An adaptive deviation correction method for photovoltaic cleaning robots based on a three-level feedback regulation mechanism

Abstract

The present invention belongs to the field of photovoltaic cleaning robots. To address the problem of correcting the operation of rail-mounted photovoltaic cleaning robots on a confined track, an adaptive correction method for photovoltaic cleaning robots based on a three-level feedback adjustment mechanism is proposed. Based on the fusion of multiple data including the cleaning robot's motor pulses, operating posture angles, and motor current characteristic values, a hierarchical feedback adjustment system is established, including a speed loop, an angle loop, and a current loop. These three independent closed-loop control systems form a hierarchical feedback mechanism, enabling real-time monitoring and step-by-step reporting of the robot's status. This method achieves mutual correction of data sources during the robot's operation, accurately detecting and correcting tilt in the robot's operation and ensuring the cleaning efficiency of the photovoltaic cleaning robot.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic cleaning robots, and in particular to a photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism of a cleaning robot motor pulse, an operating posture angle, and a motor current characteristic value. Background Art

[0002] With the rapid advancement of solar technology, photovoltaic power plants have become one of the preferred renewable energy solutions in many regions around the world. However, the performance of photovoltaic systems is affected by a variety of external factors, most notably dust, bird droppings, and other contaminants covering the photovoltaic panels, which significantly reduce the photovoltaic cells' photoelectric conversion efficiency. Therefore, ensuring the cleanliness of photovoltaic panel surfaces is crucial for maintaining the efficient operation of the entire photovoltaic power generation system. To effectively address this challenge, a series of cleaning robots designed specifically for photovoltaic panels have been introduced. Through automated cleaning methods, these robots not only improve cleaning efficiency but also reduce the cost and risk of manual maintenance, becoming an indispensable technical means to enhance the performance of photovoltaic power plants.

[0003] Figure 1 As shown in the figure, a rail-mounted photovoltaic cleaning robot is a common type of photovoltaic panel cleaning equipment. Rail-mounted photovoltaic cleaning robots are currently a common device used for maintaining and cleaning photovoltaic panels. This type of robot is designed to move along the long edges of rectangular photovoltaic panels. Its structure consists of a bar-shaped main body and two wheels mounted at each end. These wheels are driven by a master motor and a slave motor, respectively, allowing the entire device to move smoothly across the photovoltaic panel. Furthermore, the bar-shaped main body is equipped with stop mechanisms at both ends. These mechanisms adhere closely to the sides of the photovoltaic panel, ensuring that the robot can move freely on the surface without slipping. Limit sensors are installed on the front and rear of the stop mechanisms to ensure that the robot accurately stops at its designated stop position. To ensure smooth operation, the distance between the stop mechanisms is slightly larger than the actual width of the photovoltaic panel. This ingenious design gives rail-mounted photovoltaic cleaning robots a wide coverage area and efficient cleaning capabilities, making them widely used. However, in actual operation, rail-mounted photovoltaic cleaning robots sometimes deviate from their track due to factors such as uneven edges of the photovoltaic panel, accumulation of external dirt, wind, and mechanical failure. Such deviation may not only lead to poor cleaning results, but in severe cases may even cause damage to the photovoltaic panels.

[0004] To address the problem of deviation correction for rail-mounted photovoltaic cleaning robots on a confined track, the industry is currently exploring a variety of methods and devices, as shown in the table below. These include adding external sensors, steering structures, and other devices, and using angle sensors, ultrasonic sensors, and other sensors to collect data and perceive changes in the robot's posture.

[0005]

[0006] Patent [1] mainly uses data acquisition modules and steering mechanisms to monitor and correct the path deviation of the photovoltaic robot in real time. However, this research method does not take into account the displacement path deviation caused by the robot slipping, accidental jamming, etc. This deviation cannot truly reflect the robot's posture deviation.

[0007] Patent [2] mainly adopts follower arms, side wheels, swing detection elements and elastic reset parts to prevent the robot from getting stuck. However, this research method does not completely solve the problem of deflection. It only avoids the risk of the robot getting stuck during operation, and does not consider the cleaning efficiency and operating power of the robot in the case of deflection.

[0008] Patent [3] mainly calculates the robot's azimuth during operation by obtaining the acceleration value and angular velocity value during operation, and monitors and executes correction instructions in real time by comparing with the preset azimuth angle. However, this research method does not consider that the value of the Z-axis angular acceleration of the robot during operation will have cumulative errors. As the operation time increases, the accuracy of its azimuth angle decreases, and it cannot accurately reflect the robot's posture during operation. At the same time, it also does not propose how to calibrate the data.

[0009] Patent [4] mainly uses mechanical means to detect the tilt angle between the upper and lower walking mechanisms, and amplifies the tilt angle of the photovoltaic cleaning robot so that the encoder can quickly detect the tilt angle and correct it. However, this research method does not take into account the wear error of the machine and the interference factors of the external environment. As the use time increases, the tilt angle cannot accurately reflect whether the two ends of the robot are actually deflected.

[0010] Patent [5] uses an edge detection sensor installed on the cleaning roller brush bracket at the front end of the walking cleaning robot to identify the edge of the component frame on the power station bracket and determine whether the robot has deviated from the cleaning route due to the imbalance of the driving wheel, the tilt of the power station bracket, etc. The research method proposes to install an edge detection sensor, which requires manual cleaning of the exposed edge detection sensor on a regular basis, and does not consider the use of edge detection sensors in harsh environments.

[0011] Patent [6] mainly uses ultrasonic detectors to monitor the distance between the two guide wheels on the same side and the side of the photovoltaic panel, so as to determine whether the photovoltaic cleaning robot is tilted and the tilt direction. In actual environmental testing, ultrasonic detectors have special requirements for installation height and position. On this type of rail-mounted photovoltaic cleaning robot, the distance between the guide wheel position and the photovoltaic panel surface is close to the effective detection distance of the ultrasonic detector. This research method does not take into account the blind spot problem caused by the ultrasonic detector being too close to the photovoltaic panel surface, as well as the gap error at the connection between the photovoltaic panels.

[0012] Patent [7] mainly analyzes the current change trend of the robot's main motor and slave motor in real time, compares the current difference between the main motor and the slave motor with the preset reference value, intelligently judges the motor's operating status, and detects the robot's posture. This research method uses current data to avoid the additional cost of external devices and the cumulative error caused by the external environment. However, during the research process, it was found that the method described in the patent mentioned the need for additional preset reference values, which greatly reduced the universality of the method. The motor current values ​​of the same machine in different environments and years of use have certain differences. This requires the operator to conduct multiple tests on the robot in a fixed environment and quantity each time to obtain an accurate reference value. In addition, the motor current is affected by environmental factors such as photovoltaic brackets and weather, and there is a certain range of fluctuation. Therefore, the preset value is a real-time error correction, which lacks flexibility.

[0013] Combining the above-mentioned major research directions in the industry, in order to realize robot deflection detection and correction, the research method is mainly to use external devices or sensors to collect additional data to perceive the robot's operating posture.

[0014] In the research methods of the aforementioned patents [3-4], the angle sensor data used, after field testing of the angle sensors, showed that the yaw angle in the Z-axis direction, whether using a six-axis or nine-axis sensor, is easily affected by operational vibration or magnetic fields, resulting in cumulative errors during long-term operation. Therefore, it is necessary to have a clear, fixed direction or data reflecting the robot's accurate posture every time the robot passes through a certain period of time to correct the angle data errors. However, the research methods of patents [3-4] use a single data source, with only angle data reflecting the robot's operating posture. They do not clearly address or correct the cumulative errors of the angle sensor. Therefore, when data errors occur, the data source cannot be calibrated, resulting in the data not accurately reflecting whether the robot's two ends are actually deflected. The external devices and other sensors used in the research methods of patents [1-2] and [5-6] all suffer from the same problem as patents [3-4]: a single data source cannot be used for timely data calibration when data has accumulated errors or wear errors.

[0015] Patent [7] innovatively proposed that there is no need to install additional external devices and sensors, and only the internal motor current data of the robot is used as the basis for detecting the robot's running deviation. Its core method is: collecting and analyzing the motor current change trend of the robot under different running postures, comparing it with the motor current value of the robot under normal and non-deviation running, extracting the characteristic value of the current data, and setting the corresponding threshold to achieve effective detection of the robot's deflection state. Therefore, according to the patent [7], two different robots of the same model were tested under the same conditions and the same machine was tested in different sites and at different speeds. The collected motor current change curves were compared and analyzed, and it was found that there were obvious differences in the motor current running trends in the tests under the above conditions. Obviously, as described in patent [7], the preset motor current reference value as the threshold for sensing the robot's running posture change cannot be applied to all running conditions; in the field test, it was also found that when the robot runs through the metal bridge between the photovoltaic panels, the motor current will produce data fluctuations, which will lead to the misjudgment of the robot's posture deviation. Similarly, the research method of patent [7] also uses only a single data source. Therefore, for the misjudgment problem caused by the fluctuation of motor current data due to the above abnormal situation, it is also impossible to use other data that can reflect the accurate operating posture of the robot for auxiliary judgment, and the preset baseline value method it adopts cannot cover all operating conditions. Summary of the Invention

[0016] Given the aforementioned shortcomings of existing technologies, this paper proposes an innovative solution: an adaptive deviation-correction method for photovoltaic cleaning robots based on a three-level feedback regulation mechanism. This method aims to enhance the robot's operational accuracy and stability in changing environments, offering real-time response, high fault tolerance, and self-adaptation without requiring preset thresholds.

[0017] This method combines pulse data, current characteristic values, and heading angle data from the master and slave motors to establish a hierarchical feedback control system, consisting of a speed loop, an angle loop, and a current loop. These three independent closed-loop control systems form a hierarchical speed-angle-current feedback control mechanism, enabling real-time monitoring and tiered reporting of the robot's status. The key point is that, considering the aforementioned factors, such as the cumulative error and inability to accurately correct angle data from the angle sensor, the poor universality of current data due to the preset baseline value, and misjudgment under abnormal conditions, this mechanism utilizes the stability and accuracy of the initial angle data to complete the collection and recording of the motor current characteristic baseline value. During operation, environmental interference with the current value can be detected using the angle data, thus eliminating misjudgment. Similarly, the accumulated error in the angle data after increased operation time can be corroborated by the trend change in the current characteristic value at fixed intervals during operation. If an error is found, the yaw angle baseline value is immediately updated. This achieves mutual correction of the data sources during robot operation, accurately detecting and correcting tilt in the robot's operation, and ensuring the cleaning efficiency of the photovoltaic cleaning robot.

[0018] By utilizing a three-level feedback control mechanism based on motor pulses, motor current, and operating attitude angle, a low-cost, high-efficiency, and fault-tolerant adaptive deflection correction method for a photovoltaic cleaning robot based on a three-level feedback control mechanism is implemented. This mechanism collects and analyzes motor pulses, motor current, and operating attitude angle data, and utilizes a hierarchical feedback control mechanism to monitor and determine the robot's operating status in real time, promptly detecting and correcting tilt issues. This ensures the cleaning quality of photovoltaic panels and the safe operation of the robot. This method not only ensures accurate deflection correction under ideal operating conditions but also utilizes the hierarchical feedback control mechanism to verify data accuracy and perform timely error correction in exceptional circumstances. This method provides high data reliability, significantly reduces additional human intervention, and improves the operational efficiency of photovoltaic power plants.

[0019] The present invention relates to an adaptive deviation correction method for a photovoltaic cleaning robot based on a three-level feedback regulation mechanism. By performing independent closed-loop analysis on the pulse data of the main motor and the slave motor, the motor current characteristic value data and the robot heading angle data and establishing a hierarchical feedback regulation mechanism, the method realizes accurate detection and correction of the robot's posture during operation. First, in the operation preparation stage, the heading angle of the robot when it is at the parking position is collected as the heading angle reference value to ensure the initial state of the attitude angle. Then, the running distance between the main motor and the slave motor is calculated using the number of pulses generated by the motor operation, and the closed-loop control of the incremental PID technology is adopted to ensure that the running speed of the slave motor can be detected and compensated in real time. At the same time, when the robot is running stably and the collection conditions are met, the motor current data is collected and the collected current data is preprocessed to improve the accuracy and stability of the data. On this basis, the motor current characteristic reference value is determined. Finally, an independent closed-loop control system—the speed loop, angle loop, and current loop—is established through three data sets. This system then establishes a hierarchical feedback control mechanism for the current, angle, and speed loops. This mechanism monitors the robot's operating status in real time and provides feedback to higher-level systems. This feedback is then received by the highest-level feedback control mechanism, which then performs data analysis and compensation, achieving real-time monitoring of the robot's operating status and effective error correction. This method is characterized by real-time performance, high fault tolerance, and adaptability. It eliminates the need for manually set thresholds and enables real-time error correction and updates through the mutual verification of multiple data points. It is particularly suitable for complex environments such as photovoltaic power plants, where track accuracy is critical, significantly improving the robot's cleaning efficiency and operational stability.

[0020] The present invention provides a photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism.

[0021] The adaptive deviation correction method includes a three-level feedback adjustment mechanism, the first level is based on the current loop method, the second level is based on the angle loop method, and the third level is based on the speed loop method;

[0022] The speed loop method includes: real-time acquisition of the number of pulses generated by the operation of the master motor and the slave motor, judging whether deflection occurs based on the pulse number, and if deflection occurs, triggering the speed loop to start the following adjustment mechanism: using the master motor as a reference, using the incremental PID control method to compensate and adjust the operating speed of the slave motor to achieve deviation correction;

[0023] The angle loop method includes: collecting the current heading angle in real time, judging whether deflection occurs based on the current heading angle and the heading angle reference value, and if deflection occurs, triggering the angle loop to start the following adjustment mechanism:

[0024] The current loop method includes: collecting the current characteristic value in real time, judging whether deflection occurs based on the current characteristic value and the current characteristic value reference value, and if deflection occurs, triggering the current loop to start the following regulation mechanism: executing the second-level feedback regulation mechanism; this method includes the following steps:

[0025] S1: If the heading angle of the robot at the parking position is within the preset range, the heading angle of the robot at the parking position is collected as the heading angle reference value, and the main motor and the slave motor are started to enter the first stage of operation;

[0026] S2, based on the actual working scenario of the robot, defines the area without photovoltaic panels within X meters of the parking position as the parking position area. When the robot operates in the parking position area, the second-level feedback adjustment mechanism is executed to correct the deviation;

[0027] S3: When the robot travels a distance greater than X meters, it leaves the parking area and enters the second stage of operation. The second stage of operation is in the photovoltaic panel area. During the first T time of the second stage of operation, the robot's operating speed has not yet accelerated to the set value, and the motor current data has not reached the peak-to-peak value. Therefore, its motor current characteristic data cannot accurately represent the robot's operating posture. Therefore, the second-level feedback adjustment mechanism is executed to correct the deviation.

[0028] S4: When the robot runs for longer than T in the second stage, it continuously collects the master motor current and slave motor current data while executing the second-level feedback adjustment mechanism to correct the deviation, and converts them into motor current characteristic values;

[0029] S5, determining whether the angle loop is continuously triggered to start the adjustment mechanism within a preset time; if not, taking the current motor current characteristic value as the motor current characteristic reference value;

[0030] S6, if yes, record the current motor current characteristic value, wait for a preset time and then repeat step S5; if still yes, record the current motor current characteristic value, wait for a preset time and then repeat step S5, repeat up to M times, and take the average of the M recorded current characteristic values ​​as the motor current characteristic reference value;

[0031] S7, after obtaining the motor current characteristic reference value, the robot enters the third stage of operation. In this stage, the robot still operates in the photovoltaic panel area and simultaneously executes the first-level feedback adjustment mechanism and the second-level feedback adjustment mechanism to jointly correct the deviation.

[0032] Preferably, the step S7 further includes the following steps:

[0033] S8: When the robot is in the third stage of operation, if neither the current loop start-up adjustment mechanism nor the angle loop start-up adjustment mechanism is triggered, that is, the robot has not deviated, in this case, the robot updates the current motor current characteristic value and heading angle to the motor current characteristic reference value and heading angle reference value every time it runs a preset distance;

[0034] S9, if the current loop start adjustment mechanism is not triggered, but the angle loop start adjustment mechanism is triggered: the misjudgment is caused by the motor current data, and the motor current characteristic reference value is updated to the current motor current characteristic value;

[0035] S10, if the current loop is triggered to start the adjustment mechanism, but the angle loop is not triggered to start the adjustment mechanism: the misjudgment is caused by the angle data, and the heading angle reference value is updated to the current heading angle;

[0036] At the same time, in S11, during the third stage of operation, it is detected whether there is a limit sensor trigger. If not, steps S8-10 are repeated. If triggered, the robot reaches the end stop position, the robot changes its running direction, and steps S1-11 are repeated.

[0037] Preferably, in S1, the preset range is 90°±3°.

[0038] Preferably, in S4, the conversion into motor current characteristic values ​​specifically includes preprocessing the current data including reducing noise, smoothing data, and processing abnormal values; and converting the preprocessed current data into motor current characteristic values ​​through a root mean square calculation formula.

[0039] Preferably, in S5 and S6, the preset time is 3 seconds.

[0040] Preferably, the output is a signal for triggering the speed loop start adjustment mechanism; specifically, it includes: outputting a fixed number of pulses to increase or decrease the number of pulses of the main motor, thereby triggering the speed loop start adjustment mechanism, performing slave motor speed compensation, and thus correcting the robot's running posture.

[0041] Preferably, in S8, the preset distance is recorded as Y, and the setting value interval of Y is [5, 10]; the setting value interval of M is [3, 5]; the setting value interval of X is [1, 1.2]; and the setting value interval of T is [10, 15] seconds.

[0042] Preferably, the rail-mounted photovoltaic cleaning robot is used for cleaning rectangular photovoltaic panels.

[0043] Preferably, the heading angle is the attitude angle in the Z-axis direction calculated by a six-axis sensor carried on the robot main control board.

[0044] Preferably, the determining whether deflection occurs based on the number of pulses specifically includes:

[0045] Get the circumference of the driving wheel and the driven wheel respectively;

[0046] The distance traveled by one end of the fuselage where the driving wheel is located is calculated by multiplying the circumference of the driving wheel by the number of pulses of one revolution of the main motor;

[0047] The distance traveled by the end of the fuselage where the driven wheel is located is calculated by multiplying the circumference of the driven wheel by the number of pulses in one revolution of the slave motor;

[0048] Whether or not a skew occurs is determined based on the difference in the distances.

[0049] The rail-mounted photovoltaic cleaning robot is used for cleaning rectangular photovoltaic panels;

[0050] The hanging rails are the edges of the long sides of the rectangular photovoltaic panel. The hanging rail type photovoltaic cleaning robot includes a bar-shaped fuselage, and a driving wheel and a driven wheel are respectively provided at both ends of the fuselage; the driving wheel and the driven wheel are driven by the main motor and the slave motor respectively; limiting structures are also provided at both ends of the fuselage, and the limiting structures at both ends of the fuselage are stuck on the edges of the long sides of the rectangular photovoltaic panel, so that the photovoltaic cleaning robot can move on the photovoltaic panel without falling from the photovoltaic panel; the distance between the limiting structures at both ends of the fuselage is greater than the distance between the long sides of the rectangular photovoltaic panel.

[0051] The parking area and the photovoltaic panel area are a bracket photovoltaic panel group composed of rectangular photovoltaic panels used in a rail-mounted photovoltaic robot, with the two ends of the bracket being the parking area and the middle being the photovoltaic panel area.

[0052] Preferably, the angle formed by the robot in the stationary state before departure from the parking position and the side bracket of the parking position is 90°, with an acceptable error range of ±3°, and according to quantitative tests, the robot's posture will not deviate within 20 seconds from the time the robot accelerates to a stable state after leaving the parking position in the absence of external abnormalities.

[0053] Specifically, when a robot fails or stops running manually, it will enter the fault state or standby state accordingly. In this state, the robot will enter different loop feedback mechanism states according to its location and corresponding recovery conditions. For details, see Figure 2 .

[0054] The beneficial effects of the present invention are as follows: 1) Real-time state detection of the movement of the rail-mounted photovoltaic cleaning robot is realized, and data accuracy is mutually verified by establishing and utilizing a hierarchical feedback adjustment mechanism, and data error correction is performed in a timely manner, and the robot's movement posture is compensated in real time to ensure that the robot can maintain a normal posture for cleaning operation. 2) Closed-loop control using incremental PID technology ensures real-time detection and compensation of the motor's operating speed. 3) No additional devices are required, and deviation correction processing is performed through multi-data fusion adaptive scenarios. 4) Closed-loop control system, establishing an independent closed-loop control system, namely a speed loop-angle loop-current loop system, to ensure the data stability of the robot under various operating conditions, thereby improving the operating efficiency and reliability of the overall system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a structural diagram of the rail-mounted cleaning robot of the present invention;

[0056] Figure 2 The following are curves showing the changes in heading angle data when the robot of the present invention is started at the parking position in three different postures;

[0057] Figure 3 The following is a curve diagram of the filter current change of the main motor at 200 window points when the robot of the present invention is started in three different postures at the parking position;

[0058] Figure 4 The following is a curve diagram of the filter current change from the motor at 200 window points when the robot of the present invention is started in three different postures at the parking position;

[0059] Figure 5 1. The present invention is a graph showing the change in the RMS value of the main motor current when the robot is started at the parking position in three different postures;

[0060] Figure 6 The graph is a curve diagram of the change of the RMS value of the motor current when the robot of the present invention is started at the parking position in three different postures. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The following description of at least one exemplary embodiment is actually merely illustrative and is by no means intended to limit the present application and its application or use. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application. In the following description, many specific details are set forth to facilitate a full understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.

[0062] In this embodiment, the rail-mounted photovoltaic cleaning robot is referred to as the robot. In this embodiment, when the robot is not experiencing a fault or an intentional shutdown, it simply stops and waits at either end. Specifically, in this embodiment, the robot's starting position is referred to as the initial position, hereinafter referred to as the stop position; the robot's arrival position is referred to as the limit position. When the robot stops at a stop position or a limit position, it has completed its cleaning task or is waiting to restart its cleaning task.

[0063] In this embodiment, the adaptive deviation correction method includes a three-level feedback adjustment mechanism, the first level is based on a current loop method, the second level is based on an angle loop method, and the third level is based on a speed loop method.

[0064] In this embodiment, the speed loop method includes: real-time collection of the number of pulses generated by the operation of the master motor and the slave motor, judging whether deviation occurs based on the number of pulses, and if deviation occurs, triggering the speed loop to start the following adjustment mechanism: using the master motor as a reference, using the incremental PID control method to compensate and adjust the operating speed of the slave motor to achieve deviation correction.

[0065] In this embodiment, the angle loop method includes: real-time acquisition of the current heading angle, judging whether deflection occurs based on the current heading angle and a heading angle reference value, and if deflection occurs, triggering the angle loop to start the following adjustment mechanism.

[0066] In this embodiment, the current loop method includes: real-time collection of the current characteristic value, judging whether a deviation occurs based on the current characteristic value and the current characteristic value reference value, and if a deviation occurs, triggering the current loop to start the following regulation mechanism: executing the second-level feedback regulation mechanism.

[0067] The robot is in the operation preparation stage, specifically, Figure 2As shown, within 200 sampling intervals, or 20 seconds, the robot's heading angle will not change by more than 2°, regardless of whether it is started and operated in any of the three initial postures at the parking position: no deviation, main motor deviation, or slave motor deviation. Furthermore, in this embodiment, after testing, the distance traveled by the robot at the minimum speed required by safety regulations for 20 seconds is fully sufficient for the robot to enter the second stage of operation. Therefore, when the robot's heading angle at the parking position is within a preset range, i.e., the angle formed between the robot and the side bracket of the parking position in its static state before departure is 90±3°, the heading angle at the current robot position is collected and used as the heading angle reference value. Specifically, in this embodiment, based on the actual robot operating scenario, an area within X meters of the parking position without photovoltaic panels is defined as the parking position area. When the robot operates within the parking position area, the second-level feedback adjustment mechanism is implemented to correct the deviation.

[0068] In this embodiment, when the robot's running distance is greater than X meters, it leaves the parking area and enters the second stage of operation. The second stage of operation is in the photovoltaic panel area. During the first T time of the second stage of operation, the robot's running speed has not yet accelerated to the set value, and the motor current data has not reached the peak-to-peak value. Therefore, its motor current characteristic data cannot correctly represent the robot's running posture, so the second-level feedback adjustment mechanism is executed to correct the deviation.

[0069] In this embodiment, when the robot's second-stage operating time exceeds T, while the second-level feedback adjustment mechanism is executed for deviation correction, the main motor and slave motor current data are continuously collected and converted into motor current characteristic values. Specifically, when the robot's second-stage operating time exceeds T, the main motor and slave motor current data are continuously collected and averaged, with a sliding window filter of 200 sampling points. Outliers are removed from the main motor and slave motor current data to reduce noise and transient interference. The pre-processed current data is converted into motor current characteristic values ​​using a root mean square (RMS) calculation formula. Simultaneously, a determination is made as to whether the current overall robot state meets the recording conditions for the motor current characteristic reference value. Specifically, a determination is made as to whether the angle loop activation adjustment mechanism is continuously triggered within a preset time of 3 seconds. If so, the current motor current characteristic value is recorded, and step S5 is repeated after a preset time. If the result is still positive, the current motor current characteristic value is recorded, and step S5 is repeated after a preset time of 3 seconds. This is repeated up to M times, and the average of the M recorded current characteristic values ​​is taken as the motor current characteristic reference value.

[0070] like Figure 3 、 Figure 4 、 Figure 5 、 Figure 6As shown in the figure, after testing, it is found that the use of the root mean square value of the current characteristic value of the robot's main motor and slave motor can more accurately and clearly express the operation status of the robot in different postures than directly using the motor current value. Specifically, when using the motor current to judge the different postures of the robot's operation, the current data is difficult to intuitively reflect its trend characteristics. Obviously, when using the root mean square value of the motor current, it can intuitively reflect the trend characteristics of different operating postures, and it is easier to judge whether the robot is in a skewed state, and it can accurately judge whether it is in the main motor skew or the slave motor skew. In comparison, Figure 5 、 Figure 6 High accuracy and low error rate.

[0071] In this embodiment, 150 windows are selected as the motor current characteristic value. Specifically, the collected motor current characteristic baseline values ​​directly reflect the motor current characteristic baseline values ​​when the robot is operating with no deviation, eliminating the need for manually setting thresholds. Next, the current characteristic values ​​of the robot's master and slave motors are compared to see if they are abnormal from the current characteristic baseline values. The comparison results are used as input to the current loop control system, setting this independent loop system as the first-level feedback regulation mechanism.

[0072] In this embodiment, after completing the collection of the motor current characteristic baseline value, the robot enters the third stage of operation. In this stage, the robot still operates within the photovoltaic panel area, and simultaneously executes the first-level feedback adjustment mechanism and the second-level feedback adjustment mechanism to jointly perform correction. Specifically, combined with the above, the first judgment point changes from the second-level feedback adjustment mechanism to the first-level feedback adjustment mechanism.

[0073] In this embodiment, when executing the first-level feedback regulation mechanism, if the current loop triggers the regulation mechanism, that is, the motor current characteristic value differs from the reference value, the system simultaneously detects whether the angle loop triggers the regulation mechanism. If not, it is assumed that the robot is experiencing jolting, vibration, or other disturbances during operation. At this point, provided the robot is operating stably and the angle loop regulation mechanism is not continuously triggered, the current motor current characteristic data is updated to the new motor current characteristic reference value. If the angle loop regulation mechanism is triggered, it is assumed that the robot is deflected, and the incremental PID control method is used to compensate and adjust the operating speed of the slave motors, using the master motor as the reference, to achieve deviation correction.

[0074] Similarly, in this embodiment, when the current loop regulation mechanism is not triggered and the angle loop regulation mechanism is triggered, it is considered that the robot's heading angle data has accumulated errors during operation and needs to be cleared. At this time, when the robot is stable and the regulation mechanism is not triggered, the current heading angle data is updated to the new heading angle reference value. During the third stage of operation, when neither the current loop regulation mechanism nor the angle loop regulation mechanism is triggered, it is considered that the robot has not deflected. In this case, the current motor current characteristic data is updated to the new motor current characteristic reference value, and the current heading angle data is updated to the new heading angle reference value every Y meters the robot travels.

[0075] In this embodiment, the robot is considered to have a fault or be stopped midway due to human intervention, entering a fault or standby state. In this case, when it restarts, the historical log is first retrieved to trace the last distance traveled. Specifically, the robot's position is determined, and different hierarchical adjustment mechanisms are used according to different positions to ensure cleaning efficiency and error correction accuracy. Obviously, after the robot leaves its initial stop position, a limit sensor is activated for detection. When the limit sensor is triggered, the robot reaches the final stop position, at which point the robot reverses its direction and repeats the above steps to continue the cleaning task. The present invention proposes an adaptive error correction method for a photovoltaic cleaning robot based on a three-level feedback adjustment mechanism, which enables real-time monitoring of the robot's operating status and effective error correction. This method is characterized by real-time, high fault tolerance, and adaptability. It does not require manual threshold setting and achieves real-time error correction and update of data through the mutual correction of multiple data. It is particularly suitable for complex environments such as photovoltaic power plants and where track accuracy is required, greatly improving the robot's cleaning efficiency and operational stability.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism, characterized in that: The adaptive deviation correction method includes a three-level feedback adjustment mechanism, the first level is based on the current loop method, the second level is based on the angle loop method, and the third level is based on the speed loop method; The speed loop method includes: real-time acquisition of the number of pulses generated by the operation of the master motor and the slave motor, judging whether deflection occurs based on the pulse number, and if deflection occurs, triggering the speed loop to start the following adjustment mechanism: using the master motor as a reference, using the incremental PID control method to compensate and adjust the operating speed of the slave motor to achieve deviation correction; The angle loop method includes: collecting the current heading angle in real time, judging whether deflection occurs based on the current heading angle and a heading angle reference value, and if deflection occurs, triggering the angle loop to start an adjustment mechanism; The current loop method includes: collecting the current current characteristic value in real time, judging whether a deflection occurs based on the current current characteristic value and a current characteristic value reference value, and if a deflection occurs, triggering the current loop to start the following regulation mechanism: executing a second-level feedback regulation mechanism; The adaptive deviation correction method comprises the following steps: S1: If the heading angle of the robot at the parking position is within the preset range, the heading angle of the robot at the parking position is collected as the heading angle reference value, and the main motor and the slave motor are started to enter the first stage of operation; S2, based on the actual working scenario of the robot, defines the area without photovoltaic panels within X meters of the parking position as the parking position area. When the robot operates in the parking position area, the second-level feedback adjustment mechanism is executed to correct the deviation; S3, when the robot runs a distance greater than X meters, during the first T time of the second stage of operation, the second-level feedback adjustment mechanism is executed to correct the deviation; S4: When the robot runs for longer than T in the second stage, it continuously collects the master motor current and slave motor current data while executing the second-level feedback adjustment mechanism to correct the deviation, and converts them into motor current characteristic values; S5, determining whether the angle loop is continuously triggered to start the adjustment mechanism within a preset time; if not, taking the current motor current characteristic value as the motor current characteristic reference value; S6, if yes, record the current motor current characteristic value, wait for a preset time and then repeat step S5; if still yes, record the current motor current characteristic value, wait for a preset time and then repeat step S5, repeat up to M times, and take the average of the M recorded current characteristic values ​​as the motor current characteristic reference value; S7, after obtaining the motor current characteristic reference value, executing the first-level feedback adjustment mechanism and the second-level feedback adjustment mechanism to jointly perform deviation correction.

2. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 1, characterized in that: After step S7, the following steps are also included: S8: When the robot is in the third stage of operation, if neither the current loop start adjustment mechanism nor the angle loop start adjustment mechanism is triggered, the robot updates the current motor current characteristic value and heading angle to the motor current characteristic reference value and heading angle reference value each time the robot runs a preset distance. S9, if the current loop start adjustment mechanism is not triggered, but the angle loop start adjustment mechanism is triggered: the motor current characteristic reference value is updated to the current motor current characteristic value; S10, if the current loop is triggered to start the adjustment mechanism, but the angle loop is not triggered to start the adjustment mechanism: the heading angle reference value is updated to the current heading angle; At the same time, in S11, during the third stage of operation, it is detected whether there is a limit sensor trigger. If not, steps S8-10 are repeated. If triggered, the robot reaches the end stop position, the robot changes its running direction, and steps S1-11 are repeated.

3. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 1, characterized in that: In S1, the preset range is 90°±3°.

4. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 1, characterized in that: In S4, the conversion into motor current characteristic values ​​specifically includes preprocessing the current data including reducing noise, smoothing data, and processing abnormal values; and converting the preprocessed current data into motor current characteristic values ​​through a root mean square calculation formula.

5. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 1, characterized in that: In S5 and S6, the preset time is 3 seconds.

6. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 1, characterized in that: The output is used to trigger the speed loop start adjustment mechanism signal, specifically including: outputting a fixed number of pulses to increase or decrease the number of pulses of the main motor, thereby triggering the speed loop start adjustment mechanism to compensate the slave motor speed, thereby correcting the robot's running posture.

7. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 2, characterized in that: In S8, the preset distance is recorded as Y, and the setting value interval of Y is [5, 10] meters; the setting value interval of M is [3, 5] times; the setting value interval of X is [1, 1.2] meters; and the setting value interval of T is [10, 15] seconds.

8. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 2, characterized in that: The photovoltaic cleaning robot is used for cleaning rectangular photovoltaic panels.

9. The method for adaptive deviation correction of a photovoltaic cleaning robot based on a three-level feedback adjustment mechanism according to claim 2, characterized in that: The heading angle is the attitude angle in the Z-axis direction calculated by the six-axis sensor mounted on the robot main control board.

10. The photovoltaic cleaning robot adaptive deviation correction method based on a three-level feedback adjustment mechanism according to claim 2, characterized in that: The determining whether skew occurs based on the number of pulses specifically includes: Get the circumference of the driving wheel and the driven wheel respectively; The distance traveled by one end of the fuselage where the driving wheel is located is calculated by multiplying the circumference of the driving wheel by the number of pulses of one revolution of the main motor; The distance traveled by the end of the fuselage where the driven wheel is located is calculated by multiplying the circumference of the driven wheel by the number of pulses in one revolution of the slave motor; Whether or not a skew occurs is determined based on the difference in the distances.

Images