significant factor that must be considered [7]. Bai et al. [8] developed a lightweight dual-mode automatic variable-speed fertilization device and control system for strip fertilization. Liu et al. [9] put forth a servo motor control fertilization scheme based on weighing feedback and developed a variable rate fertilization control system and a machine forward speed detection module. A performance test of the fertilization mechanism controlled by the servo motor yielded results indicating an accuracy of ≥95%. Wang et al. [10] devised a two-level orchard spiral quantitative fertilization distributor and identified the optimal operational parameters, which were subsequently implemented in the fertilization process between wide rows and narrow plants in southern Xinjiang to achieve precise and uniform distribution. Yang, S. [11] et al. proposed a target fertilization method for Chinese wolfberry orchards. This method involves calculating the amount of fertilizer to be applied based on soil nutrients or crop information obtained in real time by infrared photoelectric sensors. Yuan et al. [12] employed Gaussian process regression (GPR) to construct a fertilization control model based on a genetic algorithm (GA) optimization control combination. The fertilizer machine designed by Zhang et al. [13] integrated GPS and GIS to facilitate decision making regarding fertilization, thereby enabling precise, automated variable fertilization through the regulation of the fertilizer machine shaft’s rotation speed. Additionally, scholars have conducted research on fertilization strategies based on diverse PID control methodologies with the objective of minimizing fertilization errors and enhancing fertilization accuracy [14,15,16,17,18,19].

In conclusion, the current standardized orchard precision fertilization technology has reached a high level of maturity in terms of the precise control of fertilization amount, and is capable of accurately supplying the required amount of fertilizer. However, in the process of strip intermittent fertilization, the uniformity of fertilization, specifically the distribution of fertilizer in the soil, directly affects the absorption and utilization of fertilizer by fruit trees. This is a crucial factor influencing the normal growth and yield of fruit trees. To date, there has been a paucity of research in this area. Moreover, the precise control of fertilization position is currently a topic of research, and the implementation method is more complex. Accordingly, this paper presents the design of an intermittent fertilization control system based on fruit tree crown detection. The crown width of the fruit tree was determined by means of an ultrasonic sensor, and precise intermittent fertilization was achieved by adjusting the start and stop of the fertilization motor (as illustrated in Figure 1). The control system is equipped with a mathematical model to control the speed of the servo motor, thereby ensuring uniform fertilization. The system meets the agronomic requirements of the precision and uniformity of intermittent fertilization, and provides an innovative method for the design and optimization of orchard intermittent fertilization control systems.

2. Materials and Methods

2.1. Test Materials

Citrus is the largest fruit category in the world [20], with China ranking first in terms of both planting area and yield [21]. China is responsible for approximately one-third of the global citrus production [22]. It is, therefore, of significant value to select citrus fruit trees(Chongqing, China) as materials for fertilization tests. The planting of 15-year-old citrus trees in southwest China is characterized by a sticky soil, a row spacing of approximately 6 m, a plant spacing of around 3 m, and a height of 2 to 2.5 m for the citrus trees. The agronomic requirements for the application of a base fertilizer in a 15-year-old citrus orchard are as follows [23]: the base fertilizer per plant is approximately 2 kg, the ditching depth is 40 cm, the fertilization method is strip ditching fertilization, and the fertilization is at the crown drip line on one side of the fruit tree.

2.2. The Structure and Operational Principle

The fertilization device primarily comprises three principal components: the fertilization apparatus, the detection apparatus, the human–computer interaction module, and the integrated machine control system. The overall structure diagram is illustrated in Figure 2, and the operational principle is depicted in Figure 3.

The fertilization component was constructed with a concentric double-axis spiral structure. The feeding spiral winch and the discharging spiral winch were positioned within a U-shaped groove of the winch. A vertical double-bearing seat was connected to ensure that the axes were aligned, thereby enabling the independent movement of the two winch shafts. The feed screw auger shaft was affixed to the drive sprocket by a total bond. The discharge screw auger was driven by a servo motor (SDGA-10C21BD, driver model TSDA-C21B, bipolar planetary reducer 80TDF-199070-L2-H, reduction ratio 15. Tode Automation, Shenzhen, China) through a coupling. The detection device primarily comprised an ultrasonic sensor and a Beidou navigation and positioning system, which were responsible for detecting the crown width information and the real-time operation speed of fruit trees, respectively. Two sets of ultrasonic sensors (UC30-21416A, Sick Sensor Intelligence, Changzhou, China) and a Beidou navigation antenna (TAU1312, Allystar Technology, Shenzhen, China) were affixed to a fixed bracket in a vertical orientation. The installation height of the ultrasonic sensor can be adjusted according to the height of the fruit tree in order to facilitate the application of fertilization operations throughout the entirety of the fruit tree’s growth cycle. The control system was based on the STM32f103 microcontroller(Xingyi Electronic Technology, Guangzhou, China) and incorporated the human–computer interaction module (ATK-MD0430, Xingyi Electronic Technology, Guangzhou, China) into the control box, which was located on the side of the fertilizer box.

The operational principle is illustrated in Figure 3. The external diameters of the two spiral auger blades of the fertilization component are identical, yet the pitches are distinct. An increase in pitch size results in enhanced fertilizer delivery efficiency, whereas a reduction in pitch size leads to augmented fertilization stability. Prior to the commencement of the fertilization process, the target fertilization amount for fruit trees was calculated based on the specific characteristics of the trees, the extent of their canopies, and the types of fertilizers utilized. The tractor traction device traverses the designated fertilization route. The fruit tree crown detection module employs an ultrasonic sensor to ascertain the crown information of the fruit trees in real time, while the speed detection module utilizes the Beidou navigation and positioning system to determine the tractor’s operational velocity in real time. The fertilization motor is controlled according to the crown detection information system, with the mathematical model of uniform fertilization established in the control system calculating the corresponding fertilization motor speed based on the real-time operation speed.

(a) The detection of fruit tree crown information is an important part of the intermittent fertilization of fruit trees. The timely and accurate collection of fruit tree crown information is the data basis for parameter control and execution. In this study, ultrasonic sensors are used, which have low cost and high detection accuracy [24]. The technical parameters are shown in

Table 1. Performance parameters of ultrasonic sensor.

Resolution	Repeat Accuracy	Accuracy	Response Time	Switching Frequency	Detonation Output Time	Ultrasonic Frequency
≥0.18 mm	±0.15%	±1%	180 ms	4 Hz	43 ms	120 kHz

.

In order to ascertain the width of the crown of fruit trees, the distance is calculated and uploaded to the data acquisition system in accordance with the time interval (Δt) between the emission and reception of the ultrasound after contact with the crown. The formula for ultrasonic ranging is presented in Formula (1).

$$L=c\mathsf{\Delta }t/2$$

(1)

where L is the detection distance, m; c is the ultrasonic propagation velocity, m/s; and Δt is the time difference in echo generation, s.

The efficacy of ultrasonic sensors in detecting objects is contingent upon the prevailing environmental conditions, with temperature being the primary determinant. To guarantee the precision of the measurement outcomes, a temperature sensor is integrated into the ultrasonic sensor system to ascertain the temperature data and calibrate the output distance value. The discrepancy in ultrasonic transmission speed resulting from temperature fluctuations is rectified through the implementation of temperature compensation.

(b) In the process of detecting fruit tree crown width information, as the detection distance of the ultrasonic sensor increases, the accuracy of the detection will concomitantly decline to a certain extent. The optimal detection distance of the ultrasonic sensor utilized in this study is 340–3400 mm, with the maximum detection distance extending to 3400–5000 mm. It is essential to adjust the installation position of the ultrasonic sensor in a manner that ensures the distance between the sensor and the canopy of the fruit tree is consistently maintained within the optimal detection range of the sensor. In order to circumvent the detection of the other canopy, the effective detection distance is set at 400–2000 mm, contingent upon the diameter of the canopy and the recognition range of the ultrasonic sensor.

(c) The human–computer interaction module is a high-performance 4.3-inch LCD resistive touch screen which enables the user to set the target fertilization amount and adjust the fertilization parameters. Additionally, the display interface allows for the real-time presentation of multiple indicators, including the current vehicle speed and fertilization amount.

2.3. Control System Main Program

The fertilization device is linked to the control system via the ultrasonic sensor, which is used to ascertain the crown width of the fruit tree. The fertilization motor is controlled to start and stop based on the crown width detection information, ensuring that the fertilization device is only activated within the crown range and that the fruit tree interval is not fertilized. Once the target fertilization amount has been set through the human–computer interaction module, the control system calculates the servo motor speed based on the real-time operational speed of the machine, as detected by the operation speed detection module. This enables the control of the fertilization motor’s rotation speed, facilitating the adjustment of the fertilization amount, and enhancing the accuracy and utilization rate of the fertilizer, while ensuring the uniformity of fertilization. The comprehensive control strategy of the fertilization control system is illustrated in Figure 4.

The target fertilization amount is determined by the human–computer interaction module, and the ultrasonic transmitter continuously emits the ultrasonic pulse and initiates the timing process. In the event that the ultrasonic pulse fails to detect obstacles (i.e., the canopy) within the designated scanning range or encounters a reflected echo that exceeds the effective detection distance, the fertilization motor is not initiated. If the pulse reaches the canopy of the fruit tree within the effective detection distance, it will reflect and receive an echo. The timer is then halted in order to ascertain the time difference, which is subsequently calculated according to the ranging formula and uploaded to the human–computer interaction module. The Beidou navigation module is continuously engaged in the search for the character “GNRMC”. In the event that the character is searched, the operating speed is subjected to analysis and conversion in order to obtain the operating speed information of the machine. In the event that the character is not located, the message “No signal!” should be displayed. The control system has acquired the crown width detection data and the operational speed of the fruit tree. The fertilization motor is activated and deactivated in order to facilitate the fertilization of the fruit tree within the designated crown range while preventing the fertilization of the tree’s inter-row space. Concurrently, the velocity of the servo motor is calculated and regulated to enhance the precision and efficiency of fertilization while maintaining uniformity in the distribution of nutrients.

2.3.1. Fertilization System Communication Method

The main work of the fertilization control system includes fruit tree crown information detection, machine operation speed acquisition, and human–computer interaction module input and output. In addition, there is a fertilization motor control method based on the above information. In the following, the communication method of the control system is introduced in combination with the overall block diagram of the fertilization control system shown in Figure 5.

The fruit tree crown width detection module subroutine is responsible for the real-time detection of crown information by the ultrasonic sensor. These data are then transmitted to the control system via the IO-Link protocol. The Beidou navigation module is tasked with obtaining the operating speed information, which is achieved through the parsing of the NMEA-0183 protocol. The STM32 microcontroller and the servo motor driver are controlled by RS485 communication, with internal operation based on the MODBUS RTU communication protocol. Subsequently, the fertilization device subroutine, upon the receipt of the aforementioned crown detection information and operating speed data, proceeds to regulate the initiation and cessation of the fertilization motor, as well as its operational velocity. The display interface for the human–computer interaction module is driven by an 8080 parallel port, which is capable of receiving and displaying collected information, including real-time operational speed, fruit tree crown detection data, and the current operational conditions.

2.3.2. Mathematical Model of Uniform Fertilization

In the context of orchard fertilization operations, the evaluation of fertilization effects is contingent upon the assessment of fertilization uniformity. The total quantity of fertilizer necessary for the target fruit tree is calculated based on the tree’s growth parameters. In accordance with the agronomic requirements and field investigation, a correlation table was constructed to determine the optimal fertilization amount for citrus fruit trees at 15 years of age. The specific fertilization parameters are shown in

Table 2. Correlation table of target fertilization amount.

Fruit Species	Tree Age (Year)	Crown Area (m2)	Fertilizer Types	Target Fertilization Amount m(kg)
Ehime Orange	1~5	0.385	Compound fertilizer	0.15
			Bio-organic fertilizer	0.4
			Yard manure	1.5
	6~10	0.636	Compound fertilizer	0.18
			Bio-organic fertilizer	0.5
			Yard manure	1.7
	11~15	0.785	Compound fertilizer	0.2
			Bio-organic fertilizer	0.6
			Yard manure	2

.

In accordance with the principles of citrus fruit tree planting agronomy [25] and the general growth law of fruit trees, the theoretical fertilization amount of the fertilization device was calculated. In order to ensure the precise control of the bio-organic fertilizer application amount, the mathematical model of uniform fertilization takes into account comprehensive data regarding the fruit tree crown width, the technical parameters of the fertilization device, and the operational data of the tractor. The specific calculation method is as follows:

The spiral axial projection area A₁ of the screw auger is calculated, and the calculation formula is shown below (2).

$$A_1={\displaystyle \frac{{\pi D}^2}{4}}$$

(2)

where D is the nominal diameter of the discharge spiral auger, m;

The axial conveying velocity v (m/h) of the fertilizer in the auger [26] is shown in Formula (3).

$$v=S{n}_e$$

(3)

where S is the screw pitch of the screw auger, m; n_e is the rotational speed of the discharge spiral auger, r/h.

According to Formulas (2) and (3), the fertilizer conveying volume Q_v of the spiral auger is calculated as shown in Formula (4):

$$Q_v=A_1\times v\times \phi ={\displaystyle \frac{\pi }{4}}\phi D^{2}S{n}_e$$

(4)

where Q_v is the volume of fertilizer transported by spiral auger in unit time, m³/h. φ is the filling coefficient, and φ is 0.33;

According to the mass calculation formula, the fertilizer conveying capacity Q of the spiral auger is the product of the fertilizer conveying volume Q_v and the fertilizer density ρ. The calculation formula is shown in Formula (5):

$$Q=\rho Q_v\epsilon =\pi \rho \phi D^{2}S{n}_e\epsilon$$

(5)

where Q is the amount of fertilizer transported by the spiral auger, kg/h; ρ is the bulk density of the material to be transported, kg/m³; and ɛ is the inclined conveying coefficient, because the discharge spiral auger is placed horizontally, which is taken as 1.

When the unit time is hour, the amount of fertilizer transported by the spiral auger Q is equivalent to the amount of fertilizer discharged per unit time q (kg/h). As shown in Formula (6):

$$q=Q=\pi \rho {\phi D}^{2}S{n}_e$$

(6)

With known unit time (h), discharge screw auger fertilizer q, and target fertilization fruit tree crown area A, the calculation of fruit tree target fertilization m is as shown in Formula (7), according to the empirical formula calculation:

$$\begin{array}{c}m=q\times {\displaystyle \frac{\sqrt{4A/\pi }}{v_t}}={\displaystyle \frac{\sqrt{4A{q}^2/\pi }}{v_t}}\end{array}$$

(7)

where v_t is the operating speed of the machine that is fed back to the control system in real time, m/h.

Combined with Formulas (6) and (7), according to the target fertilization amount and crown area, the rotation speed of the spiral auger is calculated as shown in Formula (8).

$$n_e={\displaystyle \frac{m{v}_t}{3.54\phi D^{2}S\sqrt{A}}}$$

(8)

where m is the target fertilization amount, kg; A is the crown area of the target fruit tree, m²; and the crown area A of the fruit tree corresponds to the target fertilization amount m.

Formula (8) is used as the speed control rule of the fertilization motor. The control system obtains the data writing value f according to the target fertilization motor speed, as shown in Formula (9).

$$n_e={\displaystyle \frac{f}{8192}}\times 3000$$

(9)

The decimal writing value f is converted into 16-digit speed control data, and the speed of the fertilization motor is adjusted to achieve uniform fertilization.

2.3.3. Servo Motor Speed Control Algorithm

A stable control system must meet the following requirements: (1) strong anti-interference ability and minimal fluctuation; (2) excellent control accuracy and minimal speed deviation; (3) rapid response time and minimal overshoot [27]. The MATLAB Simulink software (R2022a) is used to simulate the traditional control system (see Figure 6) and the fuzzy control system (see Figure 7), and to compare the dynamic performance of the two PID control methods.

The traditional PID controller transfer function and the corresponding control equation algorithm are presented as follows:

$$G\left(s\right)=K_p+{\displaystyle \frac{K_i}{s}}+K_{d}s$$

(10)

$$u\left(t\right)=K_p\cdot e\left(t\right)+K_i\cdot {\int }_0^t e\left(t\right)dt+K_d{\displaystyle \frac{de\left(t\right)}{dt}}$$

(11)

where s is the complex variable in the Laplace transform, which is used to convert the time domain signal into the frequency domain signal. u(t) is the output signal of the PID controller and the control input of the controlled object. e(t) is the deviation signal, which is equal to the difference between the set value (target value) and the actual output value of the controlled object. K_p is the proportional gain, which determines the strength of proportional control. K_i is the integral gain, which determines the strength of the integral control action. K_d is the differential gain, which determines the strength of differential control.

3. Experiments and Results

3.1. Analysis of Simulation Test

The physical model of the vertical servo motor system is approximated by the Laplace transform to a second-order system [28]. Based on the technical parameters of the fertilization motor, the transfer function is established as shown in Formula (12).

$$G\left(s\right)={\displaystyle \frac{\theta \left(s\right)}{u\left(s\right)}}={\displaystyle \frac{4000}{s^2+800s+133}}$$

(12)

The transfer function represents the relationship between the armature voltage of the motor and the output angular velocity, which is employed as the correction object of the PID controller.

The initial PID parameters established by the fertilization motor are imported into the design module via the self-tuning module optimization. The conventional PID parameter design module is illustrated in Figure 8a. A fuzzy PID control model is constructed on the foundation of the traditional PID, and the resulting fuzzy PID design module is illustrated in Figure 8b. The selection of the input language quantity entails a bias, designated by the symbol e, and a bias rate, represented by the symbol ec. The output language variables are ΔK_p, ΔK_i, and ΔK_d. A fuzzy controller with two inputs and three outputs is constructed. The fuzzy set of linguistic variables is defined as {NB, NM, NS, 0, PS, PM, PB}. Fuzzy sets are classified into seven categories, each corresponding to a specific range of values: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large.

According to the above principles and expert experience, a fuzzy control rule table for five parameters can be obtained, as shown in

Table 3. Fuzzy control rule table.

Kp\Ki\Kd		ec
		NB	NM	NS	ZO	PS	PM	PB
e	NB	PB\NB\PS	PB\NB\NS	PM\NM\NB	PM\NM\NB	PS\NS\NB	ZO\ZO\NM	ZO\ZO\PS
	NM	PB\NB\PS	PB\NB\NS	PM\NM\NB	PS\NS\NM	PS\NS\NM	ZO\ZO\NS	NS\ZO\ZO
	NS	PM\NB\ZO	PM\NM\NS	PS\NS\NM	PS\NS\NM	ZO\ZO\NS	NS\PS\NS	NS\PS\ZO
	ZO	PM\NM\ZO	PM\NM\NS	PS\NS\NS	ZO\ZO\NS	NS\PS\NS	NM\PM\NS	NM\PM\ZO
	PS	PS\NM\ZO	PS\NS\ZO	ZO\ZO\ZO	NS\PS\ZO	NS\PS\ZO	NM\PM\ZO	NM\PB\ZO
	PM	PS\ZO\PB	ZO\ZO\NS	NS\PS\PS	NM\PS\PS	NM\PM\PS	NM\PB\PS	NB\PB\PB
	PB	ZO\ZO\PB	ZO\ZO\PM	NM\PS\PM	NM\PM\PM	NM\PM\PS	NB\PB\PS	NB\PB\PB

.

The characteristic curves corresponding to the input and output variables are shown in Figure 9.

The result of fuzzy reasoning is the fuzzy quantity of ΔK_p, ΔK_i, and ΔK_d, which is the matrix. However, a precise quantity is required for the actual control process [29,30]. Accordingly, this paper employs the center of gravity method to defuzzify the operation, thereby converting the fuzzy amount into an accurate one.

As shown in Figure 10, in the Simulink of MATLAB, three distinct simulation models have been constructed, representing the absence of PID control, traditional PID control, and fuzzy PID control, respectively. The models are illustrated in Figure 10. The amplitude of the step signal is set to 1, the duration of the simulation is 2 s, and the system sampling time is 0.1 s.

The step response curves of the three control methods are obtained by simulation based on the step response curve of the target speed of the motor, as illustrated in Figure 11.

In the absence of a PID adjustment control state, the time required for the control system to reach stability is approximately 1.4 s, and the steady-state value is less than the target value. In the traditional PID control algorithm and PID parameter self-tuning state, the PID adjustment parameters are K_p = 6.34, K_i = 37.25, K_d = 0.04, and N = 122.35 (N is the filter coefficient). The adjustment time of the speed control system is reduced to approximately 0.63 s, and a certain degree of overshoot is observed during the start-up phase, which does not exceed 20%. The steady-state value is in accordance with the target value, and there is no recurrent oscillation. A fuzzy PID automatic parameter adjustment algorithm is implemented based on the traditional PID parameter adjustment. The speed control system reaches a steady state in 0.11 s, and the steady-state value is consistent with the target value. The response is relatively rapid, with minimal overshoot.

The simulation results depicted in the figure indicate that among the three controllers, the fuzzy controller exhibits the most optimal performance, followed by the traditional PID. Conversely, the motor running state is not optimal when there is no PID control, which is unable to meet the requisite standards. In the simulation test, the fuzzy PID control algorithm demonstrated superior performance in terms of response time and overshoot when compared to the initial traditional PID control algorithm for the fertilization motor driver. The fuzzy PID control system demonstrates a rapid response time and minimal overshoot, thus meeting the requisite specifications for use in the speed control system of the fertilization motor.

3.2. Field Verification Test

In order to test whether the intermittent fertilization device and control system for fruit tree crown detection meet the requirements of orchard strip ditching fertilization agronomy and fruit tree fertilization operation, fertilization accuracy verification test and uniformity verification test were carried out, respectively.

3.2.1. Verification Test of Accuracy Fertilization

A precision verification test of fertilization was conducted at the Citrus Research Institute of Southwest University (Figure 12). Firstly, the parameters were established: the screw pitch of the screw auger was set at 40 mm, and the working speed was set at 0.281 m/s. In accordance with the agronomic requirements, two groups were initially established, with the fertilization amounts set at 1.5 kg and 2 kg, respectively. To comprehensively evaluate the overall performance of the fertilization control system, the selection range of fertilization amounts was expanded to include a 1 kg group. Once the device had completed the fertilization operation, five fruit trees were randomly selected from the fertilizer ditch, and the length of fertilization and the offset distance of the fruit tree crown center were recorded for each test. The offset distance of the fruit tree center is defined as the distance between the center of the actual fertilization distribution and the center of the fruit tree trunk in the horizontal projection direction. The length and offset of fertilization were measured with a measuring tape, taking the center of each fruit tree as the point of reference.

The test data are shown in

Table 4. The 1 kg fertilizer group.

Test Number	Group 1
	Crown Width/cm	Fertilization Length/cm	Fertilization Offset/cm	Effective Fertilization Length/cm	Effective Fertilization Rate
1	125	120	8	112	89.60%
2	145	138	15	123	84.83%
3	152	141	15	126	82.89%
4	170	172	0	172	100.00%
5	180	166	16	150	83.33%
Average	154.4	147.4	10.8	136.6	88.13%

,

Table 5. The 1.5 kg fertilizer group.

Test Number	Group 2
	Crown Width/cm	Fertilization Length/cm	Fertilization Offset/cm	Effective Fertilization Length/cm	Effective Fertilization Rate
1	119	102	9	93	78.15%
2	152	155	8	147	96.71%
3	155	135	13	122	78.71%
4	170	168	4	164	96.47%
5	171	161	11	150	87.72%
Average	153.4	135.2	9	135.2	87.55%

and

Table 6. The 2 kg fertilizer group.

Test Number	Group 3
	Crown Width/cm	Fertilization Length/cm	Fertilization Offset/cm	Effective Fertilization Length/cm	Effective Fertilization Rate
1	165	150	11	139	84.24%
2	165	160	17	143	86.67%
3	173	160	21	139	80.35%
4	186	182	7	175	94.09%
5	192	178	23	155	80.73%
Average	176.2	166	15.8	150.2	85.21%

, and the data records of the field experiment are analyzed to calculate the effective fertilization rate.

As demonstrated in Table 4, Table 5 and Table 6, when the fertilization amount is set to 1 kg, the lowest effective fertilization rate is 82.89%, the highest effective fertilization rate is 100.00%, and the average value is 88.13%. In the 1.5 kg group, the lowest effective fertilization rate was 78.15%, the highest effective fertilization rate was 96.71%, and the average value was 87.55%. In the 2 kg group, the lowest effective fertilization rate was 80.35%, the highest effective fertilization rate was 94.09%, and the average value was 85.21%. As illustrated in Figure 13a, the table reveals a notable fluctuation in the effective fertilization rate.

To ascertain the stability of fertilization efficiency, the standard deviation and coefficient of variation of the three groups of fertilization experiments were calculated and are presented in Figure 13b. As the set fertilization amount increased, the mean effective fertilization rate exhibited a downward trend. The mean effective fertilization rate for the 1 kg group was the highest, and the standard deviation and coefficient of variation were moderate. The effective fertilization rate of the 1.5 kg group was slightly lower than that of the 1 kg group, and the standard deviation and coefficient of variation were the most pronounced. In comparison to the 1.5 kg group, the average effective fertilization rate of the 2 kg group exhibited a 2.34% decline. However, the standard deviation and coefficient of variation were observed to be the lowest.

3.2.2. Verification Test of Uniform Fertilization

The fertilization uniformity verification test was conducted at the purple soil teaching and research base of Southwest University (Figure 14a). The fertilization apparatus was affixed to the tractor via a three-point suspension system at a consistent velocity, with the forward speed set at 0.281 m per second. According to the corresponding agronomic requirements, the fertilization amount of 15-year-old citrus fruit trees needs to be greater than 1 kg, so the target fertilization amount is set to be 1.5 kg and 2 kg. In accordance with the established fertilization test standard [31], a total distance of 30 m was designated for the test, with 10 m allotted for start-up and end buffer areas. Upon the completion of the aforementioned test, the remaining 10 m were divided into 30 measurement points, with an interval of 10 cm between each point for the purpose of data acquisition. The grid is employed to capture the fertilizer that falls within the designated areas, and following its removal, a precision electronic scale (JM-BL5003) is utilized to ascertain its weight. Once all the designated areas have been collected and tested, the average amount of fertilization and its uniformity will be calculated and analyzed. The formulas are presented in (13)–(15).

$$\overline{x}={\displaystyle \frac{\sum _{i=1}^n x_i}{n}}(i=\mathrm{1,2}\cdots ,30)$$

(13)

$$S=\sqrt{{\displaystyle \frac{\sum _{i=1}^n {\left(x_i-\overline{x}\right)}^2}{n-1}}}(i=\mathrm{1,2}\cdots ,30)$$

(14)

$$CV={\displaystyle \frac{S}{\overline{x}}}\times 100\%$$

(15)

where x is the average amount of fertilizer, g; n is the number of tests; x_i is the quality of fertilizer collected each time, m; S is the standard deviation of fertilizer uniformity; and CV is the coefficient of variation of fertilizer.

The relevant test results are recorded in detail in

Table 7. Table of experimental data for 1500 g of fertilizer applied at the target rate.

Experiment Number	The Amount of Fertilizer/g	Experiment Number	The Amount of Fertilizer/g	Experiment Number	The Amount of Fertilizer/g
1	155.12	11	153.34	21	154.09
2	157.725	12	155.245	22	154.79
3	151.175	13	154.5	23	149.075
4	155.365	14	150.74	24	150.79
5	153.355	15	154.095	25	149.225
6	151.78	16	156.095	26	149.83
7	151.51	17	148.61	27	157.935
8	153.295	18	148.195	28	155.6
9	153.175	19	159.57	29	153.945
10	151.335	20	148.69	30	154.06
Average	153.3835		152.908		152.934

and

Table 8. Table of experimental data for the target amount of 2000 g of applied fertilizer.

Experiment Number	The Amount of Fertilizer/g	Experiment Number	The Amount of Fertilizer/g	Experiment Number	The Amount of Fertilizer/g
1	206.56	11	206.695	21	195.16
2	202.9	12	207.065	22	196.295
3	206.48	13	209.705	23	209.775
4	199.065	14	214.065	24	199.455
5	208.16	15	215.15	25	206.29
6	195.57	16	211.105	26	212.845
7	206.4	17	209.095	27	200.725
8	204.095	18	199.06	28	210.105
9	200.69	19	197.66	29	200.66
10	206.61	20	196.455	30	198.24
Average	203.653		206.6055		202.955

.

As indicated in Table 7 and Table 8, when the target value is 1500 g, the standard deviation measured per 10 cm is 2.9467, and the coefficient of variation of fertilizer uniformity is 1.92%. When the target value is 2000 g, the standard deviation measured per 10 cm is 5.8292, and the coefficient of variation of fertilizer uniformity is 2.85%. All of the aforementioned values were found to be below 7%, thus meeting the requisite standards for orchard ditching and fertilization operations. The fertilization uniformity coefficient of variation trend is illustrated in Figure 15.

4. Conclusions

The conventional continuous strip ditching fertilization technique employed in standard orchards is beset with challenges, including the wastage of fertilizer, the lack of precision in the application of the existing intermittent fertilization devices, and the prevalence of uneven soil nutrient distribution. An intermittent fertilization control system based on fruit tree crown detection was designed and implemented. The fertilization device, which is equipped with an intermittent fertilization control system, was subjected to a trial production phase, after which its performance was evaluated. The principal findings can be summarized as follows:

(1) An intermittent fertilization control system for fruit tree crown detection was designed. The fruit tree crown detection module, working speed detection module, and human–computer interaction module were selected for construction. The software scheme of the fertilization control system, based on the STM32 single-chip microcomputer, includes the following components: fruit tree crown detection, machine forward speed acquisition, and motor drive control strategy. Collectively, these components enable the system to achieve precise, intermittent, and uniform fertilization.

(2) A Simulink simulation model of the fertilization motor was constructed for the purpose of investigating the impact of control strategies, including non-PID, traditional PID, and fuzzy PID, on the speed of the servo motor. The results demonstrated that a servo motor speed control strategy based on fuzzy theory is an effective approach for enhancing the precision of fertilization positioning and the stability of fertilization quantity throughout the fertilization process.

(3) The results demonstrated that within a specified range, the coefficient of variation of fertilizer uniformity for the intermittent fertilization device equipped with the aforementioned control system was less than 7%, and the average effective fertilization rate was greater than 85%.

This study introduces a novel approach to the strip intermittent fertilization method in orchards, offering a valuable reference for the subsequent design of an intermittent fertilization control system. Subsequently, based on the extant research, the target fertilization amount derived from the canopy scanning of fruit trees will be incorporated into the automatic control system design for the intermittent fertilization of fruit trees.