Penetrating cutting depth control method based on load pressure feedback

Abstract

The invention relates to the technical field of agricultural machinery, in particular to an in-soil cutting depth control method based on load pressure feedback, which comprises the following steps of: firstly, respectively performing test analysis on load pressures of different soil types and parameters, and taking test results as training samples; according to the method, an extreme learning machine algorithm is adopted to establish load pressure prediction models of different soil types, a sparrow optimization algorithm and a genetic algorithm are adopted to optimize the prediction models, through experimental verification, an extreme learning machine load pressure prediction model based on sparrow algorithm optimization is established, and then the load pressure prediction model is combined to determine the load pressure of different soil types. And a soil-penetrating cutting depth control system is designed, the soil-penetrating cutting depth control system controls the soil-penetrating cutting depth according to the real-time load pressure condition, the cutting quality of the sugarcane is improved, and the head breaking rate is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of agricultural machinery, and in particular to a method for controlling soil cutting depth based on load pressure feedback. Background Art

[0002] According to statistics, most parts of my country mainly use manual harvesting and incomplete small-scale mechanized harvesting methods when harvesting sugarcane. The mechanized harvesting rate of sugarcane in 2020-2021 was less than 12%, which is much lower than the mechanization level of sugarcane harvesting in other major sugarcane-growing countries. In terms of the mechanized harvesting area, the mechanized harvesting rate of sugarcane in Guangxi was only 4.6% in the 2019 / 2020 crushing season. Even in the "double-high" bases with a high degree of mechanization, the mechanized harvesting rate of sugarcane is only 15.82%. The main reason for this phenomenon is the poor cutting quality of mechanized harvesting, and cutting in the ground is an effective way to improve the cutting quality.

[0003] At present, the cutting depth of domestic sugarcane harvesters is mainly adjusted manually by visually observing the changes in the sugarcane terrain. During the harvesting process, the cab is relatively high, and the height changes of the sugarcane terrain cannot be accurately obtained, resulting in large errors; the contact terrain detection device commonly used in sugarcane harvesters is greatly affected by ground factors. my country's sugarcane is mainly grown in hilly areas, and there will be certain errors in the detection, and the modification of the prototype is relatively large; non-contact methods such as ultrasound, laser and radar are generally used in occasions with better environments, while the sugarcane planting environment is complex, and the influence of weeds, sugarcane stems and leaves, and soil in the sugarcane harvesting environment needs to be resolved. The application of this method has considerable limitations; the detection method based on load pressure feedback mainly considers the soil penetration depth and locomotive parameters, and has not established a load pressure prediction model for different soils. Summary of the invention

[0004] The purpose of the present invention is to provide a method for controlling the cutting depth into the soil based on load pressure feedback, which adopts a load pressure prediction model in conjunction with a control system to monitor and adjust the load pressure of the cutting system in real time, thereby ensuring that the cutting depth of the sugarcane harvester into the soil is within a set range, improving the cutting quality of sugarcane and reducing the breakage rate.

[0005] To achieve the above object, the present invention provides a method for controlling the soil cutting depth based on load pressure feedback, comprising the following steps:

[0006] Step 1: Establish training samples in the prediction model dataset according to L 27 (3 13 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil;

[0007] Step 2: Establish the validation sample in the prediction model dataset, according to L9(34 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil;

[0008] Step 3: Use the extreme learning machine optimized by the sparrow search algorithm to establish a load pressure prediction model and optimize it;

[0009] Step 4: Design a soil cutting depth control system in combination with pressure feedback to achieve soil cutting depth control.

[0010] Optionally, the execution process of step 3 includes the following steps:

[0011] Step 3.1: Import data and preprocess the data;

[0012] Step 3.2: Initialize the input layer and hidden layer parameters, then use the initialization parameters as the initial population of sparrows, set the number of iterations and individual position dimensions;

[0013] Step 3.3: Calculate the individual fitness value, use the position update formula to update and save the individual position;

[0014] Step 3.4: Recalculate the fitness value of the individual after the updated position, and replace the original solution with the position with a better fitness value until the fitness threshold or the maximum number of iterations is reached, and the updated position is taken as the optimal solution;

[0015] Step 3.5: Repeat steps 3.3 and 3.4. If the sparrow search algorithm finds the optimal solution or reaches the maximum number of iterations, the output result is the weight and threshold of the load pressure prediction model; if the above conditions are not met, continue to iterate and update.

[0016] Optionally, the soil cutting depth control system includes execution hardware and a mathematical model, and the execution hardware acts under the coordinated control of the mathematical model;

[0017] The control element in the execution hardware is a proportional directional valve, the execution element is a hydraulic cylinder, the execution object is a cutter, and the detection feedback device is used to reflect the cutting depth into the soil.

[0018] Optionally, the mathematical model includes a proportional directional valve mathematical model, a proportional amplifier and sensor mathematical model, a load flow equation, a flow continuity equation, a force balance equation, and a system transfer function;

[0019] The mathematical model of the proportional directional valve is used for proportional directional valve regulation, and the mathematical model of the proportional amplifier and sensor is used for proportional amplifier and sensor control. The mathematical model is coordinated through load flow equation, flow continuity and force balance equation, and finally the output is optimized by the system transfer function.

[0020] Optionally, the expression of the mathematical model of the proportional directional valve is as follows:

[0021]

[0022] Among them, K q is the flow gain of the proportional directional valve; ω v is the natural frequency of the proportional directional valve; v is the damping ratio of the proportional directional valve, ranging from 0.5 to 0.7, K q Use K q =Q0 / I, where Q0 is the no-load flow of the proportional directional valve and I is the maximum current of the proportional directional valve.

[0023] Optionally, the load flow q in the load flow equation L for:

[0024]

[0025] Among them, C d is the valve port flow coefficient; p s is the oil supply pressure, Pa; ω is the valve port area gradient, m; ρ is the oil density; p1 is the hydraulic cylinder oil inlet pressure; p2 is the hydraulic cylinder oil return pressure; x v is the displacement of the proportional directional valve core, q1 is the flow rate of the oil inlet chamber, and q2 is the flow rate of the oil return chamber.

[0026] Alternatively, the flow continuity equation is expressed as:

[0027]

[0028] Where, ΣQ in is the total inflow flow, in m 3 / s; ΣQ out is the total outflow flow, in m 3 / s; V is the volume of the hydraulic cylinder cavity, unit is m 3 ; p is the pressure inside the hydraulic cylinder chamber, the unit is Pa; β is the elastic modulus of the oil, the unit is MPa.

[0029] Optionally, the control of the cutting depth in step 4 depends on the real-time load pressure conditions, including the following conditions:

[0030] P≤P imin , the real-time pressure P is less than the reference pressure P l When the lower limit is reached, the system determines that the cutter is less than 20 mm deep into the soil, and the cutter descends to make its depth approach 20 mm.

[0031] P imin ≤P≤P imax , the implementation pressure P is at the reference pressure Pl Allowing for a range of variation, the cutter maintains this depth of penetration;

[0032] P imax ≤P≤P bmax , the real-time pressure P is greater than the reference pressure P l The upper limit of the pressure is less than the maximum pressure P allowed by the system. bmax If sugarcane is detected, the cutting depth is maintained; if no sugarcane is detected, the cutter is judged to be more than 20mm deep, and the cutter is lifted to make the depth approach 20mm;

[0033] P≥P bmax , the real-time pressure P is greater than the maximum pressure P allowed by the system bmax , the cutter needs to be lifted quickly.

[0034] The present invention provides a method for controlling the depth of soil cutting based on load pressure feedback. First, test analysis is performed on the load pressures of different soil types and parameters, and the test results are used as training samples. The load pressure prediction model of different soil types is established by using the extreme learning machine algorithm, and the prediction model is optimized by using the sparrow optimization algorithm and the genetic algorithm. Through experimental verification, the load pressure prediction model of the extreme learning machine based on the sparrow algorithm optimization is established. Then, the soil cutting depth control system is designed in combination with the load pressure prediction model. The soil cutting depth control system controls the soil cutting depth according to the real-time load pressure condition, thereby improving the cutting quality of sugarcane and reducing the head breakage rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 It is a schematic flow chart of the steps of a method for controlling the soil cutting depth based on load pressure feedback of the present invention.

[0037] Figure 2 It is a structural diagram of the extreme learning machine.

[0038] Figure 3 It is a schematic diagram of the operation process of the SSA-ELM of the present invention.

[0039] Figure 4 It is a schematic diagram of the structure of a sugarcane harvester soil cutting depth control system according to a specific embodiment of the present invention.

[0040] Figure 5 It is a schematic diagram of the process flow of the soil cutting depth control system of a specific embodiment of the present invention.

[0041] Figure 6 It is a schematic diagram of the structure of a valve-controlled hydraulic cylinder system according to a specific embodiment of the present invention.

[0042] Figure 7 It is a schematic diagram of the control system transfer function structure of a specific embodiment of the present invention.

[0043] Figure 8 The present invention is a schematic diagram of the automatic control process of the cutting depth of a sugarcane harvester in the soil according to a specific embodiment of the present invention.

[0044] Fig. 9 It is a schematic diagram comparing the measured values ​​verified by yellow soil in a specific embodiment of the present invention and the SSA-ELM predicted values.

[0045] Fig.10 It is a schematic diagram of the prediction error results of yellow soil verification in a specific embodiment of the present invention.

[0046] Fig.11 It is a schematic diagram comparing the measured values ​​verified by red soil in a specific embodiment of the present invention and the SSA-ELM predicted values.

[0047] Fig.12 It is a schematic diagram of the prediction error results of red soil verification in a specific embodiment of the present invention.

[0048] Fig.13 It is a Bode diagram of the transfer function of the mathematical model of the system of the specific embodiment of the present invention. DETAILED DESCRIPTION

[0049] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0050] The following is an explanation of the English abbreviations that appear in the text. Some areas in the text will be explained using English abbreviations. The specific meanings are as follows:

[0051] SSA: Sparrow Search Algorithm;

[0052] ELM: Extreme Learning Machine;

[0053] SSA-ELM: A load pressure prediction model established by extreme learning machine optimized with sparrow search algorithm.

[0054] See also Figure 1The present invention provides a method for controlling the soil cutting depth based on load pressure feedback, comprising the following steps:

[0055] S1: Establish training samples in the prediction model dataset, according to L 27 (3 13 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil;

[0056] S2: Establish the validation samples in the prediction model dataset, according to L9(3 4 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil;

[0057] S3: Establish and optimize the load pressure prediction model by combining the extreme learning machine optimized by the sparrow search algorithm;

[0058] S4: Combine pressure feedback to design a soil cutting depth control system to achieve soil cutting depth control.

[0059] The present invention is further described below in conjunction with specific embodiments, implementation principles and execution steps:

[0060] 1) Establish training samples in the prediction model data set, according to L 27 (3 13 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil. The levels of the experimental influencing factors were selected as shown in Tables 1 and 2.

[0061] Table 1 Yellow soil training test level table

[0062]

[0063] Table 2 Red soil training test level table

[0064]

[0065] 2) Establish the validation samples in the prediction model data set, according to L9(3 4 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil. In order to detect the accuracy of the prediction model and avoid verification errors, the selection of the levels of the experimental influencing factors should be different from that of the training sample orthogonal test, as shown in Tables 3 and 4.

[0066] Table 3 Yellow soil verification test level table

[0067]

[0068] Table 4 Red soil verification test level table

[0069]

[0070] 3) Extreme Learning Machine

[0071] Extreme learning machine is an effective learning algorithm that can quickly learn the parameters of feedforward neural networks. The parameters of traditional neural networks were adjusted iteratively, so the speed was relatively slow. Moreover, ELM was originally proposed to reduce the learning time of SLFNs. It is a learning algorithm for SLFNs and has stronger learning ability than traditional learning algorithms. The biggest innovation of ELM is that: (1) the connection weights between the input layer and the hidden layer and the threshold of the hidden layer can be set randomly and no further adjustment is required after setting; (2) the connection weights between the hidden layer and the output layer do not need to be iteratively adjusted, but are determined once by analytical equations. Studies have shown that through this change in calculation rules, the model exhibits good generalization performance and the learning speed is significantly improved. Its structure is as follows Figure 2 shown.

[0072] The principle of extreme learning machine is as follows:

[0073] For N arbitrary different samples (x i ,t i ), where x i =[x i1 ,x i2 ,…,x in ] T ∈R n , t i =[t i1 ,t i2 ,…,t in ] T ∈R m , we can know that (x i ,t i )∈R n ×R m (i=1,2,···,3), x is the input vector, t is the training set target matrix, for the extreme learning machine algorithm with L hidden layer nodes, the calculation process is as follows:

[0074]

[0075] in:

[0076] g i (x) = g(w i *x j +b i ),j=1,···,N (2)

[0077] β i =[β i1 ,β i2 ,…,β im ] T∈R m is the weight vector connecting the ith hidden node and the output node, w i *x j Indicates w i With x j The inner product of g(x), g(x) is the activation function, usually "Sigmoid function" or "RBF (Radial Basis Function)", calculate g i (x) is the output of the i-th node in the hidden layer.

[0078] T=Hβ (3)

[0079] Among them, H is the output matrix of the hidden layer,

[0080]

[0081] T is the target matrix, and L is the number of hidden layer nodes. The hidden layer weights ω and bias b are randomly input, and the hidden layer output matrix can be obtained based on the input samples. The process of training the extreme learning machine is transformed into finding the least squares solution of the equation T = Hβ, that is,

[0082]

[0083] Among them, H + is the MP (Moore-Penrose) generalized inverse matrix of the hidden layer output matrix H, and L is the number of nodes in the hidden layer. Solving the formula, we get Substitute it into the network again and the extreme learning machine algorithm is trained.

[0084] 4) Sparrow Algorithm

[0085] The Sparrow Search Algorithm (SSA) is a new type of swarm intelligence algorithm proposed by Xue Jiankai, which mainly simulates the foraging behavior of sparrows. Sparrows have three behaviors: the finder is responsible for searching for food and guiding the foraging direction of the entire population; the follower follows and monitors the finder; the sentinel is located outside the population and is responsible for warning and adjusting its own position.

[0086] In SSA, the discoverer and forager play different roles. The discoverer actively searches for the optimal solution, while the forager obtains the optimal solution. The role of the discoverer is to guide the entire population to search and forage, which can be expressed as:

[0087]

[0088] Where t is the current iteration number, is the position of the i-th discoverer in the t-th iteration of dimension d, iter maxis the maximum number of iterations; random number α∈[0,1]; Q is a random number that follows a normal distribution; R2 is randomly selected from [0,1]; ST is the warning value; L is a 1×d matrix, in which each element is 1. When R2<ST, the discoverer can conduct a wide-area search; when R2≥ST, it means that the foraging environment is dangerous and natural enemies appear, and the leader will lead the sparrow population to the next safe area to forage.

[0089] There are two behaviors of foragers: one is to follow the leader to find food, during which there are sentinels to monitor the leader's behavior and snatch the leader's food; the other is to go to other places to find food. The calculation formula for foragers is:

[0090]

[0091] In the formula, is the individual with the worst position at the tth time; To be in the best position to be the leader of the next iteration; + =E T (EE T ) -1 E is a 1×d matrix, where each element is 1 or -1; N is the number of populations. When i>N / 2, the forager does not have enough food and will go to other places to find food; when i<N / 2, it means that the forager follows the leader to continue looking for food, or snatches the leader's food.

[0092] In the sparrow algorithm model, in addition to updating the population position according to the environmental safety factor ST, there is another method for the sparrow population. Assuming that 10% to 20% of the sparrows are at the edge of the foraging area and can detect danger, they will move closer to a safer place. The starting position of these individuals is randomly given, and the update formula is:

[0093]

[0094] In the formula, is the best position of the forager in the tth iteration; λ is the learning rate of a random number with a normal distribution with a mean of 0 and a variance of 1; the random number K∈[1,1]; f i is the fitness value of the current population; f g and f w are the best fitness value and the worst fitness value respectively; ε is a small constant to avoid f i -f w =0.

[0095] 5) Extreme Learning Machine Optimized by Sparrow Algorithm

[0096] SSA has powerful local and global search capabilities and is very suitable for optimizing the weights and thresholds of ELM, thereby improving the accuracy of ELM. The operation process of SSA-ELM is as follows: Figure 3 The main steps of SSA-ELM are as follows:

[0097] (1) Import data and preprocess the data;

[0098] (2) Initialize the input layer and hidden layer parameters, then use the initialization parameters as the initial population of sparrows, set the number of iterations and the individual position dimension;

[0099] (3) Calculate the individual fitness value, use the position update formula to update and save the individual position;

[0100] (4) Recalculate the fitness value of the individual after the updated position, and replace the original solution with the position with a better fitness value until the fitness threshold or the maximum number of iterations is reached, and the updated position is taken as the optimal solution;

[0101] (5) Repeat steps 3 and 4. If the SSA optimization obtains the optimal solution or reaches the maximum number of iterations, the output result is the weight and threshold of SSA-ELM. If the above conditions are not met, continue iterative updating.

[0102] Furthermore, in order to realize the pressure feedback and control function, the soil cutting depth control system designed by the present invention is as follows:

[0103] 1) Design of control system

[0104] The structure of the control system is as follows Figure 4 As shown by Figure 4 It can be seen that the control element of the system is a proportional directional valve, the actuator is a hydraulic cylinder, the actuator object is the cutter, and the detection feedback device is used to reflect the cutting depth into the soil.

[0105] The input of the control system is the reference pressure p r , which is provided by the load pressure prediction model; the output is the cutter's penetration depth s hs ; The feedback is the load pressure p actually measured by the pressure sensor o , so the key link of closed-loop control is to determine p r As well as establishing hs With p o conversion model.

[0106] Working principle of the control system: The prediction model predicts the load pressure of different soil types and soil parameters, and uses the predicted value as the reference pressure. The PLC controller then compares the measured pressure with the reference pressure and outputs a control signal. The proportional directional valve controls the lifting and lowering of the hydraulic cylinder according to the control signal, thereby adjusting the cutting depth. According to the working principle of the control system, the corresponding scheme is designed as follows Figure 5 shown.

[0107] 2) Establish a mathematical model of the control system

[0108] The hydraulic control system can be represented by a mathematical model. Proving the stability of the control system transfer function is an important step to ensure that the control system can accurately achieve the control target. The control system uses a four-way valve-controlled asymmetric hydraulic cylinder. The structure diagram of the valve-controlled hydraulic cylinder system is as follows: Figure 6 shown.

[0109] Among them: A1, p1 and q1 are the effective area, oil pressure and flow of the rodless chamber of the hydraulic cylinder respectively; A2, p2 and q2 are the effective area, oil pressure and flow of the rod chamber of the hydraulic cylinder respectively; FL is the load force; K is the load elastic coefficient; BP is the load damping coefficient; Mt is the load mass; ps is the oil supply pressure; p0 is the return oil pressure; xv is the valve core displacement; y is the piston rod displacement.

[0110] 1. Mathematical model of proportional directional valve

[0111] The proportional directional valve is a key component of the electro-hydraulic conversion in the hydraulic control system. The internal mathematical relationship of the proportional directional valve is relatively complex, and the second-order oscillation link is often used in engineering to describe it:

[0112]

[0113] Where: K q - Flow gain of proportional directional valve; ω v - natural frequency of the proportional directional valve; v - Damping ratio of proportional directional valve (usually 0.5 to 0.7). K q Available Type K q =Q0 / I calculation, where Q0-no-load flow of proportional directional valve, I-maximum current of proportional directional valve.

[0114] 2. Proportional amplifier and sensor mathematical model

[0115] The proportional amplifier provides a specific performance current to the proportional solenoid. The transition frequency of the proportional amplifier is much higher than the natural frequency of the system, which is processed into a proportional link:

[0116]

[0117] Where: Ka - proportional amplifier gain, A / V; U-proportional amplifier input voltage, V; I-proportional amplifier input current, A.

[0118] The transition frequency of the sensor is also much higher than the natural frequency of the system, which can be simplified into a proportional link.

[0119]

[0120] Where: K f -Gain of the sensor; U f -Feedback voltage of the sensor; P m - Pressure of the sensor.

[0121] 3. Load flow equation

[0122] According to fluid mechanics, when the piston rod of the hydraulic cylinder is extended, the flow equations of the flow rate q1 of the oil inlet chamber and the flow rate q2 of the oil return chamber are as shown in formula (11).

[0123]

[0124] Where: C d -Valve port flow coefficient; p s - oil supply pressure, Pa; ω-valve port area gradient, m; ρ-oil density; p1-hydraulic cylinder oil inlet pressure; p2-hydraulic cylinder oil return pressure; x v - Proportional directional valve spool displacement.

[0125] In practical applications, the load flow q is defined as L for:

[0126]

[0127] The relationship between the speed of the piston rod and the oil inlet chamber and the oil return chamber is:

[0128]

[0129] Where: v-speed of piston rod movement.

[0130] The flow ratio between the oil inlet chamber and the oil return chamber of the hydraulic cylinder is obtained by equations (13) and (15):

[0131]

[0132] In the formula: n is the flow ratio of the two chambers of the hydraulic cylinder; A1 is the effective area of ​​the oil inlet chamber of the hydraulic cylinder; A2 is the effective area of ​​the oil return chamber of the hydraulic cylinder.

[0133] The thrust F generated by the piston is:

[0134] F=A1p1-A2p2 (17)

[0135] Load pressure p L It can be defined as shown in formula (18):

[0136]

[0137] Combining equations (16) and (18) we can get equation (19):

[0138]

[0139] Therefore, formula (20) can be rewritten as:

[0140]

[0141] Combining equations (14) and (20), we can get:

[0142]

[0143] After linearizing equation (21), we get the equation shown in equation (22):

[0144] q L =K q x v -K c p L (twenty two)

[0145] Where: K q -Proportional directional valve flow gain; K c -Proportional directional valve pressure-flow coefficient.

[0146] 4. Flow continuity equation

[0147] When studying the continuity equation of the hydraulic cylinder system, the influence of factors such as the system oil quality and pipeline dynamics is ignored, and the system flow continuity equation (38) is shown as follows:

[0148]

[0149] Where: ΣQ in -Total inflow flow, m 3 / s; ΣQ out -Total outflow flow, m 3 / s; V- hydraulic cylinder cavity volume, m 3 ; p-pressure inside the hydraulic cylinder chamber, Pa; β-elastic modulus of oil, MPa.

[0150] The liquid compression flow rate can be expressed as:

[0151]

[0152] Where: βe -Equivalent bulk modulus, MPa.

[0153] From formula (19), we can get:

[0154]

[0155] The flow rate required for piston movement is shown in formula (26):

[0156]

[0157] Where: A-piston effective area, m 2 .

[0158] The flow continuity equations of the rod chamber and rodless chamber of the hydraulic cylinder are shown in (27).

[0159]

[0160] Where: C ip -Cylinder internal leakage coefficient; C ep -Cylinder leakage coefficient; V1-oil inlet chamber volume; V2-oil return chamber volume.

[0161] Assuming that the initial volumes of the two piston chambers are equal, the flow continuity equation can be derived as shown in equation (28):

[0162]

[0163] Where: A p -Equivalent piston area; C t -Equivalent leakage coefficient; V t -Equivalent volume.

[0164] 5. Force balance equation

[0165] According to the force analysis of the piston rod, the equilibrium equation can be obtained as shown in formula (29):

[0166]

[0167] Where: m-total mass converted to the cylinder piston; B p -Viscous damping coefficient of piston and load; K s - Spring stiffness; F L -Load capacity.

[0168] By performing Laplace transform on equations (22), (28) and (29), we can obtain:

[0169] Q L =K q X v -K c P L(30)

[0170]

[0171] P L A p =ms 2 Y+B p Y+K s Y+F L (32)

[0172] Eliminate P from equations (30), (31), and (32) L and Q L , the mathematical model of the hydraulic cylinder is:

[0173]

[0174] Where K ce is the total flow-pressure coefficient, K ce =K c +C t .

[0175] Ignoring the damping load and elastic load, equation (33) can be simplified to:

[0176]

[0177] Where: p - natural frequency of hydraulic cylinder; p -Hydraulic cylinder damping ratio (generally 0.1~0.2).

[0178] In formula (34),

[0179] The transfer function of the hydraulic cylinder piston rod displacement to the valve core displacement is:

[0180]

[0181] The transfer function of the hydraulic cylinder piston rod displacement to the external load is:

[0182]

[0183] 6. Transfer function of control system

[0184] Based on the transfer functions of each link, the structural block diagram of the control system can be obtained as follows: Figure 7 shown.

[0185] The transfer function of the system can be obtained as follows:

[0186]

[0187] The basic parameters of the soil cutting depth control system of the sugarcane harvester are shown in Table 5.

[0188] Table 5 Main parameters of the system

[0189]

[0190] From the parameters of the hydraulic cylinder, we can know that the area of ​​the rod cavity is A. p The area of ​​the rodless cavity is 1001.3 mm2, the load pressure of the cylinder is about 350 kg, and the natural frequency ω of the hydraulic cylinder is obtained. p is 183 rad / s, and the damping ratio of the hydraulic cylinder is ζ p Take 0.2, the damping ratio of the proportional directional valve is ζ v Taking 0.6, substituting the above parameters into the open-loop transfer function of the system is:

[0191]

[0192] 3) Control system process

[0193] The control idea of ​​the control system is the concentrated embodiment of the entire control scheme. According to the analysis in the previous article, the control system of the cutting depth of the sugarcane harvester is as follows: Figure 8 shown.

[0194] In the figure, P is the real-time load pressure; P l is the reference pressure; P bmax The maximum pressure allowed by the system is set to 80% of the overflow pressure. The cutting system pressure needs to be the same as the reference pressure P l Compare and determine the allowable variation range of the real-time pressure P according to the single factor test in Chapter 2, that is, determine P lmin , P lmax According to the judgment result, there are the following situations:

[0195] (1)P≤P lmin , the real-time pressure P is less than the reference pressure P l When the lower limit is reached, the system determines that the cutter is less than 20 mm deep into the soil, and the cutter descends to make its depth approach 20 mm.

[0196] (2)P∈(P lmin ,P lmax ), the implementation pressure P is at the reference pressure P l A range of variation is allowed while the cutter maintains this depth.

[0197] (3)P∈(P lmax ,P bmax ), the real-time pressure P is greater than the reference pressure P lThe upper limit of the pressure is less than the maximum pressure P allowed by the system. bmax , the control system needs to determine whether there is sugarcane here. This paper uses a photoelectric sensor to detect whether there is sugarcane. If sugarcane is detected, the cutting depth into the soil is maintained. If the sensor does not detect sugarcane, the system determines that the cutter's soil depth is greater than 20mm, and the cutter is lifted to make the soil depth approach 20mm;

[0198] (4)P≥P bmax , the real-time pressure P is greater than the maximum pressure P allowed by the system bmax , then no matter there is sugarcane here or not, in order to ensure safety, the cutter needs to be quickly lifted. The maximum pressure allowed by the test cutting workbench is 100Bar. In order to ensure the safe operation of the system, take the maximum allowable pressure of the system P bmax It is 80Bar.

[0199] In order to avoid the influence of sugarcane on the collection of load pressure, it is necessary to identify the sugarcane signal. Therefore, the control system uses a photoelectric sensor to process the sugarcane signal, and the sensor is installed in front of the cutter. If the photoelectric sensor detects the signal of the sugarcane and the load pressure P∈(P lmax ,P bmax ), the system controls the cutter to remain unchanged; if the sensor does not detect the signal of sugarcane, and the load pressure P∈(P lmax ,P bmax ), the system controls the cutter to rise; if the sensor detects sugarcane, but the load pressure P≥P bmax , indicating that the cutting depth into the soil is too great. At this time, for safety reasons, the system controls the cutter to rise.

[0200] Finally, in order to prove the effectiveness of the prediction model and the stability of the control, the present invention also uses example comparison for auxiliary explanation:

[0201] (1) Effectiveness of the prediction model

[0202] In the yellow soil verification test: the prediction results of the SSA-ELM verification test are shown in Table 6, and the error analysis results are shown in Fig. 9 and Fig.10 Therefore, the maximum relative error of the SSA-ELM prediction model is 1.09%, and the average relative error is 0.63%;

[0203] Table 6 SSA-ELM prediction and verification results for yellow soil

[0204]

[0205]

[0206] In the red soil verification test: the prediction results of the SSA-ELM verification test are shown in Table 7, and the error analysis results are shown in Fig.11 and Fig.12 Therefore, the maximum relative error of the SSA-ELM prediction model is 1.16%, and the average relative error is 0.74%;

[0207] Table 7 Verification results of red soil SSA-ELM prediction

[0208]

[0209] Through the above analysis, it can be concluded that when predicting the load pressure of the sugarcane harvester cutting into the soil, the prediction model of SSA-ELM can meet the prediction requirements of load pressure.

[0210] (2) Stability of control system

[0211] After the transfer function is established, MATLAB is used to design the Bode diagram of the system mathematical model. The stability of the system is analyzed through the Bode diagram. The Bode diagram of the system is as follows: Fig.13 shown.

[0212] Depend on Fig.13 It can be seen that the amplitude margin Gm of the soil cutting depth control system is 26.1dB, and the phase margin Pm is 89°, satisfying Gm>0 and Pm>0, indicating that the system is stable.

[0213] The closed-loop transfer function obtained from the open-loop transfer function is:

[0214]

[0215] The characteristic equation of the closed-loop transfer function is:

[0216] s 5 +553.2s 4 +228625s 3 +27786720s 2 +53582440000s+1035473325=0 (40)

[0217] The coefficients of the characteristic equation are not zero, and all the coefficients are positive, which meets the conditions of system stability. The Routh table is:

[0218]

[0219]

[0220] It can be seen from the Routh table that the first column is all greater than zero, which means that the closed-loop transfer function of the system is stable.

[0221] (3) Control system test results

[0222] The designed sugarcane harvester cutting depth prediction control system was built on the test platform for testing. According to the test results, the results of the cutting depth into the soil and the recognition of sugarcane signals were statistically analyzed, and the statistical results are shown in Table 8. The response time of the control system was measured, and the response time of the system is shown in Table 9.

[0223] Table 8 Test results statistics

[0224]

[0225] Table 9 Control system response time

[0226]

[0227] The control system operates normally on the test platform and can meet the use requirements of different soils and working conditions. By analyzing the test results, it is concluded that when the target depth is 20mm, the maximum error of the system control is 5mm, the average control error is 2.5mm, and the cutting depth is basically maintained within the range of 20±5mm, which meets the requirements of actual sugarcane harvesting production. In the red soil No. 2 and No. 3 tests, the response time of the control system hydraulic cylinder is 500ms for descent and 550ms for ascent. In the yellow soil No. 2 and No. 3 tests, the response time of descent is 450ms and the response time of ascent is 500ms. The response time of the remaining tests is less than 550ms, and the response time of the control system is less than 600ms. Therefore, the response time meets the design requirements of the control system.

[0228] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0229] (1) In view of the lack of an effective cutting depth detection method for sugarcane harvesters, the present invention reflects the cutting depth through load pressure, conducts experimental analysis on load pressures of different soil types and parameters, and uses the experimental results as training samples. The extreme learning machine algorithm is used to establish a load pressure prediction model for different soil types, and the prediction model is optimized using the sparrow optimization algorithm and the genetic algorithm. Through experimental verification, an extreme learning machine load pressure prediction model based on the sparrow algorithm optimization is established, which improves the accuracy of the control system.

[0230] (2) Combined with the load pressure prediction model, an automatic control system for soil cutting depth was developed. The system can control the cutting depth and respond quickly according to different soil conditions and soil surface unevenness.

[0231] (3) The core technology of the present invention is the control system and the pressure prediction model. The existing harvester can be realized by simply modifying the control system, introducing the prediction model and adding a simple sensor. The transformation cost is low and the transformation speed is fast. The mechanical and hydraulic system structure of the existing harvester is not changed, and large-scale rapid replacement can be achieved.

[0232] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method for controlling soil cutting depth based on load pressure feedback, characterized in that: The following steps are involved: Step 1: Establish training samples in the prediction model dataset according to L 27 (3 13 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil; Step 2: Establish the validation sample in the prediction model dataset, according to L9(3 4 ) The orthogonal test table was used to conduct the load pressure orthogonal test of yellow soil and red soil; Step 3: Use the extreme learning machine optimized by the sparrow search algorithm to establish a load pressure prediction model and optimize it; Step 4: Design a soil cutting depth control system in combination with pressure feedback to achieve soil cutting depth control.

2. The method for controlling soil cutting depth based on load pressure feedback according to claim 1, characterized in that: The execution process of step 3 includes the following steps: Step 3.1: Import data and preprocess the data; Step 3.2: Initialize the input layer and hidden layer parameters, then use the initialization parameters as the initial population of sparrows, set the number of iterations and individual position dimensions; Step 3.3: Calculate the individual fitness value, use the position update formula to update and save the individual position; Step 3.4: Recalculate the fitness value of the individual after the updated position, and replace the original solution with the position with a better fitness value until the fitness threshold or the maximum number of iterations is reached, and the updated position is taken as the optimal solution; Step 3.5: Repeat steps 3.3 and 3.

4. If the sparrow search algorithm finds the optimal solution or reaches the maximum number of iterations, the output result is the weight and threshold of the load pressure prediction model; if the above conditions are not met, continue to iterate and update.

3. The method for controlling soil cutting depth based on load pressure feedback according to claim 2, characterized in that: The soil cutting depth control system includes execution hardware and a mathematical model, and the execution hardware acts under the coordinated control of the mathematical model; The control element in the execution hardware is a proportional directional valve, the execution element is a hydraulic cylinder, the execution object is a cutter, and the detection feedback device is used to reflect the cutting depth into the soil.

4. The method for controlling soil cutting depth based on load pressure feedback according to claim 3, characterized in that: The mathematical model includes a proportional directional valve mathematical model, a proportional amplifier and sensor mathematical model, a load flow equation, a flow continuity equation, a force balance equation and a system transfer function; The mathematical model of the proportional directional valve is used for proportional directional valve regulation, and the mathematical model of the proportional amplifier and sensor is used for proportional amplifier and sensor control. The mathematical model is coordinated through load flow equation, flow continuity and force balance equation, and finally the output is optimized by the system transfer function.

5. The method for controlling soil cutting depth based on load pressure feedback according to claim 4, characterized in that: The expression of the mathematical model of the proportional directional valve is as follows: Among them, K q is the flow gain of the proportional directional valve; ω v is the natural frequency of the proportional directional valve; v is the damping ratio of the proportional directional valve, ranging from 0.5 to 0.7, K q Use K q =Q0 / I, where Q0 is the no-load flow of the proportional directional valve and I is the maximum current of the proportional directional valve.

6. The method for controlling soil cutting depth based on load pressure feedback according to claim 5, characterized in that: The load flow q in the load flow equation L for: Among them, C d is the valve port flow coefficient; p s is the oil supply pressure, Pa; ω is the valve port area gradient, m; ρ is the oil density; p1 is the hydraulic cylinder oil inlet pressure; p2 is the hydraulic cylinder oil return pressure; x v is the displacement of the proportional directional valve core, q1 is the flow rate of the oil inlet chamber, and q2 is the flow rate of the oil return chamber.

7. The method for controlling soil cutting depth based on load pressure feedback according to claim 6, characterized in that: The flow continuity equation is expressed as: Among them, ∑Q in is the total inflow flow, in m 3 / s; ∑Q out is the total outflow flow, in m 3 / s; V is the volume of the hydraulic cylinder cavity, unit is m 3 ; p is the pressure inside the hydraulic cylinder chamber, the unit is Pa; β is the elastic modulus of the oil, the unit is MPa.

8. The method for controlling soil cutting depth based on load pressure feedback according to claim 7, characterized in that: The control of the cutting depth in step 4 depends on the real-time load pressure conditions, including the following: P≤P imin , the real-time pressure P is less than the reference pressure P l When the lower limit is reached, the system determines that the cutter is less than 20 mm deep into the soil, and the cutter descends to make its depth approach 20 mm. P imin ≤P≤P imax , the implementation pressure P is at the reference pressure P l Allowing for a range of variation, the cutter maintains this depth of penetration; P imax ≤P≤P bmax , the real-time pressure P is greater than the reference pressure P l The upper limit of the pressure is less than the maximum pressure P allowed by the system. bmax If sugarcane is detected, the cutting depth is maintained; if no sugarcane is detected, the cutter is judged to be more than 20mm deep, and the cutter is lifted to make the depth approach 20mm; P≥P bmax , the real-time pressure P is greater than the maximum pressure P allowed by the system bmax , the cutter needs to be lifted quickly.

Images