Photovoltaic cleaning robot path planning method fusing group cascade power generation analysis

Abstract

The invention discloses a photovoltaic cleaning robot path planning method fusing string level power generation analysis, and belongs to the technical field of data processing, and the method specifically comprises the steps: dividing historical string power generation data according to the same interval to construct an entropy model, collecting the data in real time, calculating an interval entropy, comparing the interval entropy with the model, and recognizing a suspected low power generation interval; key cleaning group strings are screened in combination with spatial neighborhood information, and a posterior probability is calculated through a Bayesian probability model to determine a cleaning priority; power station layout constraints and robot motion characteristics are combined, and an optimal cleaning path is generated by adopting a path planning algorithm; through multi-dimensional data fusion and an intelligent algorithm, precise identification and path optimization of photovoltaic string cleaning requirements are realized, the cleaning efficiency is effectively improved, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular to a photovoltaic cleaning robot path planning method integrating group cascade power generation analysis. Background Art

[0002] With the rapid development of photovoltaic power generation technology, the scale of photovoltaic power stations continues to expand. During the operation of photovoltaic power stations, dust and dirt on the surface of photovoltaic panels will reduce the power generation efficiency of photovoltaic panels, so regular cleaning of photovoltaic panels is necessary. At present, using photovoltaic cleaning robots to clean photovoltaic panels has become a common method. However, existing photovoltaic cleaning robot path planning methods lack accurate analysis of the power generation status of photovoltaic strings, resulting in unreasonable cleaning path planning, low cleaning efficiency, and high cleaning costs. For example, some methods do not fully consider the power generation differences between different photovoltaic strings and uniformly clean all photovoltaic strings, resulting in a waste of cleaning resources. Other methods, although they take power generation into consideration, the analysis is not detailed enough and cannot accurately identify the photovoltaic strings that truly need to be cleaned, thus affecting the cleaning effect and the overall power generation efficiency of the power station. Therefore, there is an urgent need for a method that can accurately analyze the power generation of photovoltaic strings and rationally plan the cleaning robot path accordingly. Summary of the Invention

[0003] In response to the shortcomings of the existing technology, the present invention proposes a photovoltaic cleaning robot path planning method that integrates string-level power generation analysis. The historical string power generation data is divided into the same interval to construct an entropy model. The interval entropy is calculated by real-time data collection and compared with the model to identify suspected low-power generation intervals. The spatial neighborhood information is combined to screen key cleaning strings, and the posterior probability is calculated through a Bayesian probability model to determine the cleaning priority. The path planning algorithm is used to generate the optimal cleaning path based on the power station layout constraints and the robot motion characteristics. The present invention realizes the accurate identification of photovoltaic string cleaning needs and path optimization through multi-dimensional data fusion and intelligent algorithms, effectively improving cleaning efficiency and reducing operation and maintenance costs.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] The photovoltaic cleaning robot path planning method, which integrates group cascade power generation analysis, includes:

[0006] Step S1: Divide the pre-acquired photovoltaic string power generation key-value pair data set into intervals, calculate the interval entropy value, and construct a string-level power generation entropy value model based on the interval entropy value statistics;

[0007] Step S2: obtaining in real time a key-value pair data set of photovoltaic string power generation at each moment within a preset time period in the photovoltaic power station, and performing interval division and interval real-time entropy value calculation;

[0008] Step S3: Compare the real-time entropy value of each interval with the group string level power generation entropy value model to obtain the suspected low power generation interval;

[0009] Step S4: Analyze each suspected low power generation interval in combination with its spatial neighborhood information, and use the entropy change trend of the neighborhood interval to determine the key string cleaning groups and mark the low power generation interval;

[0010] Step S5: Construct a Bayesian probability model, take the real-time entropy value of the key cleaning strings as input, calculate the posterior probability, and combine it with the probability threshold to obtain the priority score of the string cleaning;

[0011] Step S6: Based on the results of the entropy analysis and the Bayesian probability model, combined with the layout structure of the photovoltaic power station and the motion characteristics of the cleaning robot, a path planning algorithm is used to plan the cleaning path of the cleaning robot.

[0012] Specifically, the specific steps of step S1 include:

[0013] S1.1: Obtain a high-level data set of the photovoltaic string power generation at each moment within a preset time period from photovoltaic strings that have historically generated normal power, and perform normalization processing to obtain a data set of photovoltaic string power generation key-value pairs; each data point in the high-level data set of photovoltaic string power generation includes a string number, a timestamp, and a corresponding power generation value;

[0014] S1.2: Divide the PV string power generation key-value pair data set into N 8×8 intervals;

[0015] S1.3: Calculate the power generation distribution histogram P for each interval = {p k,1 ,...,p k,a}, and calculate the entropy value of each interval based on the power generation distribution histogram Among them, p k,i represents the probability of the power generation value i appearing in the kth interval, and i∈[1,a], k∈[1,N], a represents the power generation level number, w k,i represents the weight coefficient of the power generation value i in the kth interval, and δ represents the smoothing parameter;

[0016] S1.4: Set entropy value anomaly thresholds h1 and h2, where h1 represents the lowest entropy value anomaly and h2 represents the highest entropy value anomaly;

[0017] If E k k >h2, then eliminate the interval;

[0018] If h1≤E k ≤h2, then based on the entropy value E of each interval k ​, calculate the mean μ and standard deviation σ of all interval entropy values;

[0019] S1.5: Based on the mean μ and standard deviation σ of all interval entropy values, construct a group string power generation entropy value model, and satisfy Y = [μ - 3σ, μ + 3σ], where Y represents the range of the group string power generation entropy value model.

[0020] Specifically, the specific steps of step S3 include:

[0021] S3.1: Obtain the real-time entropy value E of each interval in the key-value pair data set of photovoltaic string power generation at each moment in the preset time period calculated in step S2 k ';

[0022] S3.2: Obtain the model range Y of the group string power generation entropy value;

[0023] S3.3: For each interval, determine its real-time entropy value E k 'Whether it is within the range Y of the group string level power generation entropy value model;

[0024] If E k ′∈[μ-3σ,μ+3σ], then the power generation of the photovoltaic string corresponding to the judgment interval is normal;

[0025] like The interval is then determined to be a suspected low power generation interval, that is, the power generation of the photovoltaic strings corresponding to the interval is affected by dust and dirt factors;

[0026] S3.4: According to the interval index value (z row ,z col ), calculate the position coordinates (x, y) of the photovoltaic string corresponding to the suspected low power generation interval = (z row ×8,z col ×8), where z row and z col Respectively represent the horizontal and vertical index values ​​of the interval;

[0027] S3.5: Record the suspected low power generation interval and the location coordinates (x, y) of the corresponding PV string.

[0028] Specifically, the specific steps of step S4 include:

[0029] S4.1: Obtain suspected low power generation interval patch m The position coordinates (x, y) of the , based on the position coordinates (x, y) extract the neighborhood interval patch ′ obtained after the 3×3 neighborhood expansion m , and calculate the suspected low power generation interval patch m Entropy and the neighborhood interval patch′m Entropy Where m represents the number of suspected low power generation intervals;

[0030] S4.2: Neighborhood-based patch′ m Entropy The mean entropy value μ of the neighborhood interval is obtained by summing and averaging lin , and use the standard deviation calculation formula to get the entropy standard deviation σ lin ;

[0031] S4.3: By calculation and The absolute value of the difference between the suspected low power generation interval and the neighborhood interval is obtained

[0032] S4.4: Set the entropy difference threshold h3;

[0033] like The PV strings corresponding to the suspected low power generation interval are identified as key cleaning strings, marked, and their location information (x′, y′) is recorded;

[0034] like The PV strings corresponding to the suspected low power generation interval are determined to be affected by non-cleaning factors and are excluded from the cleaning range.

[0035] Specifically, the specific steps of step S5 include:

[0036] S5.1: Obtain the real-time entropy value of the key cleaning string And r≤m, where r represents the number of key cleaning strings in real time;

[0037] S5.2: Define a Bayesian probability model. Based on the historical PV string power generation key-value pair data set, obtain the prior probability P(X) of a decrease in power generation in the historical PV string power generation key-value pair data set, the prior probability P(X′) of a change in power generation due to non-cleaning factors in the historical PV string power generation key-value pair data set, and fit the probability distribution of entropy under the condition of power generation decrease in the historical PV string power generation key-value pair data set. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors Where X represents the power generation reduction data in the historical photovoltaic string power generation key-value pair data set, X′ represents the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set, and r′ represents the number of power generation reduction data in the historical photovoltaic string power generation key-value pair data set. The entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set is represented. The entropy value of the power generation change data caused by non-cleaning factors in the historical PV string power generation key-value pair data set.

[0038] Specifically, the specific steps of step S5 also include:

[0039] S5.3: Probability distribution of entropy under conditions of reduced power generation based on the fitted historical PV string power generation key-value pair data set And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The entropy value of the real-time key cleaning string Substitute as input and The calculation formula of is used to calculate the probability distribution of entropy under the condition of power generation reduction in the real-time photovoltaic string power generation key-value pair data set. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors

[0040] S5.4: Through P(X) and The product of P(X′) and The marginal probability of the entropy value in the real-time photovoltaic string power generation key-value pair data set is obtained by multiplying

[0041] Specifically, the specific steps of step S5 also include:

[0042] S5.5: Through P(X) and The marginal probability of the product divided by the entropy value Get the posterior probability

[0043] S5.6: Set probability threshold h4;

[0044] like It is determined that the key cleaning string is caused by non-cleaning factors.

[0045] like It is determined that the key cleaning string is the power generation change caused by the cleaning factor, and the posterior probability Serves as a priority score for string cleaning.

[0046] Specifically, the probability distribution of entropy value under the condition of power generation reduction in the historical photovoltaic string power generation key-value pair data set in S5.2 is fitted And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The specific process includes:

[0047] S5.21: Reduce the entropy of the power generation in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set X , the standard deviation σ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set X Substituting into the formula of Gaussian distribution function, the probability distribution of entropy value under the condition of reduced power generation in the historical photovoltaic string power generation key-value pair data set is fitted.

[0048] S5.22: The entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X ′, the standard deviation σ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X′ Substituted into the formula of Gaussian distribution function, the probability distribution of entropy value in the historical photovoltaic string power generation key-value pair data set under the condition of power generation change caused by non-cleaning factors is fitted

[0049] Specifically, the specific steps of step S6 include:

[0050] S6.1: Obtain a set of all strings with abnormal entropy values, and obtain a set of all strings whose posterior probabilities exceed a probability threshold from the Bayesian probability model. Take the intersection of the two to generate a preliminary set of cleaned candidate strings.

[0051] S6.2: Map the physical coordinates of the string in the PV array according to the string number, and extract obstacle information from the PV power station layout database;

[0052] S6.3: Bayesian Posterior Probability The entropy value result and the attenuation term based on the last cleaning time are weighted and summed to obtain the cleaning priority; the entropy value result is the difference between 1 and the real-time entropy value; the real-time entropy value is the ratio of the first variable to the second variable; the first variable is the difference between the real-time entropy value of group b and the lowest abnormal entropy value; the second variable is the difference between the highest abnormal entropy value and the lowest abnormal entropy value;

[0053] S6.4: Normalize the cleaning priority to the interval [0, 1] and divide it into three levels; the threshold nodes in the division process are set to 0.4 and 0.7;

[0054] S6.5: Based on the photovoltaic array layout constraints, a path planning algorithm is used to plan the cleaning path of the cleaning robot, and the robot is controlled to perform the cleaning task according to the cleaning path.

[0055] Specifically, the photovoltaic string power generation key-value pair data set in step S2 includes power generation key-value pairs of multiple photovoltaic strings at different times, and each power generation key-value pair is a data pair of a string number and a power generation value.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. This paper proposes a path planning method for photovoltaic cleaning robots that integrates string-level power generation analysis. The method quantifies the degree of power generation anomalies through a string-level power generation entropy model, combines spatial neighborhood information to eliminate environmental interference, and avoids the misjudgment of dust and equipment failures in traditional methods. A Bayesian probability model further distinguishes between cleanable and non-cleanable factors, generating a prioritized cleaning string list. This makes cleaning targets more targeted, reduces ineffective work, and improves resource utilization.

[0058] 2. The present invention proposes a path planning method for photovoltaic cleaning robots that integrates group-level power generation analysis. Based on the layout constraints of the photovoltaic array, such as row and column structure and obstacles, a distance matrix is ​​constructed, and a heuristic function is designed in combination with priority. A cleaning path that takes into account both efficiency and energy consumption is generated through a path planning algorithm. The dynamic feedback mechanism supports real-time replanning during the cleaning process, adapts to environmental changes, and ensures that the cleaning robot completes its task with the shortest path and lowest energy consumption, thereby improving the intelligence level of photovoltaic power station operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of the photovoltaic cleaning robot path planning method integrating group cascade power generation analysis of the present invention;

[0060] Figure 2 This is a flow chart showing the principle of the photovoltaic cleaning robot path planning method integrating group cascade power generation analysis in the present invention;

[0061] Figure 3 This is a flow chart for constructing a group string power generation entropy model in the photovoltaic cleaning robot path planning method that integrates group string power generation analysis in the present invention. DETAILED DESCRIPTION

[0062] Example 1:

[0063] See also Figure 1-Figure 3 The present invention provides an embodiment of a photovoltaic cleaning robot path planning method integrating group cascade power generation analysis, comprising the following steps:

[0064] Step S1: Divide the pre-acquired photovoltaic string power generation key-value pair data set into intervals, calculate the interval entropy value, and construct a string-level power generation entropy value model based on the interval entropy value statistics;

[0065] It should be noted that the PV string power generation key-value pair data set refers to the time series data set of PV string power generation. This is time series data, consisting of the power generation values ​​of each PV string in a PV power station, its corresponding timestamp, and the string identifier. It is structured numerical data. In the string power generation key-value pair, the key is the power generation data that identifies the string at a specific moment, typically composed of a string number and a timestamp, and the value is the power generation value of the string at that moment. This data is collected in real time by the PV power station's smart combiner box, string-level inverter, or monitoring system.

[0066] Step S2: Real-time acquisition of a data set of photovoltaic string power generation key-value pairs at each moment within a preset time period within the photovoltaic power station, and performing interval division and interval real-time entropy calculation; the photovoltaic string power generation key-value pair data set includes power generation key-value pairs of multiple photovoltaic strings at different moments, each power generation key-value pair being a data pair of a string number and a power generation value;

[0067] It should be understood that when calculating the real-time entropy value of each interval in step S2, the specific process and specific formula refer to step S1.

[0068] Step S3: Compare the real-time entropy value of each interval with the group string level power generation entropy value model, and obtain the suspected low power generation interval based on the comparison result;

[0069] Step S4: Analyze each suspected low power generation interval in combination with its spatial neighborhood information, and use the entropy change trend of the neighborhood interval to determine the key string cleaning groups and mark the low power generation interval;

[0070] Step S5: Construct a Bayesian probability model, take the real-time entropy value of the key cleaning strings as input, calculate the posterior probability, and combine it with the probability threshold to obtain the priority score of the string cleaning;

[0071] Step S6: Based on the results of the entropy analysis and the Bayesian probability model, combined with the layout structure of the photovoltaic power station and the motion characteristics of the cleaning robot, a path planning algorithm is used to plan the cleaning path of the cleaning robot.

[0072] To sum up, the present invention realizes the detection of power generation reduction by analyzing the entropy value of the photovoltaic string power generation key-value pair data set, combining spatial neighborhood information and Bayesian probability model method. Moreover, all entropy value calculation formulas in this application refer to the entropy value calculation formula in S1.3.

[0073] The specific steps of step S1 include:

[0074] S1.1: Obtain a high-level data set of string power generation at each moment within a preset time period from PV strings with historically normal power generation, and perform normalization processing to obtain a data set of PV string power generation key-value pairs; each data in the PV string power generation key-value pair data set includes a string number, a timestamp, and a corresponding power generation value;

[0075] S1.2: Divide the PV string power generation key-value pair data set into N 8×8 intervals, and Among them, A row A represents the number of string power generation data sets in the string number dimension, col Represents the quantity of the string power generation data set in the time dimension, Indicates rounding down;

[0076] S1.3: Calculate the power generation distribution histogram P for each interval = {p k,1 ,...,p k,a}, and calculate the entropy value of each interval based on the power generation distribution histogram Among them, p k,i represents the probability of the power generation value i appearing in the kth interval, and i∈[1,a], k∈[1,N], a represents the power generation level number, w k,i represents the weight coefficient of the power generation value i in the kth interval, and δ represents the smoothing parameter;

[0077] in, f k,i Indicates the number of times the power generation value i appears in the kth interval.

[0078] max(f k,1 ,...,f k,a ) represents the maximum frequency of power generation value in the interval.

[0079] It should be noted that in actual calculations, there will be a situation where the probability of the power generation value i appearing in the kth interval is 0, that is, p k,i = 0, resulting in logarithmic calculation log(p k,i ) appears infinite, so the smoothing parameter δ is increased to solve the problem of p k,i =0, thereby improving the stability of the calculation; at the same time, the weight coefficient can enhance the contribution of high-frequency power generation values ​​and suppress the noise influence of low-frequency power generation values, making the entropy value calculation more consistent with the distribution characteristics of power generation reduction in actual scenarios. In summary, the modified formula can better adapt to complex environments.

[0080] S1.4: Set entropy value anomaly thresholds h1 and h2, where h1 represents the lowest entropy value anomaly and h2 represents the highest entropy value anomaly;

[0081] If E k k >h2, then eliminate the interval;

[0082] If h1≤E k ≤h2, then based on the entropy value E of each interval k , calculate the mean μ and standard deviation σ of all interval entropy values;

[0083] S1.5: Based on the mean μ and standard deviation σ of all interval entropy values, construct a group string power generation entropy value model, and satisfy Y = [μ - 3σ, μ + 3σ], where Y represents the range of the group string power generation entropy value model.

[0084] It should be noted that the string-level power generation entropy model is a statistical model based on entropy, which is used to describe the power distribution characteristics of string-level power generation in a key-value pair data set. In the string-level power generation detection method based on entropy, the string-level power generation entropy model range Y = [μ-3σ, μ+3σ]. Because the mean and standard deviation of the entropy value are statistical results based on a large number of string power generation data set intervals, they can reflect the typical entropy value distribution of string-level power generation. The mean and standard deviation can be used to determine the concentration range and dispersion of the string-level power generation entropy value. Moreover, the entropy value of string-level power generation usually fluctuates within a certain range, and the entropy value of power generation reduction often exceeds this range. By defining a range Y, it is possible to simply and efficiently determine whether the entropy value of a certain interval belongs to the typical range of string-level power generation. Therefore, in actual detection, it is only necessary to compare the entropy value of the real-time interval with the string-level power generation entropy model range Y to quickly determine whether it belongs to the string-level power generation value.

[0085] For example, assuming that the size of the photovoltaic string power generation key-value pair data set is 256×256, that is, the string number dimension and the time dimension are both 256, and the interval size is 8×8, then the number of intervals N=1024, and the number of power generation levels k=256;

[0086] (1) Calculate entropy:

[0087] 1) Calculate the power generation distribution histogram P for each interval;

[0088] 2) Calculate the entropy value E of each interval according to the entropy formula k ;

[0089] (2) Calculate the mean μ and standard deviation σ of all interval entropy values;

[0090] (3) Constructing the entropy model of group cascade power generation:​

[0091] Assuming μ = 5.2, σ = 0.8, the model range of the entropy value of the group string power generation is: [5.2-3×0.8,5.2+3×0.8] = [2.8,7.6].

[0092] The specific steps of step S3 include:

[0093] S3.1: Obtain the real-time entropy value E of each interval in the key-value pair data set of photovoltaic string power generation at each moment in the preset time period calculated in step S2 k ';

[0094] S3.2: Obtain the model range Y of the group string power generation entropy value;

[0095] S3.3: For each interval, determine its real-time entropy value E k 'Whether it is within the range Y of the group string level power generation entropy value model;

[0096] If E k ′∈[μ-3σ,μ+3σ], then the power generation of the photovoltaic string corresponding to this interval is judged to be normal;

[0097] like This interval is then determined to be a suspected low power generation interval, meaning that the power generation of the photovoltaic strings corresponding to this interval is affected by dust and dirt.

[0098] S3.4: According to the interval index value (z row ,z col ), calculate the position coordinates (x, y) of the photovoltaic string corresponding to the suspected low power generation interval = (z row ×8,z col ×8), where z row and z col Respectively represent the horizontal and vertical index values ​​of the interval;

[0099] S3.5: Record the suspected low power generation interval and the location coordinates (x, y) of the corresponding PV string.

[0100] The specific steps of step S4 include:

[0101] S4.1: Obtain suspected low power generation interval patch m The position coordinates (x, y) of the , based on the position coordinates (x, y) extract the neighborhood interval patch ′ obtained after the 3×3 neighborhood expansion m , and calculate the suspected low power generation interval patch m Entropy and the neighborhood interval patch′ m Entropy Where m represents the number of suspected low power generation intervals;

[0102] S4.2: Neighborhood-based patch′ m Entropy The mean entropy value μ of the neighborhood interval is obtained by summing and averaging lin , and use the standard deviation calculation formula to get the entropy standard deviation σ lin , wherein the standard deviation calculation formula is the existing technical content in this field and is not an inventive solution of this application, and is not described here in detail;

[0103] S4.3: By calculation and The absolute value of the difference between the suspected low power generation interval and the neighborhood interval is obtained

[0104] S4.4: Set the entropy difference threshold h3;

[0105] like The photovoltaic string corresponding to the suspected low power generation interval is determined as the key cleaning string, marked, and its location information (x′, y′) is recorded;

[0106] like The PV string corresponding to the suspected low power generation interval is determined to be affected by non-cleaning factors and is excluded from the cleaning range.

[0107] The specific steps of step S5 include:

[0108] S5.1: Obtain the real-time entropy value of the key cleaning string And r≤m, where r represents the number of key cleaning strings in real time;

[0109] S5.2: Define a Bayesian probability model based on the historical PV string power generation key-value pair data set to obtain the prior probability of power generation reduction in the historical PV string power generation key-value pair data set The prior probability P(X′)=1-P(X) of the change in power generation due to non-cleaning factors in the historical PV string power generation key-value pair data set is fitted, and the probability distribution of entropy under the condition of power generation reduction in the historical PV string power generation key-value pair data set is fitted. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors Where X represents the power generation reduction data in the historical photovoltaic string power generation key-value pair data set, X′ represents the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set, and r′ represents the number of power generation reduction data in the historical photovoltaic string power generation key-value pair data set. The entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set is represented. The entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set;

[0110] S5.3: Probability distribution of entropy under conditions of reduced power generation based on the fitted historical PV string power generation key-value pair data set And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The entropy value of the real-time key cleaning string Substitute as input and The calculation formula of is used to calculate the probability distribution of entropy under the condition of power generation reduction in the real-time photovoltaic string power generation key-value pair data set. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors

[0111] S5.4: Through P(X) and The product of P(X′) and The marginal probability of the entropy value in the real-time photovoltaic string power generation key-value pair data set is obtained by multiplying

[0112] S5.5: Through P(X) and The marginal probability of the product divided by the entropy value Get the posterior probability

[0113] S5.6: Set probability threshold h4;

[0114] like It is determined that the power generation of the key cleaning string is affected by non-cleaning factors;

[0115] like It is determined that the key cleaning string is the power generation change caused by the cleaning factor, and the posterior probability As the priority score for cleaning this string.

[0116] Probability distribution of entropy value under the condition of reduced power generation in fitting the historical PV string power generation key-value pair data set in S5.2 And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The specific process includes:

[0117] S5.21: Reduce the entropy of the power generation in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set X , the standard deviation σ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data setX Substituting into the formula of Gaussian distribution function, the probability distribution of entropy value under the condition of reduced power generation in the historical photovoltaic string power generation key-value pair data set is fitted.

[0118] S5.22: The entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X′ , the standard deviation σ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X′ Substituted into the formula of Gaussian distribution function, the probability distribution of entropy value in the historical photovoltaic string power generation key-value pair data set under the condition of power generation change caused by non-cleaning factors is fitted

[0119] Example 2:

[0120] In this embodiment, the specific steps of step S6 include:

[0121] S6.1: Obtain a set of all strings with abnormal entropy values, and obtain a set of all strings whose posterior probabilities exceed a probability threshold from the Bayesian probability model. Take the intersection of the two to generate a preliminary set of cleaned candidate strings.

[0122] S6.2: Map the physical coordinates of the string in the PV array according to the string number, and extract obstacle information from the PV power station layout database;

[0123] S6.3: Bayesian Posterior Probability The entropy value result and the attenuation term based on the last cleaning time are weighted and summed to obtain the cleaning priority; the entropy value result is the difference between 1 and the real-time entropy value; the real-time entropy value is the ratio of the first variable to the second variable; the first variable is the difference between the real-time entropy value of group b and the lowest abnormal entropy value; the second variable is the difference between the highest abnormal entropy value and the lowest abnormal entropy value;

[0124] S6.4: Normalize the cleaning priority to the interval [0, 1] and divide it into three levels; set the threshold nodes in the division process to 0.4 and 0.7. The normalization method is the prior art content in this field and does not constitute the inventive solution of this application, and is not described here in detail;

[0125] S6.5: In combination with the photovoltaic array layout constraints, a path planning algorithm is used to plan the cleaning path of the cleaning robot, and the robot is controlled to perform the cleaning task according to the cleaning path. The path planning algorithm is the existing technical content in this field and is not an inventive solution of this application, so it will not be elaborated here.

[0126] Furthermore, the photovoltaic array layout constraints are:

[0127]

[0128] in, Indicates the constraint distance from string b to c, represents the Euclidean distance from string b to c, θ represents the weight coefficient of layout constraint penalty, LayoutPenalty b,c represents the layout penalty term, b and c represent the group string index.

[0129] Among them, for the layout penalty item LayoutPenalty b,c ,satisfy:

[0130] If obstacles need to be crossed from string b to c, LayoutPenalty b,c = C1, if there is no need to cross obstacles from string b to c, then LayoutPenalty b,c =0, where C1 represents the obstacle constraint penalty value;

[0131] If the path from string b to c passes through an obstacle area, then LayoutPenalty b,c = C2, if the path from string b to c does not pass through the obstacle area, then LayoutPenalty b,c =0, where C2 represents the path constraint penalty value;

[0132] If the robot turning cost needs to be considered when going from string b to c, then LayoutPenalty b,c =η×Δψ, if the robot turning cost does not need to be considered when going from string b to c, then LayoutPenalty b,c =0, where η represents the direction constraint penalty value and Δψ represents the steering angle.

[0133] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific embodiments. The above-mentioned specific embodiments are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention. These are all protected by the present invention.

Claims

1. A photovoltaic cleaning robot path planning method integrating group cascade power generation analysis is characterized by: include: Step S1: Divide the pre-acquired photovoltaic string power generation key-value pair data set into intervals, calculate the interval entropy value, and construct a string-level power generation entropy value model based on the interval entropy value statistics; Step S2: obtaining in real time a key-value pair data set of photovoltaic string power generation at each moment within a preset time period in the photovoltaic power station, and performing interval division and interval real-time entropy value calculation; Step S3: Compare the real-time entropy value of each interval with the group string level power generation entropy value model to obtain the suspected low power generation interval; Step S4: Analyze each suspected low power generation interval in combination with its spatial neighborhood information, and use the entropy change trend of the neighborhood interval to determine the key string cleaning groups and mark the low power generation interval; Step S5: Construct a Bayesian probability model, take the real-time entropy value of the key cleaning strings as input, calculate the posterior probability, and combine it with the probability threshold to obtain the priority score of the string cleaning; Step S6: Based on the results of the entropy analysis and the Bayesian probability model, combined with the layout structure of the photovoltaic power station and the motion characteristics of the cleaning robot, a path planning algorithm is used to plan the cleaning path of the cleaning robot.

2. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 1, characterized in that: The specific steps of step S1 include: S1.1: Obtain a high-level data set of the photovoltaic string power generation at each moment within a preset time period from photovoltaic strings that have historically generated normal power, and perform normalization processing to obtain a data set of photovoltaic string power generation key-value pairs; each data point in the high-level data set of photovoltaic string power generation includes a string number, a timestamp, and a corresponding power generation value; S1.2: Divide the PV string power generation key-value pair data set into N 8×8 intervals; S1.3: Calculate the power generation distribution histogram P for each interval = {p k,1 ,...,p k,a }, and calculate the entropy value of each interval based on the power generation distribution histogram Among them, p k,i represents the probability of the power generation value i appearing in the kth interval, and i∈[1,a], k∈[1,N], a represents the power generation level number, w k,i represents the weight coefficient of the power generation value i in the kth interval, and δ represents the smoothing parameter; S1.4: Set entropy value anomaly thresholds h1 and h2, where h1 represents the lowest entropy value anomaly and h2 represents the highest entropy value anomaly; If E k k >h2, then eliminate the interval;​ If h1≤E k ≤h2, then based on the entropy value E of each interval k , calculate the mean μ and standard deviation σ of all interval entropy values; S1.5: Based on the mean μ and standard deviation σ of all interval entropy values, construct a group string power generation entropy value model, and satisfy Y = [μ - 3σ, μ + 3σ], where Y represents the range of the group string power generation entropy value model.

3. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 2, characterized in that: The specific steps of step S3 include: S3.1: Obtain the real-time entropy value E′ of each interval in the key-value pair data set of photovoltaic string power generation at each moment within the preset time period calculated in step S2 k ; S3.2: Obtain the model range Y of the group string power generation entropy value; S3.3: For each interval, determine its real-time entropy value E′ k Whether it is within the range Y of the string-level power generation entropy model; If E′ k ∈[μ-3σ,μ+3σ], then the power generation of the photovoltaic string corresponding to the judgment interval is normal; like The interval is then determined to be a suspected low power generation interval, that is, the power generation of the photovoltaic strings corresponding to the interval is affected by dust and dirt factors; S3.4: According to the interval index value (z row ,z col ), calculate the position coordinates (x, y) of the photovoltaic string corresponding to the suspected low power generation interval = (z row ×8,z col ×8), where z row and z col Respectively represent the horizontal and vertical index values ​​of the interval; S3.5: Record the suspected low power generation interval and the location coordinates (x, y) of the corresponding PV string.

4. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 3, characterized in that: The specific steps of step S4 include: S4.1: Obtain suspected low power generation interval patch m The position coordinates (x, y) of the , based on the position coordinates (x, y) extract the neighborhood interval patch ′ obtained after the 3×3 neighborhood expansion m , and calculate the suspected low power generation interval patch m Entropy and the neighborhood interval patch′ m Entropy Where m represents the number of suspected low power generation intervals; S4.2: Neighborhood-based patch′ m Entropy The mean entropy value μ of the neighborhood interval is obtained by summing and averaging lin , and use the standard deviation calculation formula to get the entropy standard deviation σ lin ; S4.3: By calculation and The absolute value of the difference between the suspected low power generation interval and the neighborhood interval is obtained S4.4: Set the entropy difference threshold h3; like The PV strings corresponding to the suspected low power generation interval are identified as key cleaning strings, marked, and their location information (x′, y′) is recorded; like The PV strings corresponding to the suspected low power generation interval are determined to be affected by non-cleaning factors and are excluded from the cleaning range.

5. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 4, characterized in that: The specific steps of step S5 include: S5.1: Obtain the real-time entropy value of the key cleaning string And r≤m, where r represents the number of key cleaning strings in real time; S5.2: Define a Bayesian probability model. Based on the historical PV string power generation key-value pair data set, obtain the prior probability P(X) of a decrease in power generation in the historical PV string power generation key-value pair data set, the prior probability P(X′) of a change in power generation due to non-cleaning factors in the historical PV string power generation key-value pair data set, and fit the probability distribution of entropy under the condition of power generation decrease in the historical PV string power generation key-value pair data set. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors Where X represents the power generation reduction data in the historical photovoltaic string power generation key-value pair data set, X′ represents the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set, and r′ represents the number of power generation reduction data in the historical photovoltaic string power generation key-value pair data set. The entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set is represented. The entropy value of the power generation change data caused by non-cleaning factors in the historical PV string power generation key-value pair data set.

6. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 5, characterized in that: The specific steps of step S5 also include: S5.3: Probability distribution of entropy under conditions of reduced power generation based on the fitted historical PV string power generation key-value pair data set And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The entropy value of the real-time key cleaning string Substitute as input and The calculation formula of is used to calculate the probability distribution of entropy under the condition of power generation reduction in the real-time photovoltaic string power generation key-value pair data set. And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors S5.4: Through P(X) and The product of P(X′) and The marginal probability of the entropy value in the real-time photovoltaic string power generation key-value pair data set is obtained by multiplying 7. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 6, characterized in that: The specific steps of step S5 also include: S5.5: Through P(X) and The marginal probability of the product divided by the entropy value Get the posterior probability S5.6: Set probability threshold h4; like It is determined that the key cleaning string is caused by non-cleaning factors. like It is determined that the key cleaning string is the power generation change caused by the cleaning factor, and the posterior probability Serves as a priority score for string cleaning.

8. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 7, characterized in that: The probability distribution of entropy value under the condition of reduced power generation in the historical photovoltaic string power generation key-value pair data set in S5.2 is fitted And the probability distribution of entropy value under the condition of power generation change caused by non-cleaning factors The specific process includes: S5.21: Reduce the entropy of the power generation in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set X , the standard deviation σ of the entropy value of the power generation reduction data in the historical photovoltaic string power generation key-value pair data set X Substituting into the formula of Gaussian distribution function, the probability distribution of entropy value under the condition of power generation reduction in the historical photovoltaic string power generation key-value pair data set is fitted S5.22: The entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set The mean value μ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X′ , the standard deviation σ of the entropy value of the power generation change data caused by non-cleaning factors in the historical photovoltaic string power generation key-value pair data set X′ Substituted into the formula of Gaussian distribution function, the probability distribution of entropy value in the historical photovoltaic string power generation key-value pair data set under the condition of power generation change caused by non-cleaning factors is fitted 9. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 8, characterized in that: The specific steps of step S6 include: S6.1: Obtain a set of all strings with abnormal entropy values, and obtain a set of all strings whose posterior probabilities exceed a probability threshold from the Bayesian probability model. Take the intersection of the two to generate a preliminary set of cleaned candidate strings. S6.2: Map the physical coordinates of the string in the PV array according to the string number, and extract obstacle information from the PV power station layout database; S6.3: Bayesian Posterior Probability The entropy value result and the attenuation term based on the last cleaning time are weighted and summed to obtain the cleaning priority; the entropy value result is the difference between 1 and the real-time entropy value; the real-time entropy value is the ratio of the first variable to the second variable; the first variable is the difference between the real-time entropy value of group b and the lowest abnormal entropy value; the second variable is the difference between the highest abnormal entropy value and the lowest abnormal entropy value; S6.4: Normalize the cleaning priority to the interval [0,1] and divide it into three levels; S6.5: Based on the photovoltaic array layout constraints, a path planning algorithm is used to plan the cleaning path of the cleaning robot, and the robot is controlled to perform the cleaning task according to the cleaning path.

10. The photovoltaic cleaning robot path planning method integrating group cascade power generation analysis as claimed in claim 9, characterized in that: The photovoltaic string power generation key-value pair data set in step S2 includes power generation key-value pairs of multiple photovoltaic strings at different times, and each power generation key-value pair is a data pair of a string number and a power generation value.

Images