Intelligent management system for periodic cleaning based on data analysis
Through the intelligent cycle cleaning management system based on data analysis, the problem of unreasonable dust accumulation and cleaning cycle of photovoltaic panels in the operation and maintenance management of photovoltaic power stations is solved. Through data compensation and feature analysis, a prediction model is built to determine the optimal cleaning cycle, which improves the The operation and maintenance and power generation efficiency of photovoltaic power stations have optimized the cleaning strategy.
Patent Information
- Application Number
- CN202411155375.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-22
AI Technical Summary
In the operation and maintenance management of photovoltaic power stations, the ash accumulation, pollution and cleaning cycle of photovoltaic panels are unreasonable, resulting in reduced power generation efficiency and increased operation and maintenance costs. The time series analysis is complex, making it difficult to flexibly adjust the cleaning strategy.
Intelligent management system for periodic cleaning based on data analysis, including multi-dimensional data adaptation adjustment module, feature analysis and extraction module, periodic cleaning prediction module and prediction evaluation optimization module. By compensating and processing the historical data of photovoltaic power stations, a multi-dimensional feature space is built and a multi-dimensional feature space is established. The predictive model determines the optimal cleaning cycle and evaluates and analyzes it with real-time monitoring data to optimize the cleaning strategy.
It significantly improves the operation and maintenance efficiency and power generation efficiency of photovoltaic power stations, ensures that the photovoltaic panels are in the best working state for a long time, optimizes the cleaning strategy, reduces the operation and maintenance costs, and improves the accuracy and generalization capabilities of the prediction model.
Smart Images

Figure CN118691262B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data analysis, and in particular to an intelligent management system for periodic cleaning based on data analysis. Background Art
[0002] Data analysis technology refers to the technology that uses knowledge of statistics, computer science and related fields to process, analyze and interpret collected data in order to extract valuable information and provide support for decision-making.
[0003] With the continuous advancement of technology and the continuous expansion of application fields, the photovoltaic power generation industry is also developing steadily. However, the following problems still exist in the operation and maintenance management of photovoltaic power stations: dust accumulation and pollution of photovoltaic panels. After long-term operation, pollutants such as dust, bird droppings, and leaves will accumulate on the surface of photovoltaic panels. Pollutants will block light and reduce the transmittance and photoelectric conversion efficiency of photovoltaic panels; the cleaning cycle is unreasonable. Traditional manual cleaning or regular cleaning methods often cannot flexibly adjust the cleaning cycle according to the actual pollution of photovoltaic panels, resulting in too frequent cleaning or untimely cleaning, increasing operation and maintenance costs or reducing power generation efficiency; time series analysis complexity. Data such as the power generation of photovoltaic power stations have obvious time series characteristics and are affected by many factors, such as seasonal changes, weather conditions, etc., which increases the complexity of time series analysis. To this end, the present invention proposes an intelligent management system for periodic cleaning based on data analysis. Summary of the invention
[0004] The purpose of the present invention is to solve the problems in the background technology and to propose an intelligent management system for periodic cleaning based on data analysis.
[0005] In order to achieve the above object, the present invention adopts the following technical solutions:
[0006] The intelligent management system for periodic cleaning based on data analysis includes: multivariate data adaptation and adjustment module, feature analysis and extraction module, periodic cleaning prediction module and prediction evaluation and optimization module;
[0007] The multivariate data adaptation and adjustment module is used to compensate and process the historical data of the photovoltaic power station; the multivariate data adaptation and adjustment module includes a first composite compensation unit, a second composite compensation unit and a cleaning frequency processing unit;
[0008] The feature analysis and extraction module is used to extract features according to the compensation processing data and construct a multi-dimensional feature space; the feature analysis and extraction module includes a first composite feature extraction unit, a second composite feature extraction unit and a cleaning activity feature extraction unit;
[0009] The cycle cleaning prediction module is used to perform time series analysis based on the constructed multi-dimensional feature space and establish a prediction model to determine the optimal cleaning cycle;
[0010] The prediction evaluation and optimization module is used to evaluate and analyze the optimal cleaning cycle of the prediction model and the real-time monitoring data, and to determine the accuracy of the model prediction.
[0011] It should be noted that the application object of the intelligent management system for periodic cleaning based on data analysis in the embodiment of the present invention can be the operation and maintenance management of photovoltaic power stations in the field of clean energy industry, and can be used to monitor the operating status of photovoltaic panels in photovoltaic power stations. Specifically, it can be to use data analysis technology to comprehensively analyze and accurately evaluate the historical power generation data of photovoltaic power stations, environmental factor data (such as temperature, humidity, light intensity, etc.) and cleaning record data to predict and determine the optimal cleaning cycle of photovoltaic panels. Through data processing, feature extraction, model training, prediction and evaluation and other links, a complete intelligent management system is formed, which enables the system to significantly improve the operation and maintenance efficiency and power generation efficiency of photovoltaic power stations, ensure that photovoltaic panels are in the best working state for a long time, so as to meet the needs of efficient utilization of clean energy and sustainable development.
[0012] Furthermore, the process of the first composite compensation unit performing compensation analysis on the acquired historical temperature data and historical humidity data includes:
[0013] The historical temperature data and historical humidity data of the photovoltaic power station are marked as and ;
[0014] Extract the values of historical temperature and humidity data and perform compound compensation analysis:
[0015] , where Indicates the output power after historical temperature and humidity composite compensation, reflecting the impact of temperature and humidity on the output power of photovoltaic panels; Represents the original output power of the photovoltaic panel under the historical temperature and humidity observation conditions; represent the sensitivity coefficients of temperature and humidity respectively; Respectively represent the reference temperature value and the reference humidity value;
[0016] It is understandable that changes in temperature and humidity can affect the performance of photovoltaic panels. Reflects the actual output power level of the photovoltaic panel under the current temperature and humidity conditions before any compensation adjustment is made; Perform compound compensation to obtain the compensation adjusted , in order to accurately reflect the effect of temperature and humidity on the output power of photovoltaic panels.
[0017] Furthermore, the process of the second composite compensation unit for performing compensation analysis on the acquired historical light intensity and historical angle data includes:
[0018] The historical light intensity data and historical angle data of the photovoltaic power station are marked as and ;
[0019] Extract the values of historical light intensity and historical angle data, and perform composite compensation analysis:
[0020] , where It represents the output power after the historical light intensity and historical angle are compositely compensated; represents the raw output power of the photovoltaic panel under historical light intensity and historical angle conditions; Indicates the light intensity reference value; represents the optimal angle; Represent the sensitivity index of light intensity and angle respectively;
[0021] It is understandable that due to the light intensity and angle The change of has a direct impact on the amount of solar radiation received by the photovoltaic panel, which in turn affects its output power. Perform compound compensation based on light intensity and angle to obtain the compensated power , accurately reflects the performance of photovoltaic panels under specific light intensity and angle conditions;
[0022] Among them, the original output power is adjusted through a composite compensation formula to more accurately reflect the performance of photovoltaic panels under actual working conditions and more accurately predict the performance changes of photovoltaic panels. This compensation process is of great significance for optimizing the operation and maintenance of photovoltaic power stations, especially in the formulation of periodic cleaning strategies, so as to formulate more effective cleaning plans and ensure the efficient operation of photovoltaic power stations.
[0023] Furthermore, the process of the cleaning frequency processing unit processing and analyzing the acquired historical cleaning frequency includes:
[0024] Set Existence Historical cleaning records, each cleaning record contains the cleaning time point , Average output power of photovoltaic panels before and after cleaning And cleaning costs ;
[0025] Use the cleaning frequency processing formula for analysis:
[0026] , , where It represents the cleaning optimization power index, which comprehensively reflects the cleaning frequency, cleaning effect, cost and long-term impact, and provides a comprehensive evaluation for the subsequent extraction of cleaning activity feature values; Indicates The average output power of photovoltaic panels increased after cleaning. It indicates the ratio of the output power increase after each cleaning to the cleaning cost, that is, the increase in output power after cleaning per unit cost; is the attenuation factor, which is used to indicate the degree of attenuation of the cleaning effect over time, quantify the attenuation of the cleaning effect over time, and thus evaluate the long-term impact of cleaning activities on the performance of photovoltaic panels; represents the cleaning effect decay rate, and ; Indicates the time interval between two cleanings; The average value representing the long-term decay of the cleaning effect is used to evaluate the long-term impact of cleaning activities on the performance of photovoltaic panels; Indicates the cleaning time point identifier or index;
[0027] It is understandable that through the cleaning frequency processing formula, the cleaning frequency processing unit can analyze the historical cleaning records more deeply, so as to subsequently extract the comprehensive characteristic values of the cleaning activities, and provide data support for optimizing the cleaning cycle and task scheduling of the photovoltaic power station. This analysis method not only considers the direct effect and cost of cleaning, but also considers the long-term impact of cleaning on the performance of photovoltaic panels, which helps to achieve more efficient and economical photovoltaic power station management.
[0028] Furthermore, the process of the first composite feature extraction unit for extracting features according to the compensation analysis result of the first composite compensation unit includes:
[0029] Output power after temperature and humidity compound compensation , get some output power The data set is taken and the first composite feature extraction is performed:
[0030] , where represents the first composite eigenvalue; represents the number of observations; Represents the observation index subscript; Indicates Set historical temperature data; Indicates Set historical humidity data; Indicates The output power of the group's historical observations.
[0031] Furthermore, the process in which the second composite feature extraction unit is used to extract features according to the compensation analysis result of the second composite compensation unit includes:
[0032] Collect the output power of photovoltaic panels after compound compensation under different light intensities and angles and the corresponding timestamp ;
[0033] Assuming the output power after composite compensation Affected by the periodic changes in light intensity and angle, a periodic function is used to capture the periodic changes in light intensity and angle on the output power of the photovoltaic panel:
[0034] , , , , where express is approximated as a Fourier series; represents the order of the Fourier series; Indicates the order index of the harmonic component, used to traverse from 1 to All harmonic components of Indicates the cycle length; are Fourier coefficients, which respectively represent the average value of the output power, the even harmonic components of the output power varying with time, and the odd harmonic components of the output power varying with time, reflecting the periodic characteristics of the output power varying with time; Indicates A small increment of the variable, for timestamps Integrate the differential of ;
[0035] Define the second composite eigenvalue is the Fourier coefficient vector:
[0036] .
[0037] Furthermore, the process of the cleaning activity feature extraction unit for extracting features according to the processing and analysis results of the cleaning frequency processing unit includes:
[0038] Optimize power index by combining cleaning Perform computational analysis:
[0039] , where Indicates the characteristic value of cleaning activity; Indicates the instant power recovery rate, that is, the rapid recovery of the photovoltaic panel power after cleaning, which is used to reflect the immediate effect of the cleaning activity; It represents the long-term performance stability coefficient, which is used to analyze the long-term stability of the performance of the photovoltaic panel after cleaning, that is, how long the cleaning effect can last without being significantly affected;
[0040] Integrate the first composite eigenvalue, the second composite eigenvalue and the cleaning activity eigenvalue to form a multidimensional eigenvector set , and constitute a multidimensional feature space; wherein each feature vector is a point in the multidimensional feature space; the multidimensional feature space is composed of multiple feature vectors, each feature vector contains the integrated result of the first composite feature value, the second composite feature value and the cleaning activity feature value; at the same time, the multidimensional feature vector set jointly defines a multidimensional data space, which provides a comprehensive and rich feature representation for subsequent data analysis and learning model training.
[0041] Furthermore, the cycle cleaning prediction module performs time series analysis based on the constructed multi-dimensional feature space, and the process of establishing a prediction model to determine the optimal cleaning cycle includes:
[0042] Multiple training sample sets are extracted from the original data set, and each sample set is used to build a decision tree; wherein, all multidimensional feature vector sets, i.e., points in the multidimensional feature space, are extracted from a single sample of the original data set; when building each decision tree, a random selection is made. Some features are used as candidate feature sets, and the tree is grown based on the candidate features; among them, Indicates the number of randomly selected features as the size of the candidate feature set. For example, if there are 100 features in the original data set, the user only randomly selects 10 or 20 features each time to grow the tree;
[0043] Each decision tree makes a prediction based on the input feature vector, i.e. a point in the multidimensional feature space, and outputs a prediction result;
[0044] The process of using random forest for prediction is:
[0045] , where Represents random forest in time series The final prediction value for predicting the optimal cleaning cycle; A function representing the random forest prediction process, receiving a multidimensional feature vector and the set of all decision trees in the random forest , and as input, output the predicted value of the optimal cleaning cycle ; Indicates that in time series The multi-dimensional feature vector includes a first composite feature value, a second composite feature value, and a cleaning activity feature value; represents the total number of trees in the random forest;
[0046] Among them, since random forest works by integrating the prediction results of multiple decision trees, the function is further expanded to describe its working mechanism more specifically:
[0047] , where Represents the first decision tree; Indicates A decision tree is constructed based on the feature vector For prediction, the prediction value of the entire random forest is the average of all tree prediction values; ;
[0048] It can be understood that the random forest model uses samples in the original data set to train multiple decision trees and makes a final prediction by integrating the prediction results of all decision trees to determine the optimal cleaning cycle; therefore, the final prediction result of the random forest is based on the integration of the prediction results of all decision trees.
[0049] Furthermore, the prediction evaluation optimization module evaluates and analyzes the optimal cleaning cycle of the prediction model and the real-time monitoring data, and the process of judging the accuracy of the model prediction includes:
[0050] Define the real-time monitoring data set as:
[0051] ;
[0052] in, Represents the photovoltaic panel in the time series No. Monitoring data, Indicates the identification or index number of the monitoring data. represents the amount of monitoring data, and ; Real-time monitoring data is the degree of dirt accumulation; the average thickness of the dirt layer, the fluid inlet and outlet temperature difference, the dirt thermal resistance and the change in the weight of the deposition plate in the direct weighing method are used to characterize the degree of dirt accumulation; the dirt thermal resistance is an important parameter to measure the effect of dirt deposition on heat exchange efficiency, and is indirectly measured by comparing the total heat transfer coefficient in the clean state and the contaminated state; the average thickness of the dirt layer refers to the average thickness of the layer formed by dirt on the surface of the equipment; the fluid inlet and outlet temperature difference refers to the temperature difference before and after the fluid passes through the equipment, and the change in the fluid inlet and outlet temperature difference is monitored to reflect the change in the amount of dirt deposition; the change in the weight of the deposition plate in the direct weighing method refers to setting an easily disassembled deposition plate on the equipment, regularly removing it for drying and weighing it, and determining the amount of scale deposition by the change in the weight of the deposition plate;
[0053] The optimal cleaning cycle predicted by the random forest prediction model is compared and analyzed with the actual real-time monitoring data to evaluate whether the change in the degree of dirt accumulation at the current moment is in line with expectations;
[0054] The predicted optimal cleaning cycle is defined as , and the real-time monitoring of the degree of dirt accumulation is ,but Based on historical data, the The degree of dirt accumulation at a certain moment can be evaluated and analyzed as follows:
[0055] , where represents the acceptable range of prediction error, that is, if the difference between the predicted optimal cleaning cycle corresponding to the degree of dirt accumulation (estimated or known) and the real-time monitored degree of dirt accumulation is within If the prediction is within the range, the prediction is considered accurate;
[0056] If the predicted optimal cleaning cycle corresponds to the degree of dirt accumulation With real-time monitoring of the degree of dirt accumulation The difference is at the threshold If the predicted optimal cleaning cycle at the current moment is within the range, the predicted optimal cleaning cycle is consistent with the dirt accumulation degree monitored in real time, and the predicted result is considered to be consistent with the actual situation. A cleaning condition including a cleaning start signal is generated, and the cleaning start signal is sent to the cleaning robot to allocate and start the cleaning task.
[0057] The cleaning start signal includes the cleaning time, cleaning area and cleaning intensity. The cleaning conditions are:
[0058] , where Indicates the cleaning task priority. A higher value indicates a higher priority; Indicates the set constant item to ensure the priority of cleaning tasks range, i.e. ; Indicates resource allocation; Represents the resource allocation coefficient, which is used to adjust the resource allocation ratio to ensure range, i.e. ;
[0059] If the difference between the predicted dirt accumulation degree corresponding to the optimal cleaning cycle and the real-time monitored dirt accumulation degree exceeds the threshold range, the predicted optimal cleaning cycle at the current moment does not conform to the real-time monitored dirt accumulation degree, and it is considered that there is a problem with the prediction result, and the model adjustment parameters are re-optimized;
[0060] Among them, re-optimize the model adjustment parameters for:
[0061] , where Represents the learning rate, which is used to control the step size of parameter adjustment; Represents a time series, that is, random forest in the time series Make predictions on represents the gradient with respect to the prediction model parameters; Function representing the random forest prediction process; Indicates that in time series The multidimensional feature vector of .
[0062] Compared with the existing technology, the advantages of the intelligent management system and method for periodic cleaning based on data analysis provided by the present invention are:
[0063] The present invention compensates and processes the historical data of photovoltaic power plants, adjusts the deviation of power generation data caused by extreme weather or seasonal changes, and more accurately evaluates the impact of lighting conditions on power generation. Based on the effectiveness of the evaluation of cleaning activities, it can provide a valuable reference for optimizing the cleaning strategy.
[0064] The present invention extracts features based on the compensation processing data and constructs a multidimensional feature space, thereby revealing the nonlinear relationship between environmental factors and power generation, analyzing the impact of cleaning activities on the long-term performance of photovoltaic panels, and improving the training efficiency and prediction performance of the model;
[0065] The present invention establishes a prediction model to determine the optimal cleaning cycle by performing time series analysis based on the constructed multidimensional feature space, thereby improving the accuracy and generalization ability of the prediction model and making the prediction result more reliable and practical; by evaluating and analyzing the optimal cleaning cycle of the prediction model with real-time monitoring data and judging the accuracy of the model prediction, the continuous evaluation and optimization process ensures that the prediction model always maintains a high accuracy and stability.
[0066] To sum up, the present invention collects historical data of photovoltaic power stations to predict cleaning cycles, and conducts comprehensive data analysis and precise monitoring based on the actual situation of photovoltaic power stations. The various modules in the invention cooperate and work together to form an efficient and intelligent periodic cleaning management and operation system, which can clean photovoltaic panels in time, ensure their cleanliness and power generation efficiency, further improve economic benefits, and ensure the efficient and stable operation of the subsequent periodic cleaning intelligent management system based on data analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0067] Figure 1 This is a module diagram of the intelligent management system for periodic cleaning based on data analysis proposed by the present invention. DETAILED DESCRIPTION
[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0069] Reference Figure 1 , an intelligent management system for periodic cleaning based on data analysis, the system includes a multivariate data adaptation and adjustment module, a feature analysis and extraction module, a periodic cleaning prediction module, and a prediction evaluation and optimization module;
[0070] The multivariate data adaptation and adjustment module is used to compensate and process the historical data of the photovoltaic power station; the multivariate data adaptation and adjustment module includes a first composite compensation unit, a second composite compensation unit and a cleaning frequency processing unit; wherein the first composite compensation unit is used to perform compensation analysis on the acquired historical temperature data and historical humidity data, the second composite compensation unit is used to perform compensation analysis on the acquired historical light intensity and historical angle data, and the cleaning frequency processing unit is used to process and analyze the acquired historical cleaning frequency;
[0071] The feature analysis and extraction module is used to extract features according to the compensation processing data and construct a multi-dimensional feature space; the feature analysis and extraction module includes a first composite feature extraction unit, a second composite feature extraction unit and a cleaning activity feature extraction unit; wherein the first composite feature extraction unit is used to extract features according to the compensation analysis result of the first composite compensation unit, the second composite feature extraction unit is used to extract features according to the compensation analysis result of the second composite compensation unit, and the cleaning activity feature extraction unit is used to extract features according to the processing analysis result of the cleaning frequency processing unit;
[0072] The cycle cleaning prediction module is used to perform time series analysis based on the constructed multi-dimensional feature space and establish a prediction model to determine the optimal cleaning cycle;
[0073] The prediction evaluation and optimization module is used to evaluate and analyze the optimal cleaning cycle of the prediction model and the real-time monitoring data, and to determine the accuracy of the model prediction.
[0074] It should be noted that the application object of the intelligent management system for periodic cleaning based on data analysis in the embodiment of the present invention can be the operation and maintenance management of photovoltaic power stations in the field of clean energy industry, and can be used to monitor the operating status of photovoltaic panels in photovoltaic power stations. Specifically, it can be to use data analysis technology to comprehensively analyze and accurately evaluate the historical power generation data of photovoltaic power stations, environmental factor data (such as temperature, humidity, light intensity, etc.) and cleaning record data to predict and determine the optimal cleaning cycle of photovoltaic panels. Through data processing, feature extraction, model training, prediction and evaluation and other links, a complete intelligent management system is formed, which enables the system to significantly improve the operation and maintenance efficiency and power generation efficiency of photovoltaic power stations, ensure that photovoltaic panels are in the best working state for a long time, so as to meet the needs of efficient utilization of clean energy and sustainable development.
[0075] The steps of the multivariate data adaptation adjustment module for compensating and processing the historical data of the photovoltaic power station include:
[0076] Step 101: Mark the historical temperature data and historical humidity data of the photovoltaic power station as and ;
[0077] Step 102: extract the values of historical temperature and humidity data and perform composite compensation analysis:
[0078] , where Indicates the output power after historical temperature and humidity composite compensation, reflecting the impact of temperature and humidity on the output power of photovoltaic panels; Represents the original output power of the photovoltaic panel under the historical temperature and humidity observation conditions; represent the sensitivity coefficients of temperature and humidity respectively; Respectively represent the reference temperature value and the reference humidity value;
[0079] In step 102, since changes in temperature and humidity can affect the performance of the photovoltaic panels, Reflects the actual output power level of the photovoltaic panel under the current temperature and humidity conditions before any compensation adjustment is made; Perform compound compensation to obtain the compensation adjusted , in order to accurately reflect the effect of temperature and humidity on the output power of photovoltaic panels;
[0080] Step 103: Mark the historical light intensity data and historical angle data of the photovoltaic power station as and ;
[0081] Step 104: extract the values of historical light intensity and historical angle data, and perform composite compensation analysis:
[0082] , where It represents the output power after the historical light intensity and historical angle are compositely compensated; represents the raw output power of the photovoltaic panel under historical light intensity and historical angle conditions; Indicates the light intensity reference value; represents the optimal angle; Represent the sensitivity index of light intensity and angle respectively;
[0083] In step 104, due to the light intensity and angle The change of has a direct impact on the amount of solar radiation received by the photovoltaic panel, which in turn affects its output power. Perform compound compensation based on light intensity and angle to obtain the compensated power , accurately reflects the performance of photovoltaic panels under specific light intensity and angle conditions;
[0084] The original output power is adjusted through a composite compensation formula to more accurately reflect the performance of the photovoltaic panels under actual working conditions and more accurately predict the performance changes of the photovoltaic panels. This compensation process is of great significance for optimizing the operation and maintenance of photovoltaic power plants, especially in the formulation of periodic cleaning strategies, so as to formulate more effective cleaning plans and ensure the efficient operation of photovoltaic power plants.
[0085] Step 105: Set existence Historical cleaning records, each cleaning record contains the cleaning time point , Average output power of photovoltaic panels before and after cleaning And cleaning costs ;
[0086] Step 106: Use the cleaning frequency processing formula for analysis:
[0087] , , where It represents the cleaning optimization power index, which comprehensively reflects the cleaning frequency, cleaning effect, cost and long-term impact, and provides a comprehensive evaluation for the subsequent extraction of cleaning activity feature values; Indicates The average output power of photovoltaic panels increased after cleaning. It indicates the ratio of the output power increase after each cleaning to the cleaning cost, that is, the increase in output power after cleaning per unit cost; is the attenuation factor, which is used to indicate the degree of attenuation of the cleaning effect over time, quantify the attenuation of the cleaning effect over time, and thus evaluate the long-term impact of cleaning activities on the performance of photovoltaic panels; represents the cleaning effect decay rate, and ; Indicates the time interval between two cleanings; The average value representing the long-term decay of the cleaning effect is used to evaluate the long-term impact of cleaning activities on the performance of photovoltaic panels; Indicates the cleaning time point identifier or index;
[0088] In step 106, through the cleaning frequency processing formula, the cleaning frequency processing unit can analyze the historical cleaning records more deeply, so as to subsequently extract the comprehensive characteristic values of the cleaning activities, and provide data support for optimizing the cleaning cycle and task scheduling of the photovoltaic power station. This analysis method not only considers the direct effect and cost of cleaning, but also considers the long-term impact of cleaning on the performance of photovoltaic panels, which helps to achieve more efficient and economical photovoltaic power station management.
[0089] The feature analysis and extraction module extracts features based on the compensated processing data, and the steps of constructing a multi-dimensional feature space include:
[0090] Step 201: Output power after temperature and humidity composite compensation , get some output power The data set is taken and the first composite feature extraction is performed:
[0091] , where represents the first composite eigenvalue; represents the number of observations; Represents the observation index subscript; Indicates Set historical temperature data; Indicates Set historical humidity data; Indicates Output power of the group's historical observations;
[0092] Step 202: Collect the output power of the photovoltaic panel after compound compensation under different light intensities and angles and the corresponding timestamp ;
[0093] Step 203: Assume that the output power after composite compensation is Affected by the periodic changes in light intensity and angle, a periodic function is used to capture the periodic changes in light intensity and angle on the output power of the photovoltaic panel:
[0094] , , , , where express is approximated as a Fourier series; represents the order of the Fourier series; Indicates the order index of the harmonic component, used to traverse from 1 to All harmonic components of Indicates the cycle length; are Fourier coefficients, which respectively represent the average value of the output power, the even harmonic components of the output power varying with time, and the odd harmonic components of the output power varying with time, reflecting the periodic characteristics of the output power varying with time; Indicates A small increment of the variable, for timestamps Integrate the differential of ;
[0095] Step 204: Define the second composite eigenvalue is the Fourier coefficient vector:
[0096] ;
[0097] Step 205: Optimizing power index in combination with cleaning Perform computational analysis:
[0098] , where Indicates the characteristic value of cleaning activity; Indicates the instant power recovery rate, that is, the rapid recovery of the photovoltaic panel power after cleaning, which is used to reflect the immediate effect of the cleaning activity; It represents the long-term performance stability coefficient, which is used to analyze the long-term stability of the performance of the photovoltaic panel after cleaning, that is, how long the cleaning effect can last without being significantly affected;
[0099] In step 205, the instantaneous power recovery rate measures the rate of power recovery after cleaning, and the long-term performance stability coefficient measures the degree to which the power generation efficiency and overall performance of the photovoltaic panel remain stable over a long period of time after the cleaning process.
[0100] , where They represent the power after cleaning is completed and stabilized and the power before cleaning respectively; Indicates the time required from the start of cleaning to the power recovery to a stable state; Indicates the real-time power of the photovoltaic panel within a period of time g after cleaning; Indicates the time point after cleaning is completed and stabilized; Indicates is a microvariable of the variable, Integrate the differential of ;
[0101] Step 206: Integrate the first composite eigenvalue, the second composite eigenvalue, and the cleaning activity eigenvalue to form a multidimensional eigenvector set. , and constitute a multidimensional feature space; wherein each feature vector is a point in the multidimensional feature space; the multidimensional feature space is composed of multiple feature vectors, each feature vector contains the integrated result of the first composite feature value, the second composite feature value and the cleaning activity feature value; at the same time, the multidimensional feature vector set jointly defines a multidimensional data space, which provides a comprehensive and rich feature representation for subsequent data analysis and learning model training.
[0102] The cycle cleaning prediction module performs time series analysis based on the constructed multi-dimensional feature space, and the steps of establishing a prediction model to determine the optimal cleaning cycle include:
[0103] Step 301: extract multiple training sample sets from the original data set, and each sample set is used to construct a decision tree;
[0104] In step 301, all multidimensional feature vector sets, i.e., points in the multidimensional feature space, are extracted from a single sample of the original data set;
[0105] Step 302: When constructing each decision tree, randomly select The partial features are used as candidate feature sets, and the tree is grown based on the candidate features;
[0106] In step 302, Indicates the number of randomly selected features as the size of the candidate feature set. For example, if there are 100 features in the original data set, the user only randomly selects 10 or 20 features each time to grow the tree;
[0107] Step 303, each decision tree makes a prediction based on the input feature vector, i.e., a point in the multidimensional feature space, and outputs a prediction result;
[0108] Step 304: The process of using random forest for prediction is:
[0109] , where Represents random forest in time series The final prediction value for predicting the optimal cleaning cycle; A function representing the random forest prediction process, receiving a multidimensional feature vector and the set of all decision trees in the random forest , and as input, output the predicted value of the optimal cleaning cycle ; Indicates that in time series The multi-dimensional feature vector includes a first composite feature value, a second composite feature value, and a cleaning activity feature value; represents the total number of trees in the random forest;
[0110] In step 304, since the random forest works by integrating the prediction results of multiple decision trees, the function is further expanded to describe its working mechanism in more detail:
[0111] , where Represents the first decision tree; Indicates A decision tree is constructed based on the feature vector For prediction, the prediction value of the entire random forest is the average of all tree prediction values; ;
[0112] The random forest model uses samples from the original data set to train multiple decision trees, and makes a final prediction by integrating the prediction results of all decision trees to determine the optimal cleaning cycle; therefore, the final prediction result of the random forest is based on the integration of the prediction results of all decision trees. The prediction evaluation and optimization module evaluates and analyzes the optimal cleaning cycle of the prediction model with real-time monitoring data, and the steps to determine the accuracy of the model prediction include:
[0113] Step 401: define the real-time monitoring data set as:
[0114] ;
[0115] in, Represents the photovoltaic panel in the time series No. Monitoring data, Indicates the identification or index number of the monitoring data. represents the amount of monitoring data, and ;
[0116] In step 401, the real-time monitoring data is the degree of dirt accumulation; the degree of dirt accumulation is characterized by parameters such as the average thickness of the dirt layer, the fluid inlet and outlet temperature difference, the dirt thermal resistance, and the weight change of the deposition plate in the direct weighing method; the dirt thermal resistance is an important parameter for measuring the effect of dirt deposition on heat exchange efficiency, and is indirectly measured by comparing the total heat transfer coefficient in the clean state and the contaminated state; the average thickness of the dirt layer refers to the average thickness of the layer formed by the dirt on the surface of the equipment; the fluid inlet and outlet temperature difference refers to the temperature difference before and after the fluid passes through the equipment, and the change in the fluid inlet and outlet temperature difference is monitored to reflect the change in the amount of dirt deposition; the weight change of the deposition plate in the direct weighing method refers to setting an easily disassembled deposition plate on the equipment, regularly removing it for drying and weighing it, and determining the amount of dirt deposition by the change in the weight of the deposition plate;
[0117] Step 402: Compare and analyze the optimal cleaning cycle predicted by the random forest prediction model with the actual real-time monitoring data to evaluate whether the change in the degree of dirt accumulation at the current moment is in line with expectations;
[0118] Step 403: define the predicted optimal cleaning cycle as , and the real-time monitoring of the degree of dirt accumulation is ,but Based on historical data, the The degree of dirt accumulation at a certain moment can be evaluated and analyzed as follows:
[0119] , where represents the acceptable range of prediction error, that is, if the difference between the predicted optimal cleaning cycle corresponding to the degree of dirt accumulation (estimated or known) and the real-time monitored degree of dirt accumulation is within If the prediction is within the range, the prediction is considered accurate;
[0120] Step 404: If the predicted optimal cleaning cycle corresponds to the degree of dirt accumulation With real-time monitoring of the degree of dirt accumulation The difference is at the threshold If the predicted optimal cleaning cycle at the current moment is within the range, the predicted optimal cleaning cycle is consistent with the dirt accumulation degree monitored in real time, and the predicted result is considered to be consistent with the actual situation. A cleaning condition including a cleaning start signal is generated, and the cleaning start signal is sent to the cleaning robot to allocate and start the cleaning task.
[0121] In step 404, the cleaning start signal includes the cleaning time, cleaning area and cleaning intensity, and the cleaning conditions are:
[0122] , where Indicates the cleaning task priority. A higher value indicates a higher priority; Indicates the set constant item to ensure the priority of cleaning tasks range, i.e. ; Indicates resource allocation; Represents the resource allocation coefficient, which is used to adjust the resource allocation ratio to ensure range, i.e. ;
[0123] Step 405: If the difference between the predicted dirt accumulation degree corresponding to the optimal cleaning cycle and the real-time monitored dirt accumulation degree exceeds the threshold range, the predicted optimal cleaning cycle at the current moment does not conform to the real-time monitored dirt accumulation degree, and it is considered that there is a problem with the prediction result, and the model adjustment parameters are re-optimized;
[0124] In step 405, re-optimize the model adjustment parameters for:
[0125] , where Represents the learning rate, which is used to control the step size of parameter adjustment; Represents a time series, that is, random forest in the time series Make predictions on represents the gradient with respect to the prediction model parameters; Function representing the random forest prediction process; Indicates that in time series The multidimensional feature vector of .
[0126] In the embodiment of the present invention, by performing compensation analysis based on historical temperature data and historical humidity data, the direct impact of changes in temperature and humidity on the power generation efficiency of photovoltaic panels is reduced; by performing compensation analysis on historical light intensity and historical angle data, the impact of lighting conditions on power generation can be more accurately reflected; by processing and analyzing historical cleaning frequency data, the accuracy of subsequent predictions is improved; by extracting features related to temperature and humidity based on the compensation result of the first composite compensation unit, it is helpful to understand how environmental factors affect power generation efficiency; by extracting features related to light intensity and angle based on the compensation result of the second composite compensation unit, these features are directly related to the power generation capacity of photovoltaic panels; by extracting features related to cleaning activities based on the results of the cleaning frequency processing unit, the impact of cleaning activities on the power generation efficiency of photovoltaic panels is reflected; by establishing a prediction model, the model can predict the optimal cleaning cycle based on historical data and current environmental conditions; by feedback and evaluation analysis of real-time data, the prediction model is continuously corrected and improved to ensure that the model always maintains a high prediction accuracy. In summary, the example of the present invention solves the problems of low cleaning efficiency and insufficient intelligent management of current photovoltaic power stations. In actual situations, more data and context information may be needed to make specific decisions and optimization plans.
[0127] In addition, the formulas involved in the above are all calculated by removing dimensions and taking their numerical values. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The proportional coefficient in the formula and the various preset thresholds in the analysis process are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data; the size of the proportional coefficient is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the proportional coefficient depends on the amount of sample data and the preliminary setting of the corresponding processing coefficient for each group of sample data by technical personnel in this field; as long as it does not affect the proportional relationship between the parameter and the quantized value.
[0128] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically based on the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0129] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0130] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0131] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0132] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0133] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0134] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;
[0135] Finally: The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. Intelligent management system for periodic cleaning based on data analysis, characterized by: It includes multivariate data adaptation and adjustment module, feature analysis and extraction module, cycle cleaning prediction module and prediction evaluation and optimization module; The multivariate data adaptation and adjustment module is used to compensate and process the historical data of the photovoltaic power station; the multivariate data adaptation and adjustment module includes a first composite compensation unit, a second composite compensation unit and a cleaning frequency processing unit; The feature analysis and extraction module is used to extract features based on the compensation processing data and construct a multi-dimensional feature space; The feature analysis and extraction module includes a first composite feature extraction unit, a second composite feature extraction unit, and a cleaning activity feature extraction unit; The cycle cleaning prediction module is used to perform time series analysis based on the constructed multi-dimensional feature space and establish a prediction model to determine the optimal cleaning cycle; The prediction evaluation and optimization module is used to evaluate and analyze the optimal cleaning cycle of the prediction model and the real-time monitoring data, and to determine the accuracy of the model prediction; The process in which the first composite compensation unit is used to perform compensation analysis on the acquired historical temperature data and historical humidity data includes: The historical temperature data and historical humidity data of the photovoltaic power station are marked as and ; Extract the values of historical temperature and humidity data and perform compound compensation analysis: , where Indicates the output power after historical temperature and humidity composite compensation, reflecting the impact of temperature and humidity on the output power of photovoltaic panels; Represents the original output power of the photovoltaic panel under the historical temperature and humidity observation conditions; represent the sensitivity coefficients of temperature and humidity respectively; Respectively represent the reference temperature value and the reference humidity value; The process of the second composite compensation unit performing compensation analysis on the acquired historical light intensity and historical angle data includes: The historical light intensity data and historical angle data of the photovoltaic power station are marked as and ; Extract the values of historical light intensity and historical angle data, and perform composite compensation analysis: , where It represents the output power after the historical light intensity and historical angle are compositely compensated; represents the raw output power of the photovoltaic panel under historical light intensity and historical angle conditions; Indicates the light intensity reference value; represents the optimal angle; Represent the sensitivity index of light intensity and angle respectively; The cleaning frequency processing unit is used to process and analyze the acquired historical cleaning frequency, including: Set Existence Historical cleaning records, each cleaning record contains the cleaning time point , Average output power of photovoltaic panels before and after cleaning And cleaning costs ; Use the cleaning frequency processing formula for analysis: , , where Indicates the cleaning optimization power index; Indicates The average output power of photovoltaic panels increased after cleaning. It indicates the ratio of the output power increase after each cleaning to the cleaning cost, that is, the increase in output power after cleaning per unit cost; is the attenuation factor, which is used to indicate the degree of attenuation of the cleaning effect over time; represents the cleaning effect decay rate, and ; Indicates the time interval between two cleanings; The average value representing the long-term decay of the cleaning effect is used to evaluate the long-term impact of cleaning activities on the performance of photovoltaic panels; Indicates the cleaning time point identifier or index; The process of the first composite feature extraction unit for extracting features according to the compensation analysis result of the first composite compensation unit includes: Output power after temperature and humidity compound compensation , get some output power The data set is taken and the first composite feature extraction is performed: , where represents the first composite eigenvalue; represents the number of observations; Represents the observation index subscript; Indicates Set historical temperature data; Indicates Set historical humidity data; Indicates Output power of the group's historical observations; The process of the second composite feature extraction unit for extracting features according to the compensation analysis result of the second composite compensation unit includes: Collect the output power of photovoltaic panels after compound compensation under different light intensities and angles and the corresponding timestamp ; Assuming the output power after composite compensation Affected by the periodic changes in light intensity and angle, a periodic function is used to capture the periodic changes in light intensity and angle on the output power of the photovoltaic panel: , , , , where express is approximated as a Fourier series; represents the order of the Fourier series; Indicates the order index of the harmonic component, used to traverse from 1 to All harmonic components of Indicates the cycle length; are Fourier coefficients, which respectively represent the average value of the output power, the even harmonic components of the output power varying with time, and the odd harmonic components of the output power varying with time, reflecting the periodic characteristics of the output power varying with time; Indicates A small increment of the variable, for timestamps Integrate the differential of ; Define the second composite eigenvalue is the Fourier coefficient vector: ; The process of the cleaning activity feature extraction unit for extracting features according to the processing and analysis results of the cleaning frequency processing unit includes: Optimize power index by combining cleaning Perform computational analysis: , where Indicates the characteristic value of cleaning activity; Indicates the instant power recovery rate, that is, the rapid recovery of the photovoltaic panel power after cleaning, which is used to reflect the immediate effect of the cleaning activity; Indicates the long-term performance stability coefficient, which is used to analyze the long-term stability of photovoltaic panel performance after cleaning; Integrate the first composite eigenvalue, the second composite eigenvalue and the cleaning activity eigenvalue to form a multidimensional eigenvector set , and form a multidimensional feature space; where each feature vector is a point in the multidimensional feature space; The cycle cleaning prediction module performs time series analysis based on the constructed multi-dimensional feature space and establishes a prediction model to determine the optimal cleaning cycle. The process includes: Extract multiple training sample sets from the original data set, each sample set is used to build a decision tree; wherein, all multidimensional feature vector sets, i.e., points in the multidimensional feature space, are extracted from a single sample of the original data set; wherein the original data set represents a historical data set used to train the prediction model, including historical temperature, humidity, light intensity, angle, and cleaning frequency data; When building each decision tree, randomly select Some features are used as candidate feature sets, and the tree is grown based on the candidate features; among them, represents the number of randomly selected features as the size of the candidate feature set; Each decision tree makes a prediction based on the input feature vector, i.e. a point in the multidimensional feature space, and outputs a prediction result; The process of using random forest for prediction is: , where Represents random forest in time series The final prediction value for predicting the optimal cleaning cycle; A function representing the random forest prediction process, receiving a multidimensional feature vector and the set of all decision trees in the random forest , and as input, output the predicted value of the optimal cleaning cycle ; Indicates that in time series The multi-dimensional feature vector includes a first composite feature value, a second composite feature value, and a cleaning activity feature value; Represents the total number of trees in the random forest.
2. The intelligent management system for periodic cleaning based on data analysis according to claim 1 is characterized in that: The prediction evaluation and optimization module evaluates and analyzes the optimal cleaning cycle of the prediction model and the real-time monitoring data, and the process of judging the accuracy of the model prediction includes: Define the real-time monitoring data set as: ; in, Represents the photovoltaic panel in the time series No. Monitoring data, Indicates the identification or index number of the monitoring data. represents the amount of monitoring data, and ; The real-time monitoring data is the degree of dirt accumulation; the optimal cleaning cycle predicted by the random forest prediction model is compared and analyzed with the actual real-time monitoring data to evaluate whether the change in the degree of dirt accumulation at the current moment is in line with expectations; The predicted optimal cleaning cycle is defined as , and the real-time monitoring of the degree of dirt accumulation is ,but Based on historical data, the The degree of dirt accumulation at a certain moment can be evaluated and analyzed as follows: , where Indicates the acceptable range of prediction error; if the predicted optimal cleaning cycle corresponds to the degree of dirt accumulation With real-time monitoring of the degree of dirt accumulation The difference is at the threshold If the predicted optimal cleaning cycle at the current moment is within the range, the predicted optimal cleaning cycle is consistent with the dirt accumulation degree monitored in real time, and the predicted result is considered to be consistent with the actual situation. A cleaning condition including a cleaning start signal is generated, and the cleaning start signal is sent to the cleaning robot to allocate and start the cleaning task. The cleaning start signal includes the cleaning time, cleaning area and cleaning intensity. The cleaning conditions are: , where Indicates the cleaning task priority. A higher value indicates a higher priority; Indicates the set constant item to ensure the priority of cleaning tasks range, i.e. ; Indicates resource allocation; Represents the resource allocation coefficient, which is used to adjust the resource allocation ratio to ensure range, i.e. ; If the difference between the predicted dirt accumulation degree corresponding to the optimal cleaning cycle and the real-time monitored dirt accumulation degree exceeds the threshold range, the predicted optimal cleaning cycle at the current moment does not conform to the real-time monitored dirt accumulation degree, and it is considered that there is a problem with the prediction result, and the model adjustment parameters are re-optimized; Among them, re-optimize the model adjustment parameters for: , where Represents the learning rate, which is used to control the step size of parameter adjustment; Represents a time series, that is, random forest in the time series Make predictions on represents the gradient with respect to the prediction model parameters; Function representing the random forest prediction process; Indicates that in time series The multidimensional feature vector of .
