Photovoltaic power station AI automatic cleaning system and method

Abstract

The invention provides an AI automatic cleaning system and method for a photovoltaic power station. A photovoltaic panel environment sensing subsystem of the system is responsible for acquiring environment data and identifying position distribution, hardness and type of stains. And the photovoltaic panel fusion decision subsystem makes a cleaning strategy and generates an instruction by using the information. The cleaning robot execution system executes the instructions to clean stains. And the feedback subsystem feeds back the cleaning result to the photovoltaic panel fusion decision subsystem to optimize the strategy. And the upper computer monitors and manages the states of other subsystems in real time. Through comprehensive consideration of deep environment data around a photovoltaic panel, optimization decision of a cleaning path and a mode is carried out, a cleaning strategy is intelligently recommended to realize AI automatic cleaning, deep cleaning and accurate identification of positions and types of stains to optimize the cleaning strategy, and dynamic adjustment is carried out according to a real-time result to ensure an optimal cleaning effect. The remote monitoring function enables a user to know the working state in real time, improves the stability and reliability of the system, and reduces the labor cost at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent cleaning, in particular to an AI automatic cleaning system, method, device and storage medium for a photovoltaic power station. Background Art

[0002] Currently, the cleaning work of a photovoltaic power station mainly relies on manual labor or traditional automatic cleaning equipment. Although manual cleaning has relatively good cleaning effects, it has problems such as low efficiency, high labor costs, and high safety risks. Traditional automatic cleaning equipment, such as cleaning robots, although improving the cleaning efficiency to a certain extent, still has deficiencies such as insufficient intelligence and limited adaptability to complex environments, resulting in poor cleaning effects.

[0003] Therefore, in the face of the cleaning work of photovoltaic panels in a photovoltaic power station, the traditional cleaning methods mainly have the following application deficiencies:

[0004] Limited stain removal ability, for example, stubborn stains are not thoroughly removed: the removal rate of bird droppings / oil stains is only 68% (92% for manual cleaning);

[0005] Shortcomings in the perception system. Currently, the intelligent cleaning only perceives the type, position, etc. of stains on the surface of the photovoltaic panel, and makes cleaning strategy decisions based on this. Therefore, the strategy for controlling its robot to perform automatic cleaning can only be combined with the type and position of stains and other obstacles for cleaning, without comprehensively considering the depth environment data around the photovoltaic panel for decision-making, resulting in insufficiently refined and in-depth cleaning strategies and incomplete cleaning;

[0006] Unable to generate a cleaning decision-making plan based on depth information, resulting in incomplete stain cleaning. Summary of the Invention

[0007] In order to solve the technical problem of poor cleaning effects in the prior art, the present invention provides the following technical solutions:

[0008] On the one hand, an AI automatic cleaning system for a photovoltaic power station is provided. The system includes:

[0009] A photovoltaic panel environment perception subsystem, configured to obtain environmental data, extract the position distribution information and hardness information of stains from the environmental data, and judge the type information of the stains;

[0010] A photovoltaic panel fusion decision-making subsystem, connected to the photovoltaic panel environment perception subsystem, configured to automatically formulate a cleaning strategy by using the position distribution information, hardness information and type information of the stains, and generate a cleaning instruction;

[0011] A cleaning robot execution system, connected to the photovoltaic panel fusion decision-making subsystem, configured to receive and execute the cleaning instruction to clean the stains of the photovoltaic power station;

[0012] A feedback subsystem, connected to the cleaning robot execution system, is configured to obtain cleaning results and feed back the cleaning results to the photovoltaic panel fusion decision subsystem, and the photovoltaic panel fusion decision subsystem optimizes the cleaning strategy based on the cleaning results;

[0013] A host computer, connected to the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system, is configured to receive in real time the status data of the photovoltaic panel environment perception subsystem, the status data of the photovoltaic panel fusion decision subsystem, and the status data of the cleaning robot execution system, and remotely monitor and manage the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system according to the status data of the photovoltaic panel environment perception subsystem, the status data of the photovoltaic panel fusion decision subsystem, and the status data of the cleaning robot execution system.

[0014] As an optional embodiment of the present invention, optionally, the photovoltaic panel environment perception subsystem includes:

[0015] An image acquisition subsystem, configured to acquire image information on the surface of the photovoltaic panel;

[0016] An ultrasonic detection subsystem, configured to detect ultrasonic feedback information of stains around the photovoltaic panel and calculate hardness data of the corresponding stains;

[0017] A stain detection subsystem, connected to the image acquisition subsystem, is configured to identify stains in the image information, obtain position information of the stains, and count the position distribution density of stains around on the photovoltaic panel;

[0018] A stain type judgment subsystem, connected to the stain detection subsystem, is configured to obtain the cleaning mode of the stains.

[0019] As an optional embodiment of the present invention, optionally, the stain type judgment subsystem identifies the cleaning mode of the stains through a stain recognition judgment model;

[0020] [[ID= 26]]The stain recognition judgment model includes an input layer, a hidden layer, and an output layer;

[0021] The expression of the input layer is:

[0022] x = [x1, x2,..., x n

[0023] where x represents a combined feature vector of the input stain type and stain hardness, and n represents the number of combined feature vectors of the input stain type and stain hardness;

[0024] The expression of the hidden layer is: ​

[0025]

[0026] Among them, z j represents the linear combination result of the j-th neuron in the hidden layer matching the corresponding cleaning mode for stain type and stain hardness, represents the weight from the input layer to the j-th neuron in the hidden layer, x i represents the combined feature vector of the stain type and stain hardness of the i-th input, represents the bias of the j-th neuron, m represents the number of neurons in the hidden layer, a j represents the activation output of the j-th neuron in the hidden layer, σ() represents the Sigmoid activation function, exp() represents the exponential function, and α represents the parameter for adjusting the shape of the Sigmoid activation function;

[0027] The expression of the output layer is:

[0028]

[0029] Among them, z′ k represents the linear combination result of the k-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness, represents the weight from the hidden layer to the k-th neuron in the output layer, represents the bias of the k-th neuron in the output layer, p k represents the prediction probability that the stain type or stain hardness is of the k-th type, exp() represents the activation function of the output layer, β represents the parameter for adjusting the shape of the activation function of the output layer, z l ′ represents the linear combination result of the l-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness, Ξ represents the stain type or stain hardness, represents taking the k maximum value of p

[0030] As an alternative embodiment of the present invention, optionally, the photovoltaic panel fusion decision subsystem includes:

[0031] A data preprocessing subsystem for preprocessing the type information of the stain, the stain hardness data, and the position distribution density of the surrounding stains on the photovoltaic panel;

[0032] A strategy formulation subsystem, connected to the data preprocessing subsystem, for formulating the cleaning strategy based on the preprocessed type information of the stain, the stain hardness data, and the position distribution density of the surrounding stains on the photovoltaic panel. The cleaning strategy includes a distributed cleaning path from high density to low density formulated according to the position distribution density of the stains, and the cleaning intensity and cleaning time formulated according to the stain type and hardness data;

[0033] Optimization subsystem, connected to the policy formulation subsystem, for optimizing the cleaning policy based on the cleaning results sent by the feedback subsystem, generating an optimized cleaning instruction, and sending the optimized cleaning instruction to the cleaning robot execution system;

[0034] Resource scheduling subsystem, connected to the policy formulation subsystem, for scheduling the cleaning robot execution system based on the cleaning policy.

[0035] As an alternative embodiment of the present invention, optionally, the policy formulation subsystem formulates the cleaning policy through a genetic algorithm.

[0036] As an alternative embodiment of the present invention, optionally, the optimization subsystem optimizes the cleaning policy through an optimization algorithm;

[0037] The expression of the optimization algorithm is:

[0038]

[0039] Wherein, represents the updated velocity vector of the cleaning robot execution system, ω represents the inertia weight, v id represents the velocity vector of the cleaning robot execution system before update, c1 and c2 represent learning factors, r1 and r2 represent random numbers uniformly distributed in the range [0,1], pbest id represents the component of the individual optimal position of the i-th particle in the d-th dimension, x id represents the position of the i-th particle in the d-th dimension, gbest d represents the component of the global optimal position of the entire particle swarm in the d-th dimension, represents the new position of the i-th particle in the d-th dimension.

[0040] As an alternative embodiment of the present invention, optionally, the cleaning robot execution system includes:

[0041] Instruction receiving subsystem, for receiving the cleaning instruction;

[0042] Motion control subsystem, connected to the instruction receiving subsystem, for controlling the motion trajectory of the cleaning subsystem according to the cleaning instruction;

[0043] Cleaning subsystem, connected to the motion control subsystem, for executing the control instruction of the motion control subsystem, the control instruction including spraying water or / and brushing or / and blowing.

[0044] On the other hand, the present invention also provides an AI automated cleaning method for a photovoltaic power station. The method includes the AI automated cleaning system of the photovoltaic power station, and the method further includes:

[0045] S1. Obtain the environmental data of the photovoltaic power station, extract the position distribution information and hardness information of the stains from the environmental data, and judge the type information of the stains;

[0046] S2. Automatically formulate a cleaning strategy based on the position distribution information, hardness information, and type information of the stains, and generate a cleaning instruction;

[0047] S3. Receive and execute the cleaning instruction to clean the stains on the photovoltaic power station;

[0048] S4. Obtain the cleaning result of the photovoltaic power station in real time, and optimize the cleaning strategy based on the cleaning result;

[0049] S5. Obtain the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system in real time, and remotely monitor and manage the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system according to the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system.

[0050] On the other hand, a cleaning device is provided. The cleaning device includes: a processor; a memory, on which computer-readable instructions are stored. When the computer-readable instructions are executed by the processor, the above-mentioned AI automated cleaning method for a photovoltaic power station is implemented.

[0051] On the other hand, a computer-readable storage medium is provided. At least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned AI automated cleaning method for a photovoltaic power station.

[0052] The beneficial effects brought by the technical solution provided in the embodiment of the present invention are as follows: Firstly, through the environmental data collected by the photovoltaic panel environment perception subsystem, the present invention can accurately identify the position and type of stains, so as to formulate a more reasonable cleaning strategy. During the cleaning process, the system can dynamically adjust the cleaning strategy according to the cleaning result obtained in real time to ensure that the cleaning effect reaches the best. In addition, by comprehensively considering the depth environmental data around the photovoltaic panel for optimizing the decision-making of the cleaning path and mode, intelligently recommending cleaning strategies to achieve AI automated cleaning, deeply cleaning, accurately identifying the position and type of stains to optimize the cleaning strategy, and dynamically adjusting according to the real-time results for more thorough cleaning. The remote monitoring function enables users to understand the working status in real time, improves the stability and reliability of the system, and reduces the labor cost at the same time. Brief Description of the Drawings

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a schematic structural diagram of the AI automatic cleaning system of a photovoltaic power station provided by an embodiment of the present invention;

[0055] Figure 2 It is a schematic structural diagram of the perception subsystem in an embodiment of the present invention;

[0056] Figure 3 It is a schematic structural diagram of the photovoltaic panel fusion decision-making subsystem in an embodiment of the present invention;

[0057] Figure 4 It is a flowchart of the AI automatic cleaning of a photovoltaic power station provided by an embodiment of the present invention;

[0058] Figure 5 It is a schematic structural diagram of the cleaning equipment in an embodiment of the present invention. Detailed Embodiments

[0059] The technical solutions in the present invention will be described below with reference to the drawings.

[0060] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or more advantageous than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0061] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0062] In the embodiments of the present invention, sometimes subscripts such as W1 may be miswritten as non-subscript forms such as W1. When the difference is not emphasized, the meanings they express are the same.

[0063] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0064] As Figure 1 shown, an AI automatic cleaning system for a photovoltaic power station is provided in an embodiment of the present invention. The system includes:

[0065] A photovoltaic panel environmental perception subsystem, configured to obtain environmental data, extract the position distribution information and hardness information of stains from the environmental data, and determine the type information of the stains;

[0066] It should be noted that the photovoltaic panel environmental perception subsystem can collect image information on the surface of the photovoltaic power station through sensor devices such as high-definition cameras and infrared sensors, and use the YOLOV5 image recognition algorithm and machine learning algorithm to analyze and process the collected data, so as to accurately identify the position and type of stains. For example, for different types of stains, such as dust, bird droppings, leaves, etc., their colors, shapes, textures, etc. will be different, and the system can accurately judge the type of stains by learning and analyzing these features.

[0067] A photovoltaic panel fusion decision-making subsystem, connected to the photovoltaic panel environmental perception subsystem, configured to automatically formulate a cleaning strategy using the position distribution information, hardness information, and type information of the stains, and generate a cleaning instruction;

[0068] It should be noted that the photovoltaic panel fusion decision-making subsystem can intelligently select the most suitable cleaning method and path according to the type and position of the stains and the actual situation of the photovoltaic power station. For example, for dust adhering to the surface of the photovoltaic panel, the system may choose the method of spraying water for cleaning; while for more stubborn stains such as bird droppings or leaves, it will use the method of brushing or a combination of spraying water and brushing for cleaning. At the same time, the photovoltaic panel fusion decision-making subsystem will also optimize the cleaning path according to the data (cleaning results) fed back by the feedback subsystem to ensure that the cleaning work is both efficient and safe.

[0069] A cleaning robot execution system, connected to the photovoltaic panel fusion decision-making subsystem, configured to receive and execute the cleaning instruction to clean the stains of the photovoltaic power station;

[0070] It should be noted that the cleaning robot execution system is the core execution component of the AI automated cleaning system for a photovoltaic power station. It is responsible for receiving cleaning instructions from the photovoltaic panel fusion decision subsystem and precisely controlling the movement trajectory and cleaning operations of the cleaning equipment according to these instructions. The high precision and flexibility of the cleaning robot execution system ensure the quality and efficiency of the cleaning work. When performing the cleaning task, the cleaning robot execution system first receives the cleaning instructions from the photovoltaic panel fusion decision subsystem through the instruction receiving subsystem. These instructions contain detailed cleaning strategies, such as cleaning paths, cleaning intensities, and cleaning times. Subsequently, the motion control subsystem calculates the movement trajectory of the cleaning subsystem based on the received cleaning instructions. In this process, the motion control subsystem will fully consider factors such as the layout of the photovoltaic power station, the location and type of stains, etc., to ensure the optimization of the cleaning path. The cleaning subsystem is responsible for executing the control instructions of the motion control subsystem and removing the stains from the surface of the photovoltaic power station by means of spraying water, brushing, or blowing. While performing the cleaning task, the cleaning robot execution system also monitors its own operating status in real time, including parameters such as the speed, position, and cleaning intensity of the cleaning equipment. These data will be fed back to the photovoltaic panel fusion decision subsystem in real time so that the photovoltaic panel fusion decision subsystem can dynamically adjust the cleaning strategy according to the cleaning effect and actual situation. In this way, a tight closed-loop control system is formed between the cleaning robot execution system and the photovoltaic panel fusion decision subsystem, ensuring the precision and efficiency of the cleaning work. During the cleaning process, if the cleaning robot execution system detects any abnormal situations, such as cleaning equipment failures or movement trajectory deviations, it will immediately stop the cleaning operation and issue an alarm signal. This can not only prevent damage to the cleaning equipment and the photovoltaic power station but also ensure the safety of the operators.

[0071] A feedback subsystem, connected to the cleaning robot execution system, is used to obtain the cleaning result and feed back the cleaning result to the photovoltaic panel fusion decision subsystem, and the photovoltaic panel fusion decision subsystem optimizes the cleaning strategy based on the cleaning result;

[0072] It should be noted that in this embodiment, the feedback subsystem captures the surface state after cleaning in real time through a high-definition camera or other sensor devices and compares it with the image before cleaning to evaluate the cleaning effect. This real-time feedback mechanism enables the system to promptly detect problems during the cleaning process, such as incomplete cleaning or excessive cleaning, and immediately adjust the cleaning strategy. For example, if the feedback subsystem detects that the cleaning effect in a certain area is not good, the photovoltaic panel fusion decision subsystem may increase the cleaning intensity or extend the cleaning time in that area to ensure that all stains are completely removed. In this way, the system can continuously learn and optimize the cleaning strategy, improving the cleaning efficiency and effect.

[0073] The host computer, which is connected to the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision-making subsystem, and the cleaning robot execution system, is used to receive in real time the status data of the photovoltaic panel environment perception subsystem, the status data of the photovoltaic panel fusion decision-making subsystem, and the status data of the cleaning robot execution system, and remotely monitor and manage the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision-making subsystem, and the cleaning robot execution system according to the status data of the photovoltaic panel environment perception subsystem, the status data of the photovoltaic panel fusion decision-making subsystem, and the status data of the cleaning robot execution system.

[0074] It should be noted that the host computer can monitor the working status of each subsystem in real time, and promptly discover and handle possible problems. For example, if the host computer detects that the sensor device of the photovoltaic panel environment perception subsystem fails or the data is abnormal, it will immediately issue an alarm and prompt the maintenance personnel to conduct inspections and repairs. Similarly, if problems such as a decline in performance or incorrect operations occur in the photovoltaic panel fusion decision-making subsystem or the cleaning robot execution system, the host computer will also promptly issue an alarm and take corresponding measures to correct them. In addition, the host computer can also collect and analyze the working data of each subsystem, providing strong support for the further optimization and improvement of the system. This remote monitoring and management function not only improves the stability and reliability of the system, but also reduces the maintenance cost and time cost, bringing a better user experience to users.

[0075] Specifically, the host computer in this embodiment is a remote management platform based on cloud computing. This platform can receive and process data from the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision-making subsystem, and the cleaning robot execution system in real time. Through cloud computing technology, the host computer can efficiently store, analyze, and display this data, enabling users to intuitively understand the operating status of the system. In addition, this platform also provides rich management functions, such as subsystem configuration, data query, alarm management, etc., facilitating users to comprehensively monitor and manage the system.

[0076] In summary, the AI automated cleaning system of the photovoltaic power station in this embodiment first collects the image information of the surface of the photovoltaic power station through the photovoltaic panel environment perception subsystem, and uses advanced image recognition algorithms and machine learning technologies to achieve accurate recognition of the stain position and type. This ability provides a solid foundation for formulating subsequent cleaning strategies, ensuring the pertinence and effectiveness of the cleaning work. When performing the cleaning task, the cleaning robot execution system, with its highly accurate motion control and flexible cleaning operations, can accurately clean the stains on the surface of the photovoltaic power station according to the cleaning strategy formulated by the photovoltaic panel fusion decision subsystem. At the same time, through the real-time feedback mechanism of the feedback subsystem, the system can continuously learn and optimize the cleaning strategy to meet the needs of different stain types and cleaning environments. In addition, the introduction of the upper computer enables users to monitor the operation status of the system in real time, discover and handle possible problems in a timely manner, and further improve the stability and reliability of the system. This intelligent cleaning equipment and method not only improves the cleaning efficiency and quality of the photovoltaic power station, but also reduces the labor cost and maintenance cost, bringing a new solution for the operation and maintenance management of the photovoltaic power station.

[0077] As an alternative embodiment of the present invention, optionally, the photovoltaic panel environment perception subsystem includes:

[0078] An image acquisition subsystem for acquiring image information of the surface of the photovoltaic panel;

[0079] As [[ID= shown, it should be noted that the image acquisition subsystem uses a high-definition camera or other image acquisition devices to perform high-resolution image acquisition of the surface of the photovoltaic panel. This image information will be used for subsequent analysis and processing to identify the position and type of stains.

[0080] An ultrasonic detection subsystem for detecting the ultrasonic feedback information of the stains around the photovoltaic panel and calculating the hardness data of the corresponding stains; through the ultrasonic detection probe deployed on the robot, the ultrasonic feedback information of the stains around the photovoltaic panel can be detected, and the system further performs hardness detection according to the ultrasonic feedback information of the stains. Specifically:

[0081] The ultrasonic probe carried by the robot emits high-frequency sound waves (frequency range 20 kHz - 1 MHz) to the surface of the photovoltaic panel. The acoustic impedance difference between the stain and the photovoltaic panel substrate will cause changes in the reflection intensity, phase, and spectral characteristics of the echo signal. By analyzing the echo time delay difference and frequency domain attenuation characteristics, the stain type (such as dust, bird droppings, oil stains) can be distinguished and its hardness can be estimated. Example: The acoustic wave reflection intensity of hard stains (such as bird droppings) is 30% - 50% higher than that of soft stains (such as dust), and the high-frequency components in the spectrum attenuate more significantly.

[0082] And its hardness algorithm is as follows:

[0083]

[0084] H is the stain hardness index; it can be determined by combining the industry or the hardness index method;

[0085] A max is the echo peak amplitude; detected by the ultrasonic system;

[0086] A base is the reference amplitude of the clean surface of the photovoltaic panel; detected by the ultrasonic system;

[0087] S(f) is the spectral energy density of the echo signal; detected by the ultrasonic system;

[0088] α, β are weight coefficients (calibrated through experiments).

[0089] Therefore, the cleaning parameters can be dynamically adjusted according to the stain hardness: hard stains trigger high-pressure water jets (pressure > 5 MPa) or mechanical brushing; soft stains use low-pressure spraying (pressure < 2 MPa) combined with a rotating brush. Ultrasonic detection does not require direct contact with the surface of the photovoltaic panel, reducing physical damage (such as scratches and corrosion) to the photovoltaic coating during the cleaning process and extending the life of the component. A heat map of the stain distribution is generated in real time, and the cleaning path is optimized through the edge computing module to improve efficiency (the cleaning speed is significantly improved compared to the traditional fixed path cleaning).

[0090] The stain detection subsystem, connected to the image acquisition subsystem, is used to identify the stains in the image information and obtain the position information of the stains, and to count the position distribution density of the surrounding stains on the photovoltaic panel;

[0091] It should be noted that the stain detection subsystem analyzes and processes the image information collected by the image acquisition subsystem through the YOLOV5 image recognition algorithm. This algorithm can accurately identify the stains in the image and obtain the position information of the stains.

[0092] The stain type judgment subsystem, connected to the stain detection subsystem, is used to obtain the cleaning mode of the stains.

[0093] It should be noted that the stain type judgment subsystem further analyzes and judges the stains identified by the stain detection subsystem through machine learning algorithms to determine the type of stains (or hardness, which is the same as the treatment of stain types here, so it will not be elaborated further). This step is crucial for formulating subsequent cleaning strategies because different types of stains may require different cleaning methods and intensities. For example, for relatively loose stains such as dust, the system may choose the method of spraying water for cleaning; while for relatively stubborn stains such as bird droppings or leaves, the method of brushing or a combination of spraying water and brushing will be used for cleaning. In this way, the system can ensure the pertinence and effectiveness of the cleaning work, and improve the cleaning efficiency and effect.

[0094] As an optional embodiment of the present invention, optionally, the stain type judgment subsystem identifies the cleaning mode of the stain through a stain recognition and judgment model;

[0095] The stain recognition and judgment model includes an input layer, a hidden layer, and an output layer;

[0096] The expression of the input layer is:

[0097] x = [x1, x2,..., x n

[0098] where x represents the combined feature vector of the input stain type and stain hardness, and n represents the number of combined feature vectors of the input stain type and stain hardness;

[0099] The expression of the hidden layer is:

[0100]

[0101] where z j represents the linear combination result of the j-th neuron in the hidden layer for the stain type and stain hardness to match the corresponding cleaning mode, represents the weight from the input layer to the j-th neuron in the hidden layer, x i represents the i-th combined feature vector of the input stain type and stain hardness, represents the bias of the j-th neuron, m represents the number of neurons in the hidden layer, a j represents the activation output of the j-th neuron in the hidden layer, σ() represents the Sigmoid activation function, exp() represents the exponential function, and α represents the parameter for adjusting the shape of the Sigmoid activation function;

[0102] The expression of the output layer is:

[0103]

[0104] where z′ k ​It represents the linear combination result of the k-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness. It represents the weight from the hidden layer to the k-th neuron in the output layer. It represents the bias of the k-th neuron in the output layer, p k It represents the predicted probability that the stain type or stain hardness is the k-th category. exp() represents the activation function of the output layer, β represents the parameter that adjusts the shape of the output layer activation function, z l ′ represents the linear combination result of the l-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness. Ξ represents the stain type or stain hardness. It represents taking p k The maximum value.

[0105] It should be noted that the stain recognition and judgment model is a deep learning model trained with a large amount of data and can accurately identify the type of stain. During the training process, the model gradually forms the ability to judge the stain type by learning the characteristics of different stains, such as color, shape, texture, etc. When new image information is input, the model can quickly extract the features in the image and match them with the existing stain types or hardness or their combinations, so as to obtain the most likely stain type or the linear combination structure with hardness. This method of judging stain type based on deep learning not only improves the accuracy and efficiency of judgment, but also provides strong support for the subsequent formulation of cleaning strategies. It can recommend corresponding cleaning modes according to different stain types or the linear combination with hardness (that is, the "cleaning intensity and cleaning time" recommended according to the stain type or the linear combination with hardness), so as to achieve intelligent cleaning decision-making, enabling the robot to adaptively adjust the cleaning intensity and cleaning time according to the stain type and hardness on the surface of the photovoltaic panel. By continuously optimizing and improving the structure and parameters of the model, the accuracy of stain type judgment can be further improved, providing a more intelligent solution for the cleaning work of the photovoltaic power station.

[0106] The basic model of the stain recognition and judgment model can be an RF model or other classification decision models. For the input data set of the input layer, it can be prepared with reference to the above description and can be prepared by the user himself (the specific model training, application, verification, and deployment can refer to the application principle of traditional classification models).

[0107] As an optional embodiment of the present invention, optionally, the photovoltaic panel fusion decision subsystem includes:

[0108] A data preprocessing subsystem for preprocessing the type information of the stain, the stain hardness data, and the position distribution density of the surrounding stains on the photovoltaic panel.

[0109] Such as ​As shown, it should be noted that the data preprocessing subsystem first cleans and organizes the type information and location information of the stains, removes redundant and incorrect data, and ensures the accuracy and reliability of subsequent analysis. This step is crucial for improving the performance of the photovoltaic panel fusion decision-making subsystem because accurate data is the basis for formulating effective cleaning strategies. Subsequently, the data preprocessing subsystem also encodes and formats the stain information to meet the requirements of algorithms and models in the photovoltaic panel fusion decision-making subsystem. For example, for the location information of the stains, the data preprocessing subsystem may convert it into coordinates or area identifiers for convenient subsequent calculation and analysis. In this way, the data preprocessing subsystem provides high-quality data input for the photovoltaic panel fusion decision-making subsystem, ensuring the scientificity and effectiveness of the cleaning strategy. Next, the strategy formulation subsystem formulates specific cleaning strategies based on the preprocessed stain information. In this process, the strategy formulation subsystem fully considers factors such as the type, location, size of the stains, the layout of the photovoltaic power station, and the capabilities of the cleaning equipment to ensure the rationality and feasibility of the cleaning strategy. For example, for different types of stains, the strategy formulation subsystem may choose different cleaning methods and intensities; for large areas or difficult-to-clean areas, it may adopt strategies such as area-by-area cleaning or multiple cleaning. In this way, the strategy formulation subsystem provides clear cleaning instructions for the cleaning robot execution system, ensuring the smooth progress of the cleaning work. At the same time, the strategy formulation subsystem also dynamically adjusts the cleaning strategy according to the cleaning effect and actual situation to meet the needs of different stain types and cleaning environments. This intelligent decision-making method not only improves the efficiency and effect of the cleaning work but also brings a new solution for the operation and maintenance management of the photovoltaic power station. <> <>

[0110] The strategy formulation subsystem, connected to the data preprocessing subsystem, is used to formulate the cleaning strategy based on the type information of the preprocessed stains, the stain hardness data, and the position distribution density of the surrounding stains on the photovoltaic panel. The cleaning strategy includes a distributed cleaning path from high density to low density formulated according to the position distribution density of the stains, and the cleaning intensity and cleaning time formulated according to the stain type and hardness data; <> <>

[0111] It should be noted that when formulating the cleaning strategy, the strategy formulation subsystem will comprehensively consider the location distribution information, hardness information, and type information of the stains. Through precise calculation and analysis, the strategy-making subsystem can generate the optimal cleaning path (specifically, a distributed cleaning path from high density to low density is formulated based on the location and density of the stains. This achieves distributed cleaning, avoids irregular cleaning paths, ensures that high-density stains are cleaned first, and avoids the inability to effectively and extensively clean high-density stains due to wear, power issues, and other issues in the execution system. During specific implementation: the location of each stain in the surrounding area is identified through images, and the system then records the location distribution of all surrounding stains to generate a location density distribution heat map. The photovoltaic array can be divided into 1m×1m grid cells, and the density level of each cell is annotated in conjunction with the GIS system to generate a cleaning priority map. Next, based on the Greedy Algorithm, high-density areas are prioritized and then connected to low-density areas (areas are divided and labeled from high to low by the user) with the shortest path, minimizing the robot's movement distance, thereby generating a distributed cleaning path from high density to low density) to ensure that the cleaning equipment can efficiently cover all distribution areas that need to be cleaned. The strategy-making subsystem also selects the appropriate cleaning intensity and duration based on the stubbornness (hardness) of the stain to avoid unnecessary damage to the photovoltaic panels. This intelligent decision-making approach not only improves the accuracy and efficiency of cleaning, but also effectively extends the service life of the photovoltaic panels, bringing better economic benefits to users.

[0112] an optimization subsystem connected to the strategy formulation subsystem, configured to optimize the cleaning strategy based on the cleaning results sent by the feedback subsystem, generate optimized cleaning instructions, and send the optimized cleaning instructions to the cleaning robot execution system;

[0113] It's important to note that after receiving the cleaning results from the feedback subsystem, the optimization subsystem conducts a detailed analysis and evaluation. During this process, the optimization subsystem focuses on key metrics such as cleaning effectiveness, cleaning time, and resource consumption to identify potential deficiencies in the cleaning strategy. For example, if the cleaning results are poor, the optimization subsystem will determine that the cleaning force is insufficient or the cleaning path is inappropriate, and will adjust the cleaning strategy accordingly. Similarly, if the cleaning time is too long or resource consumption is excessive, the optimization subsystem may also optimize the cleaning strategy to improve cleaning efficiency and reduce costs. Through continuous learning and optimization, the optimization subsystem can gradually improve the scientific nature and effectiveness of the cleaning strategy, providing a more intelligent solution for cleaning in photovoltaic power plants. This feedback-based optimization approach not only improves the quality and efficiency of cleaning operations but also provides strong support for continuous improvement and optimization of the system.

[0114] A resource scheduling subsystem, connected to the policy formulation subsystem, is used to schedule the cleaning robot execution system based on the cleaning policy.

[0115] It should be noted that after receiving the cleaning policy formulated by the policy formulation subsystem, the resource scheduling subsystem will immediately start executing the resource scheduling task. It will first determine the required cleaning equipment according to the specific requirements of the cleaning policy, such as the cleaning path, cleaning intensity, and cleaning time. During the resource allocation process, the resource scheduling subsystem will also fully consider the performance and efficiency of the equipment to ensure that the cleaning task can be completed efficiently and accurately. In this way, the resource scheduling subsystem provides strong support for the cleaning robot execution system and ensures the smooth progress of the cleaning work.

[0116] As an optional embodiment of the present invention, optionally, the policy formulation subsystem formulates the cleaning policy through a genetic algorithm.

[0117] It should be noted that the genetic algorithm is an optimization algorithm based on the principles of biological evolution. It searches for the optimal solution in the solution space by simulating natural selection and genetic mechanisms. During the formulation of the cleaning policy, the policy formulation subsystem will first generate a set of initial cleaning policies as candidate solutions. These candidate solutions will undergo a series of operations such as selection, crossover, and mutation, and gradually evolve into better solutions. Specifically, the policy formulation subsystem will design a suitable fitness function according to information such as the type, location, and size of the stains, as well as constraint conditions such as the capabilities of the cleaning equipment and the layout of the photovoltaic power station to evaluate the advantages and disadvantages of each candidate solution. Then, through the selection operation, the candidate solutions with higher fitness are retained, through the crossover operation, the advantages of different candidate solutions are combined, and through the mutation operation, a new solution space is introduced, so as to continuously iterate and evolve into the optimal cleaning policy. This method of formulating the cleaning policy based on the genetic algorithm not only improves the diversity and optimization efficiency of the policy, but also brings a more intelligent solution for the cleaning work of the photovoltaic power station. Through continuous learning and evolution, the system can gradually adapt to different stain types and cleaning environments, providing more scientific and effective support for the operation and maintenance management of the photovoltaic power station.

[0118] As an optional embodiment of the present invention, optionally, the optimization subsystem optimizes the cleaning policy through an optimization algorithm;

[0119] The expression of the optimization algorithm is:

[0120]

[0121] Wherein, represents the updated velocity vector of the cleaning robot execution system, ω represents the inertia weight, v iddenotes the velocity vector before the update of the cleaning robot execution system, c1 and c2 denote learning factors, r1 and r2 denote random numbers uniformly distributed in the range [0, 1], and pbest id denotes the component of the individual optimal position of the i-th particle in the d-th dimension, and x id denotes the position of the i-th particle in the d-th dimension, and gbest d denotes the component of the global optimal position of the entire particle swarm in the d-th dimension, denotes the new position of the i-th particle in the d-th dimension.

[0122] As an alternative embodiment of the present invention, optionally, the cleaning robot execution system includes:

[0123] An instruction receiving subsystem for receiving the cleaning instruction;

[0124] It should be noted that the instruction receiving subsystem is a communication subsystem responsible for receiving cleaning instructions from the photovoltaic panel fusion decision subsystem. These cleaning instructions contain specific contents of the cleaning strategy, such as cleaning paths, cleaning intensities, and cleaning times. The instruction receiving subsystem ensures that the cleaning instructions can be accurately conveyed to the execution unit through an efficient and reliable communication method, providing clear guidance for subsequent cleaning work.

[0125] A motion control subsystem connected to the instruction receiving subsystem for controlling the motion trajectory of the cleaning subsystem according to the cleaning instruction;

[0126] It should be noted that the motion control subsystem is the controller of the cleaning subsystem, which is responsible for precisely controlling the motion trajectory of the cleaning subsystem according to information such as cleaning paths, cleaning intensities, and cleaning times in the cleaning instruction.

[0127] A cleaning subsystem connected to the motion control subsystem for executing the control instructions of the motion control subsystem, and the control instructions include spraying water or / and brushing or / and blowing.

[0128] It should be noted that the cleaning subsystem is the component that actually executes the cleaning task. It performs operations such as spraying water, brushing, or blowing according to the control instructions of the motion control subsystem to effectively remove stains on the photovoltaic panel. The spraying operation can be achieved through a high-pressure water gun or nozzle, using the impact force of the water flow to wash away the stains; the brushing operation is performed by a rotating or moving brush, generating friction with the surface of the photovoltaic panel to clean stubborn stains; the blowing operation uses high-pressure air flow to blow away dust and fine particles on the photovoltaic panel. These operations can be combined and adjusted according to actual needs to achieve the best cleaning effect. Through this intelligent cleaning method, the cleaning subsystem brings a more convenient and efficient solution for the operation and maintenance management of photovoltaic power stations.

[0129] As ​ shown, for the AI automatic cleaning method of a photovoltaic power station, the method includes the AI automatic cleaning system of the photovoltaic power station, and the method further includes:

[0130] S1. Obtain the environmental data of the photovoltaic power station, extract the position distribution information and hardness information of the stains from the environmental data, and judge the type information of the stains;

[0131] It should be noted that step S1 is the starting step of the AI automatic cleaning method of the photovoltaic power station. It first obtains the environmental data of the photovoltaic power station in real time through devices such as sensors or cameras. These data include the images of the photovoltaic panels. Then, the system processes and analyzes the images to extract the position information of the stains, such as the coordinates or areas where the stains are located. At the same time, the system also uses algorithms such as deep learning models to judge the type of the stains, such as distinguishing whether it is dust, bird droppings, leaves or other types of stains. The accuracy and efficiency of this step are crucial for the subsequent cleaning work, because it directly affects the formulation and execution of the cleaning strategy.

[0132] S2. Automatically formulate a cleaning strategy based on the position distribution information, hardness information and type information of the stains, and generate a cleaning instruction;

[0133] It should be noted that step S2 is the core step of automatically formulating a cleaning strategy based on the stain position information and type information obtained in step S1. In this step, the photovoltaic panel fusion decision subsystem will comprehensively consider factors such as the position distribution information, hardness information and type information of the stains, and generate the optimal cleaning instruction after comprehensive consideration. These cleaning instructions contain key information such as the cleaning path, cleaning intensity and cleaning time, providing clear guidance for the subsequent cleaning work. Through an intelligent decision-making method, step S2 can ensure the rationality and feasibility of the cleaning strategy, improve the efficiency and effect of the cleaning work. At the same time, step S2 will also dynamically adjust and optimize the cleaning strategy according to the real-time feedback of the cleaning results to meet the requirements of different stain types and cleaning environments. This data-driven decision-making method not only improves the intelligent level of the cleaning work, but also brings a more scientific and effective solution for the operation and maintenance management of the photovoltaic power station.

[0134] S3. Receive and execute the cleaning instruction to clean the stains of the photovoltaic power station;

[0135] It should be noted that the step S3 is the execution step of the AI - automated cleaning method for a photovoltaic power station. In this step, the cleaning robot execution system receives the cleaning instructions from the photovoltaic panel fusion decision subsystem, and precisely controls the movement trajectory and cleaning operations of the cleaning subsystem according to the information such as the cleaning path, cleaning intensity, and cleaning time in the instructions. The cleaning subsystem performs operations such as spraying water, brushing, or blowing according to the instruction requirements to effectively remove the stains on the photovoltaic panels.

[0136] S4. Obtain the cleaning result of the photovoltaic power station in real - time, and optimize the cleaning strategy based on the cleaning result;

[0137] It should be noted that the step S4 is the optimization step of the AI - automated cleaning method for a photovoltaic power station. In this step, the feedback subsystem obtains the cleaning result of the photovoltaic power station in real - time, including the images and data of the photovoltaic panels after cleaning, etc. Then, the optimization subsystem conducts a detailed analysis and evaluation of these cleaning results to identify possible deficiencies in the cleaning strategy. Based on these analysis results, the optimization subsystem makes corresponding adjustments and optimizations to the cleaning strategy to improve the cleaning efficiency and reduce costs. This optimization method based on the feedback mechanism not only improves the quality and efficiency of the cleaning work but also provides strong support for the continuous improvement and optimization of the system. Through continuous learning and optimization, the system can adapt to different stain types and cleaning environments, providing a more scientific and effective solution for the operation and maintenance management of the photovoltaic power station.

[0138] S5. Obtain the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system in real - time, and remotely monitor and manage the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system according to the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system.

[0139] It should be noted that step S5 is the monitoring and management step of the AI - automated cleaning method for a photovoltaic power station. In this step, the system will obtain the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision - making subsystem, and the cleaning robot execution system in real - time. These data include the operating status, performance parameters, and error information of each subsystem. Through the remote monitoring and management platform, the operation and maintenance personnel can view these status data in real - time, understand the overall operating conditions of the system and the working status of each subsystem. Once an abnormal situation or fault is detected, the operation and maintenance personnel can immediately take measures to handle it to ensure the stable operation of the system and the smooth progress of the cleaning work. At the same time, the remote monitoring and management platform can also provide functions such as data analysis, alarm prompts, and fault diagnosis, providing more comprehensive and convenient management means for the operation and maintenance personnel. Through this intelligent monitoring and management method, the AI - automated cleaning system of the photovoltaic power station can achieve more efficient and reliable operation and maintenance management, providing a strong guarantee for the long - term stable operation of the photovoltaic power station.

[0140] ​ It is a schematic structural diagram of a cleaning device provided by an embodiment of the present invention. Optionally, the cleaning device 410 may include a first processor 2001.

[0141] Optionally, the cleaning device 410 may further include a memory 2002 and a transceiver 2003.

[0142] Among them, the first processor 2001 is connected to the memory 2002 and the transceiver 2003, for example, through a communication bus.

[0143] Next, in combination with ​ Specific introductions will be made to the various components of the cleaning device 410:

[0144] Among them, the first processor 2001 is the control center of the cleaning device 410, which can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or can be an application - specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention, such as: one or more digital signal processors (DSPs), or one or more field - programmable gate arrays (FPGAs).

[0145] Optionally, the first processor 2001 can execute various functions of the cleaning device 410 by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0146] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as ​ CPU0 and CPU1 shown in

[0147] In a specific implementation, as an embodiment, the cleaning device 410 may also include multiple processors, such as ​ the first processor 2001 and the second processor 2004 shown in

[0148] Among them, the memory 2002 is used to store software programs for implementing the solution of the present invention and is controlled by the first processor 2001 for execution. The specific implementation manner can refer to the above method embodiments and will not be elaborated here.

[0149] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001 or exist independently and be coupled to the first processor 2001 through an interface circuit ( ​ not shown in

[0150] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0151] Optionally, the transceiver 2003 may include a receiver and a transmitter ( ​ not shown separately). Among them, the receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0152] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently, and is coupled to the first processor 2001 through an interface circuit ( ​ not shown) of the cleaning device 410. The embodiments of the present invention do not make specific limitations on this.

[0153] It should be noted that ​ the structure of the cleaning device 410 shown does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0154] In addition, the technical effects of the cleaning device 410 can refer to the technical effects of the AI automatic cleaning system of the photovoltaic power station described in the above method embodiments, and will not be elaborated here.

[0155] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0156] It should also be understood that the memory in the embodiments of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0157] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0158] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context.

[0159] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0160] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0161] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0162] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0163] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections between each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0164] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0165] In addition, the functional units in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0166] When the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0167] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. The AI automatic cleaning system for a photovoltaic power station is characterized in that The system includes: A photovoltaic panel environmental perception subsystem, which is used to obtain environmental data, extract the position distribution information and hardness information of stains from the environmental data, and judge the type information of the stains; A photovoltaic panel fusion decision-making subsystem, which is connected to the photovoltaic panel environmental perception subsystem, and is used to automatically formulate a cleaning strategy by using the position distribution information, hardness information and type information of the stains, and generate a cleaning instruction; A cleaning robot execution system, which is connected to the photovoltaic panel fusion decision-making subsystem, and is used to receive and execute the cleaning instruction to clean the stains of the photovoltaic power station; A feedback subsystem, which is connected to the cleaning robot execution system, and is used to obtain the cleaning result and feedback the cleaning result to the photovoltaic panel fusion decision-making subsystem. The photovoltaic panel fusion decision-making subsystem optimizes the cleaning strategy based on the cleaning result; A host computer, which is connected to the photovoltaic panel environmental perception subsystem, the photovoltaic panel fusion decision-making subsystem and the cleaning robot execution system, and is used to receive the status data of the photovoltaic panel environmental perception subsystem, the photovoltaic panel fusion decision-making subsystem and the cleaning robot execution system in real time, and remotely monitor and manage the photovoltaic panel environmental perception subsystem, the photovoltaic panel fusion decision-making subsystem and the cleaning robot execution system according to the status data of the photovoltaic panel environmental perception subsystem, the photovoltaic panel fusion decision-making subsystem and the cleaning robot execution system.

2. The AI automatic cleaning system for a photovoltaic power station according to claim 1, wherein The photovoltaic panel environmental perception subsystem includes: An image acquisition subsystem, which is used to acquire image information on the surface of the photovoltaic panel; An ultrasonic detection subsystem, which is used to detect the ultrasonic feedback information of the stains around the photovoltaic panel and calculate the hardness data of the corresponding stains; A stain detection subsystem, which is connected to the image acquisition subsystem, and is used to identify the stains in the image information, obtain the position information of the stains, and count the position distribution density of the stains around the photovoltaic panel; A stain type judgment subsystem, which is connected to the stain detection subsystem, and is used to obtain the cleaning mode of the stains.

3. The AI automated cleaning system for a photovoltaic power station according to claim 2, wherein, The stain type judgment subsystem identifies the cleaning mode of the stains through a stain recognition judgment model; The stain recognition judgment model includes an input layer, a hidden layer and an output layer; The expression of the input layer is: x = [x1, x2,..., x n ​ where x represents the combined feature vector of the input stain type and stain hardness, and n represents the number of combined feature vectors of the input stain type and stain hardness; The expression of the hidden layer is: Among them, z j represents the linear combination result of the j-th neuron in the hidden layer matching the corresponding cleaning mode for stain type and stain hardness, represents the weight from the input layer to the j-th neuron in the hidden layer, x i represents the combined feature vector of the stain type and stain hardness of the i-th input, represents the bias of the j-th neuron, m represents the number of neurons in the hidden layer, a j represents the activation output of the j-th neuron in the hidden layer, σ() represents the Sigmoid activation function, exp() represents the exponential function, and α represents the parameter for adjusting the shape of the Sigmoid activation function; The expression of the output layer is: Among them, z′ k represents the linear combination result of the k-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness, represents the weight from the hidden layer to the k-th neuron in the output layer, represents the bias of the k-th neuron in the output layer, p k represents the predicted probability that the stain type or stain hardness is of the k-th class, exp() represents the activation function of the output layer, β represents the parameter for adjusting the shape of the output layer activation function, z l ′ represents the linear combination result of the l-th neuron in the output layer matching the corresponding cleaning mode for stain type and stain hardness, Ξ represents the stain type or stain hardness, represents taking p k maximum value.

4. The AI automated cleaning system for a photovoltaic power station according to claim 1, wherein, The photovoltaic panel fusion decision-making subsystem includes: A data preprocessing subsystem, which is used to preprocess the type information of the stains, the stain hardness data and the position distribution density of the stains around the photovoltaic panel; A strategy formulation subsystem, which is connected to the data preprocessing subsystem, and is used to formulate the cleaning strategy based on the preprocessed type information of the stains, the stain hardness data and the position distribution density of the stains around the photovoltaic panel. The cleaning strategy includes a distributed cleaning path from high density to low density formulated according to the position distribution density of the stains, and the cleaning intensity and cleaning time formulated according to the stain type and hardness data. Optimization subsystem, connected to the policy formulation subsystem, for optimizing the cleaning policy based on the cleaning result sent by the feedback subsystem, generating an optimized cleaning instruction, and sending the optimized cleaning instruction to the cleaning robot execution system; Resource scheduling subsystem, connected to the policy formulation subsystem, for scheduling the cleaning robot execution system based on the cleaning policy.

5. The AI automated cleaning system for a photovoltaic power station according to claim 4, wherein, The policy formulation subsystem formulates the cleaning policy through a genetic algorithm.

6. The AI automated cleaning system for a photovoltaic power station according to claim 4, wherein, The optimization subsystem optimizes the cleaning policy through an optimization algorithm; The expression of the optimization algorithm is: Among them, represents the velocity vector after the cleaning robot execution system is updated, ω represents the inertia weight, and v id represents the velocity vector before the cleaning robot execution system is updated, c1 and c2 represent the learning factors, r1 and r2 represent random numbers uniformly distributed in the range [0, 1], and pbest id represents the component of the individual optimal position of the i-th particle in the d-th dimension, and x id represents the position of the i-th particle in the d-th dimension, and gbest d represents the component of the global optimal position of the entire particle swarm in the d-th dimension, represents the new position of the i-th particle in the d-th dimension.

7. The AI automated cleaning system for a photovoltaic power station according to claim 1, characterized in that, The cleaning robot execution system includes: Instruction receiving subsystem, for receiving the cleaning instruction; Motion control subsystem, connected to the instruction receiving subsystem, for controlling the motion trajectory of the cleaning subsystem according to the cleaning instruction; Cleaning subsystem, connected to the motion control subsystem, for executing the control instruction of the motion control subsystem, and the control instruction includes spraying water or / and brushing or / and blowing.

8. AI automatic cleaning method for photovoltaic power station, characterized in that, The method includes the photovoltaic power station AI automatic cleaning system according to any one of claims 1-7, and the method further includes: S1. Obtain the environmental data of the photovoltaic power station, extract the position distribution information and hardness information of the stains from the environmental data, and judge the type information of the stains; S2. Automatically formulate a cleaning policy based on the position distribution information, hardness information, and type information of the stains, and generate a cleaning instruction; S3. Receive and execute the cleaning instruction to clean the stains of the photovoltaic power station; S4. Obtain the cleaning result of the photovoltaic power station in real time, and optimize the cleaning policy based on the cleaning result; S5. Obtain the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system in real time, and remotely monitor and manage the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system according to the status data of the photovoltaic panel environment perception subsystem, the photovoltaic panel fusion decision subsystem, and the cleaning robot execution system.

9. A cleaning device, characterized in that, The cleaning device includes: Processor; Memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, the method according to claim 8 is implemented.

10. A computer-readable storage medium, characterized in that, Program code is stored in the computer-readable storage medium, and the program code can be called by the processor to execute the method according to claim 8.

Images