A control method for a tracked photovoltaic cleaning robot based on fuzzy logic and proportional derivative integral corrector

Abstract

The application provides a control method of a tracked photovoltaic cleaning robot based on fuzzy logic and proportional differential integral correctors, and relates to the field of robot control, which comprises the following steps: when the tracked photovoltaic cleaning robot is running on a photovoltaic panel, a current straight line is obtained through a camera system cooperating with an image algorithm to obtain a current yaw angle error and a lateral error, fuzzy reasoning is performed on the yaw angle error and the lateral error to obtain a coefficient, and a control signal is obtained by combining two proportional differential integral correctors, the control signal is subjected to hard constraint to obtain a correction amount, and a motor control signal is generated to control the motor of the tracked photovoltaic cleaning robot so that the tracked photovoltaic cleaning robot runs along the texture of the photovoltaic panel. The application can obtain motion information only by using the camera system, and the hardware structure is simple; the pose data obtained through the image algorithm has small fluctuation, so that the tracked photovoltaic cleaning robot can be better controlled, and the cleaning effect on the photovoltaic panel is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of robot control, in particular to a track type photovoltaic cleaning robot control method. BACKGROUND

[0002] With the increasing demand of human society for energy, and the consideration of environmental protection and sustainable development, more and more clean energy and renewable energy have entered people's life. Solar energy, as a kind of renewable clean energy, has become an important force in current energy revolution. However, due to the harsh environment of the normal deployment area of photovoltaic panels, dust and dirt are easy to accumulate on the surface of photovoltaic panels. According to scientific estimates, the power generation of photovoltaic panels without scientific cleaning can be reduced by 40% to 60%, and the power generation can be attenuated by 20% to 30%. Therefore, the cleaning robot for photovoltaic panels has incomparable significance.

[0003] Due to the characteristics of the track type photovoltaic cleaning robot, the inclination angle of the photovoltaic panel, and the conditions of rain and snow on site, the original driving direction will gradually deviate during driving. Without automatic control, it is easy to fall due to excessive deviation from the original driving direction, causing economic loss.

[0004] The error acquisition method of the existing track type photovoltaic cleaning robot mainly includes four kinds: through machine vision, through inertial navigation unit, through GPS positioning, and multi-sensor data fusion. Among them, the error accumulation problem exists in the acquisition through the inertial navigation unit, and this problem is positively correlated with the running time. In long time and high complexity cleaning tasks, it is difficult to achieve relatively accurate error acquisition. The method of GPS positioning needs to place a satellite communication device on the track type photovoltaic cleaning robot, and the requirement for signal is high. In complex environment, it is easy to lose positioning and cause system open loop. The method of multi-sensor data fusion has high precision, but it depends on the data of multiple sensors, which will significantly increase the system complexity and operation time.

[0005] The patent CN119002507A proposes a photovoltaic cleaning robot control method based on a proportional-integral-derivative controller. This method has low computational complexity and good system stability, and can achieve the control of the robot driving straight along the lines on the photovoltaic panel. However, due to the limitations of the proportional-integral-derivative controller, the flexibility of the control system is poor, and it cannot well control the yaw angle error and lateral error at the same time. The patent CN118838344A proposes a control method based on tilt parameters, which uses inertial navigation to obtain tilt parameters. However, due to the error accumulation problem of inertial navigation, it performs poorly in long-time and high-complexity cleaning tasks. The patent CN117666593A proposes a robot control optimization method based on neural networks. However, due to the high complexity and high computational power requirement of neural networks, it is not suitable for deployment on robots, which are embedded systems.

[0006] In summary, it is of great significance to design a control method with high flexibility, high system stability, low computational complexity, and good performance in long-time and high-complexity cleaning tasks. It can achieve the automatic control of the tracked photovoltaic cleaning robot and reduce the economic loss caused by yaw drop. SUMMARY

[0007] To overcome the above problems and achieve the automatic control of the tracked photovoltaic cleaning robot in complex environments, the present invention proposes a tracked photovoltaic cleaning robot control method based on fuzzy control and proportional-integral-derivative corrector. It maintains the control of lateral error while achieving the control of yaw angle. The invention also uses a discrimination method based on error change rate to further reduce the misjudgment of system error and improve the stability of the system.

[0008] To achieve the above invention goals, the technical solutions adopted are as follows:

[0009] The texture image of the photovoltaic panel surface is obtained through the image acquisition system, and the texture center line of the photovoltaic panel is set as the target yaw angle of the robot. The current heading straight line of the robot is extracted from the image using image algorithms. The current yaw angle error and lateral error of the robot are determined by comparing the heading straight line with the texture center line of the photovoltaic panel. The coefficient is obtained by fuzzy reasoning of the yaw angle error and lateral error through fuzzy logic. The control signal is obtained by combining the two proportional-integral-derivative correctors. The correction amount is obtained by limiting the control signal using hard constraints. The motor control signal is generated for controlling the motor of the tracked photovoltaic cleaning robot, so that the tracked photovoltaic cleaning robot can stably drive along the texture of the photovoltaic panel. The specific steps are as follows:

[0010] Step 1: Obtain coarse error from image: Obtain the image of the texture of the photovoltaic panel where the tracked photovoltaic cleaning robot is located by the image acquisition system, obtain the center line of the image by double threshold image binarization, remove noise and highlight path features by image erosion and image dilation, extract straight lines by Huffman change, and obtain coarse error by comparing with the center line of the image;

[0011] Step 2: Error discrimination: By analyzing the rate of change of the error, if there is a sudden change in the rate of change, it is considered that the path jumps, and the last error is used to replace the current error to realize continuous target path planning;

[0012] Step 3: Input error fuzzification: A double-input fuzzy reasoning system is used to classify the input yaw angle error and lateral error into three states through a triangular input membership function;

[0013] Step 4: Fuzzy reasoning: Through 9 fuzzy reasoning rules, the input yaw angle error and lateral error are reasoned to output 3 states;

[0014] Step 5: Defuzzification: Through the method of isosceles triangle barycentric projection, the input yaw angle error and lateral error are integrated and added, and the added result is taken as the area of the isosceles triangle with a base length of 1, the barycenter of the isosceles triangle is calculated, and the height corresponding to the barycenter is the fuzzy reasoning output;

[0015] Step 6: Correct the output: The yaw angle error and the lateral error are respectively calculated by the yaw angle proportional differential integral corrector and the lateral proportional differential integral corrector, and the correction amount of each error is obtained. After the gain of the fuzzy reasoning output, the correction amount of each error is superimposed, and the final result is the total output correction amount of the tracked photovoltaic controller;

[0016] Step 7: Hard constraint to prevent loss of control: Limit the maximum angular velocity and linear velocity of the tracked photovoltaic cleaning robot to prevent the controller from losing control and causing the tracked photovoltaic cleaning robot to lose control;

[0017] Step 8: Motor control: The angular velocity and linear velocity obtained in the above steps are calculated through the kinematics change matrix to obtain the motor control signal, thereby realizing the automatic control of the tracked photovoltaic cleaning robot driving along the photovoltaic panel texture.

[0018] Further, the three states are S (small), M (medium), and B (large), wherein S (small) is normalized error between 0 and 0.4; M (medium) is normalized error between 0.3 and 0.7; B (large) is normalized error between 0.6 and 1.

[0019] Further, the three states are R (yaw angle emphasis type), N (ordinary), and H (lateral emphasis type).

[0020] Further, the 9 fuzzy inference rules are:

[0021] Rule 1: if the yaw angle error is small and the lateral error is small, the output state is normal;

[0022] Rule 2: if the yaw angle error is small and the lateral error is medium, the output state is lateral emphasis type;

[0023] Rule 3: if the yaw angle error is small and the lateral error is large, the output state is lateral emphasis type;

[0024] Rule 4: if the yaw angle error is medium and the lateral error is small, the output state is yaw emphasis type;

[0025] Rule 5: if the yaw angle error is medium and the lateral error is medium, the output state is normal;

[0026] Rule 6: if the yaw angle error is medium and the lateral error is large, the output state is lateral emphasis type;

[0027] Rule 7: if the yaw angle error is large and the lateral error is small, the output state is yaw emphasis type;

[0028] Rule 8: if the yaw angle error is large and the lateral error is medium, the output state is yaw emphasis type;

[0029] Rule 9: if the yaw angle error is large and the lateral error is large, the output state is normal.

[0030] Further, the image acquisition system comprises an optical camera, a surrounding light blocking plate and a light supplement lamp.

[0031] Further, the control method is deployed on an Ubuntu 18.04LTS system development board and is written in Python language.

[0032] The present application proposes a tracked photovoltaic cleaning robot control method based on fuzzy control and proportional integral derivative corrector, which has the following advantages:

[0033] (1) Fuzzy control is adopted to realize the control of yaw angle and lateral error at the same time with only a small increase in calculation complexity;

[0034] (2) The fixed parameter proportional integral derivative corrector control algorithm is superimposed to reduce the complexity of fuzzy inference and greatly ensure the stability of the system;

[0035] (3) The detection device only needs one optical camera, which reduces the complexity of the system and avoids the problem of error accumulation of inertial navigation unit;

[0036] (4) The error rate-based discrimination method is adopted to avoid the error data output problem of the camera at the connection of the photovoltaic panel, and long-distance stable driving of the tracked photovoltaic cleaning robot is realized. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 The control system block diagram of the method is shown.

[0038] Figure 2 The hardware installation schematic diagram is shown.

[0039] Fig. 3 is a membership function diagram of the fuzzy controller (a) input membership; (b) output membership.

[0040] Figure 4 The fuzzy inference rule table is shown.

[0041] Figure 5 The control effect simulation diagram is shown. DETAILED DESCRIPTION

[0042] The tracked photovoltaic cleaning robot control method embodiment based on fuzzy control and proportional integral derivative corrector of the present application will be specifically described below in combination with the drawings of the specification, but the following examples are only descriptive and not restrictive.

[0043] The present application adopts the method of machine vision to obtain error. The control block diagram is shown in the accompanying Figure 1 When the tracked photovoltaic cleaning robot drives on the photovoltaic panel, the image is obtained through the image acquisition system, the current heading straight line of the tracked photovoltaic cleaning robot and the photovoltaic panel texture are obtained through image algorithm, and the current yaw angle error and the current lateral error are obtained by comparing the photovoltaic panel texture and the heading straight line; the fuzzy inference is carried out on the yaw angle error and the lateral error through fuzzy logic, the motor of the tracked photovoltaic cleaning robot is controlled through the coefficients obtained by combining the two proportional integral derivative correctors and the fuzzy inference, so as to control the tracked photovoltaic cleaning robot to drive along the photovoltaic panel texture. The image acquisition system is shown in the accompanying Figure 2 The direction indicated by the arrow is the motion direction of the tracked photovoltaic cleaning robot, and the image acquisition system includes an optical camera, a surrounding light shield and a light supplement lamp, which are installed at the front part of the tracked photovoltaic cleaning robot. When the system is started, the image acquisition system starts to work, collects the image of the texture where the tracked photovoltaic cleaning robot is located in real time, and transmits it to the controller through the USB interface; the control method is deployed on the Ubuntu 18.04 LTS system development board and written in Python language.

[0044] In each control cycle, the following steps are followed:

[0045] (1) Image acquisition error: In order to reduce the real-time operation amount of the system, the original color image is grayed, the meaningless RGB information is discarded, and then the median filter is used to remove the small noise caused by the imaging quality and picture shaking. In order to highlight the characteristics of the coarse white line in the picture, the dark background and the brighter reflection are all changed to black by using the double threshold method, and only the white line is left in the picture. Since the white line of the photovoltaic panel has thick and thin, the thin line is more and the thick line is less, the image expansion and image corrosion operation are adopted in the scheme, the image expansion is used to complete the rough edge of the white line, which is more conducive to the subsequent straight line identification, and the image corrosion is used to filter out all the thin lines and only leave the thick lines for tracking. The Canny operator is used for edge detection, and the Hough transform is used to identify all the straight lines in the image. If the maximum allowed interval (break) of two line segments in the same direction is determined to be a line segment and exceeds the set value, the two line segments are regarded as a line segment. The larger the value is, the larger the break on the line segment is allowed, and the more likely the potential straight line segment is detected. In the application scenario of the application, there is expected to be only one straight line, so the constraint of the maximum allowed interval is very loose. The straight line closest to the center of the picture is found as the target path, the center line of the picture is used as the heading straight line, and the deviation of the heading straight line and the target path is used as the output yaw error and lateral error.

[0046] (2) Error discrimination: Since the interval of the lines on different photovoltaic panels is different, and in the running process (especially in the initial state), multiple straight lines may appear in the camera field of view. If the output error of the camera is directly used as the final error, when the tracked photovoltaic cleaning robot shakes for some reason, the determined straight line may jump, causing the system to fall into a positive feedback loop. Therefore, before the lateral error input controller, the lateral error is discriminated, and the change of the lateral error change rate is used to judge whether the target trajectory jumps. If the target trajectory jumps, the lateral error will change suddenly, and the lateral error change rate will suddenly rise. If this rise is detected, it can be considered that the jump occurs. At this time, the error discriminator gives up the measurement of the lateral error this time, and uses the last lateral error for output. Through the above method, the sensor error caused by environmental interference can be largely avoided.

[0047] (3) Input error fuzzification: Through the above steps, the controller considers that the true values of the yaw angle error and lateral error have been obtained, and the heading angle error and lateral error are fuzzified. The fuzzification method adopted by the present application is to normalize the yaw angle error and lateral error for discrimination, and classify them into three states, S (small), M (medium), and B (large) through a triangular input membership function. These states correspond to different error ranges, S (small), the normalized error falls between 0 and 0.4; M (medium), the normalized error falls between 0.3 and 0.7; B (large), the normalized error falls between 0.6 and 1; for state S (small), the left end point is 0 and the vertex is 0.4; for state M (medium), the left end point is 0.3 and the vertex is 0.7; for state B (large), the left end point is 0.6 and the vertex is 1; Fig. 3(a) shows the image of the input membership function, in which each state corresponds to a triangular curve. These curves describe how the error value is mapped to the membership value.

[0048] (4) Fuzzy inference: According to the fuzzy inference rules shown in Fig. 3(b), the state of the fuzzy inference output is obtained, which is divided into three states, N (normal), H (lateral emphasis), and R (yaw emphasis). There are 9 fuzzy inference rules, which are as follows: Figure 4

[0049] Rule 1: If the yaw angle error is small and the lateral error is small, the output state is normal;

[0050] Rule 2: If the yaw angle error is small and the lateral error is medium, the output state is lateral emphasis;

[0051] Rule 3: If the yaw angle error is small and the lateral error is large, the output state is lateral emphasis;

[0052] Rule 4: If the yaw angle error is medium and the lateral error is small, the output state is yaw emphasis;

[0053] Rule 5: If the yaw angle error is medium and the lateral error is medium, the output state is normal;

[0054] Rule 6: If the yaw angle error is medium and the lateral error is large, the output state is lateral emphasis;

[0055] Rule 7: If the yaw angle error is large and the lateral error is small, the output state is yaw emphasis;

[0056] Rule 8: If the yaw angle error is large and the lateral error is medium, the output state is yaw emphasis;

[0057] Rule 9: If the yaw angle error is large and the lateral error is large, the output state is normal.

[0058] ​(5) Deblurring: According to the above steps, the output state can be obtained, and according to the output membership function shown in Figure 3(b), the output state is a triangular curve, and the projection of the obtained output state will correspond to fall within a triangle, and the present application proposes a deblurring method based on the bisectional triangle barycentric projection method, specifically: the yaw angle error and the lateral error are added to the input membership of the three states, and the projection area of the output membership corresponding state is taken as the barycentic position of an isosceles triangle with a base length of 1, and the output membership function value corresponding to the barycentic position is the output q of this fuzzy reasoning.

[0059] (6) Correcting the output: The controller commonly used in industry is a proportional derivative integral corrector, which can be divided into a position proportional derivative integral corrector and an incremental proportional derivative integral corrector; since the output of the control system of the present application is a speed, and in order to reduce the memory occupation, the present application uses an incremental proportional derivative integral corrector, and its formula is:

[0060] u(t) = Kp*(err(t)-err(t-1))+Ki*err(t)+Kd*(err(t)-2err(t-1)+err(t-2))

[0061] Where Kp is the proportional coefficient, Ki is the integral coefficient, Kd is the differential coefficient, err(t) is the current error, err(t-1) is the previous error, err(t-2) is the second previous error, and u(t) is the output correction value; the yaw angle error correction value and the lateral error correction value are multiplied by the output q of the above fuzzy reasoning and 1-q respectively, and then superimposed, and the result obtained is the final output correction value, which is a change value, not a position value, so the speed output value of this time is obtained by adding the correction value of this time to the speed of the last time.

[0062] (7) Hard constraint anti-loss of control: Due to the mechanical structure limitation of the tracked photovoltaic cleaning robot, the output speed needs to be limited within a certain range, otherwise it will cause mechanical structure damage or output current too large to burn electronic components or controller out of control to cause the tracked photovoltaic cleaning robot out of control.

[0063] (8) Motor control: The angular velocity and linear velocity obtained in the above steps are calculated through kinematics change matrix to obtain motor control signals, and the control signals are output to the motor driver, so as to realize the automatic control of the tracked photovoltaic cleaning robot driving along the texture of the photovoltaic panel.

[0064] After the above steps, that is, completing a control calculation, the calculated speed is packaged with other commands and sent to the motion control board through the USB interface, the motion control board analyzes the commands after receiving them, and the analyzed speed is calculated through a kinematic change matrix to obtain a motor control signal, thereby realizing the automatic control of the tracked photovoltaic cleaning robot. Figure 5 As shown in the simulation results of the tracked photovoltaic cleaning robot in Simulink, the lateral error reaches the control target at 4 seconds and has no obvious overshoot, and there is no obvious oscillation after entering the steady state, the yaw angle error reaches the control target at 6 seconds and has a very small overshoot, and there is no obvious oscillation after entering the steady state, so it can be known that in the control process, the tracked photovoltaic robot effectively controls the yaw angle error and the lateral error at the same time, which shows that the method proposed in the application has a fast response and good stability control effect; the actual driving distance of the tracked photovoltaic cleaning robot is 200 meters and there is no route deviation, and the results show that the method proposed in the application can realize the control of the tracked photovoltaic cleaning robot and has a good effect.

Claims

1. A control method for a tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector, characterized by, The application relates to a method for controlling a track-type photovoltaic cleaning robot to stably travel along a photovoltaic panel texture. The method comprises the following steps: Step 1: obtaining a coarse error from an image: an image of a photovoltaic panel texture where a track-type photovoltaic cleaning robot is located is acquired through an image acquisition system, a center line of the image is obtained through double-threshold image binarization, noise is removed and path features are highlighted through image erosion and image dilation, a straight line is extracted through Hough transformation, and a coarse error is obtained through comparison with the image center line; Step 2: error discrimination: if a change rate of the error is suddenly changed, it is considered that a path jump occurs, and the last error is used to replace the current error to realize continuous target path planning; Step 3: input error fuzzification: a double-input fuzzy reasoning system is adopted to classify input yaw angle error and lateral error into three states through a triangular input membership function; Step 4: fuzzy reasoning: the input yaw angle error and the lateral error are reasoned into three output states through nine fuzzy reasoning rules; Step 5: defuzzification: the input yaw angle error and the lateral error are integrated and added through an isosceles triangle barycentric projection method, the addition result is taken as the area of an isosceles triangle with a bottom length of 1, the barycenter of the isosceles triangle is calculated, and the height corresponding to the barycenter is the fuzzy reasoning output; Step 6: correction output: the yaw angle error and the lateral error are respectively calculated through a yaw angle proportional differential integral corrector and a lateral proportional differential integral corrector, the correction amount of each error is superimposed after the gain of the fuzzy reasoning output, and the final result is the total output correction amount of the track-type photovoltaic controller; Step 7: hard constraint anti-loss of control: the maximum angular velocity and linear velocity of the track-type photovoltaic cleaning robot are limited to prevent the controller from losing control and causing the track-type photovoltaic cleaning robot to lose control; Step 8: motor control: the angular velocity and linear velocity obtained in the above steps are calculated through a kinematics change matrix to obtain motor control signals, so that the track-type photovoltaic cleaning robot can travel along the photovoltaic panel texture automatically.

2. The control method of the tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector according to claim 1, characterized in that, The three states are S (small), M (medium) and B (large), wherein the normalized error falls between 0 and 0.4 for S (small), the normalized error falls between 0.3 and 0.7 for M (medium), and the normalized error falls between 0.6 and 1 for B (large). The three states are R (yaw angle emphasis type), N (ordinary) and H (lateral emphasis type).

3. The control method of the tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector according to claim 1, characterized in that, The nine fuzzy reasoning rules are:

4. The control method of the tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector according to claim 1, characterized in that, ​ Rule 1: If the yaw angle error is small and the lateral error is small, the output state is normal; Rule 2: If the yaw angle error is small and the lateral error is medium, the output state is lateral emphasis type; Rule 3: If the yaw angle error is small and the lateral error is large, the output state is lateral emphasis type; Rule 4: If the yaw angle error is medium and the lateral error is small, the output state is yaw emphasis type; Rule 5: If the yaw angle error is medium and the lateral error is medium, the output state is normal; Rule 6: If the yaw angle error is medium and the lateral error is large, the output state is lateral emphasis type; Rule 7: If the yaw angle error is large and the lateral error is small, the output state is yaw emphasis type; Rule 8: If the yaw angle error is large and the lateral error is medium, the output state is yaw emphasis type; Rule 9: If the yaw angle error is large and the lateral error is large, the output state is normal.

5. The control method of the tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector according to claim 1, characterized in that, The image acquisition system comprises an optical camera, a surrounding light blocking plate and a light supplementing lamp.

6. The control method of the tracked photovoltaic cleaning robot based on fuzzy logic and proportional-derivative-integral corrector according to claim 1, characterized in that, The control method is deployed on an Ubuntu 18.04LTS system development board and is written in Python language.

Images