Photovoltaic power station equipment operation optimization method, system, device and readable storage medium

Abstract

The present application relates to the technical field of photovoltaic power generation, and discloses a photovoltaic power station equipment operation optimization method, system, device and readable storage medium, the present application obtains a radiation type index by using horizontal plane total radiation data and component surface direct radiation data of the photovoltaic power station equipment, to construct a proportional quantization model about direct radiation and scattered radiation, to facilitate capturing the radiation mutation of the actual operation of the photovoltaic power station under the condition of sand and dust; then by inputting the current radiation type index and real-time environmental data into the trained random forest prediction model, the present application breaks through the limitation of the traditional power prediction model which only depends on total radiation, and can obtain more accurate predicted power; by predicting the power, the current radiation type index and the working mode type, an intelligent decision model is used for decision-making to obtain the optimal equipment operation rule, and the operation state of the photovoltaic power station equipment is optimized by using the optimal equipment operation rule, so as to ensure the efficient and reliable operation of the photovoltaic power station equipment.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation technology, and specifically to a method, system, equipment, and readable storage medium for optimizing the operation of photovoltaic power plant equipment. Background Technology

[0002] Photovoltaic (PV) power generation systems convert solar energy into electricity using photovoltaic (PV) panels. Their power generation efficiency is affected by various factors, including radiation intensity, temperature, the material properties of the PV modules, and the installation tilt angle. Taking a certain region as an example, this area, with its abundant solar resources, has become a hotspot for PV power plant construction. However, it also faces severe challenges such as strong sandstorms, extreme temperature differences, and high radiation fluctuations. The average number of sandstorm days per year exceeds 60 days, and sandstorm coverage reduces the light transmittance of PV panels by 20%-40%. In summer, the temperature of PV modules can reach over 70°C, leading to significant power degradation and shortened lifespan. These special climatic conditions place higher demands on the power prediction accuracy and array operation optimization of PV power generation systems.

[0003] Currently, the technical solutions for regulating photovoltaic power generation systems mainly focus on power prediction models. Power prediction models include two types: physical models and data-driven models. Physical models derive power generation through the physical formula of irradiance-temperature-power, while data-driven models, such as long short-term memory networks and support vector machines (SVM), predict power by capturing the temporal characteristics of historical data.

[0004] The environmental parameters of a photovoltaic power plant are dynamically changing during actual operation. Since power prediction using physical models relies on idealized assumptions about environmental parameters, and data-driven models rely only on the total radiation in the time-series characteristics of historical data for prediction, the power prediction methods in related technologies are difficult to predict the power generation of a photovoltaic power plant under dynamic changes in real time, and it is also difficult to ensure the efficient and reliable operation of the photovoltaic power plant. Summary of the Invention

[0005] In view of this, the present invention provides a method, system, equipment and readable storage medium for optimizing the operation of photovoltaic power plant equipment, so as to solve the technical defects of power prediction models in related technologies that make it difficult to guarantee the efficient and reliable operation of photovoltaic power plants.

[0006] In a first aspect, the present invention provides a method for optimizing the operation of photovoltaic power plant equipment, the method comprising:

[0007] Based on real-time acquisition of total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, the current radiation type index and working mode type are obtained using a preset first function. The first function includes the functional relationship between total horizontal radiation data, direct radiation data of the module surface, and radiation type index.

[0008] Based on the current radiation type index, and by comprehensively acquiring real-time environmental data, a trained random forest prediction model is used to make predictions, resulting in predicted power generation at multiple preset time intervals and their respective confidence intervals.

[0009] Based on the aforementioned random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base.

[0010] Based on the rule base, the current radiation type index, the operating mode type, and the predicted power generation, an intelligent decision-making model is used to make decisions and obtain the optimal equipment operation rules. The optimal equipment operation rules are used to generate instructions to adjust the tilt angle of the photovoltaic power station equipment.

[0011] Through the above implementation method, a radiation type index is obtained by using the total horizontal radiation data and direct radiation data of the photovoltaic power station equipment and the component surface. This index is then used to construct a quantitative model of the ratio of direct radiation to diffuse radiation, which facilitates the capture of radiation abrupt changes under sandstorm conditions during the actual operation of the photovoltaic power station. Subsequently, by inputting the current radiation type index and real-time environmental data into the trained random forest prediction model, the limitation of traditional power prediction models relying solely on total radiation is overcome, thereby obtaining a more accurate predicted power generation under sandstorm weather. By predicting the power generation, the current radiation type index, and the operating mode type, a smart decision-making model is used to make decisions and obtain the optimal equipment operation rules. The optimal equipment operation rules are then used to optimize the operating status of the photovoltaic power station equipment, thereby ensuring the efficient and reliable operation of the photovoltaic power station equipment.

[0012] In one optional implementation, training the random forest prediction model includes:

[0013] Based on the historical environmental data of the current photovoltaic power station, the temporal extension features and spatial extension features of the radiation type index are obtained using a preset feature extraction method. The temporal extension features include: the mean, variance, temperature change rate of the radiation type index over 1 hour, and the average irradiance intensity over the previous 3 hours. The spatial extension features include: the power difference coefficient between adjacent strings and the string-level radiation attenuation rate.

[0014] By combining the historical environmental data, temporal extension features, and spatial extension features, the data is input into the random forest prediction model to be trained to obtain the predicted power generation.

[0015] Based on actual and predicted power generation, the input environmental data is filtered by ranking the importance of features to obtain several key feature variables.

[0016] Based on the aforementioned key feature variables, the feature input weights of the random forest prediction model are optimized to obtain the optimized random forest prediction model, and the output is the trained random forest prediction model.

[0017] Through the above implementation method, the temporal and spatial extended features of the radiation type index are obtained by utilizing historical environmental data. The random forest prediction model is then trained using the historical environmental data, temporal and spatial extended features. The trained forest prediction model can more accurately fit the key features in the environmental data, thereby making the predicted power generation more accurate and providing a reference for optimizing the operation of photovoltaic power plants.

[0018] In one alternative implementation, the radiation type index satisfies the following:

[0019] ,

[0020] in, Indicates the radiation type index; This represents the direct radiation data on the component surface; This represents the total radiation data on the horizontal plane; This represents a constant that prevents division by zero.

[0021] Through the above implementation method, a radiation type index is obtained by using the total horizontal radiation data and direct radiation data of the photovoltaic power station equipment and the component surface. The radiation type index is used to construct a quantitative model of the ratio of direct radiation to diffuse radiation, thereby facilitating the capture of radiation abrupt changes under sand and dust conditions during the actual operation of the photovoltaic power station. By introducing the radiation type index, the perception efficiency of the environmental characteristics of different environmental conditions affecting power generation during the actual operation of the photovoltaic power station is improved, and the convenience of optimizing the equipment operation status of the photovoltaic power station is improved.

[0022] In an optional implementation, the method further includes:

[0023] Based on the current radiation type index and the historical radiation type index, the radiation type index decay rate is obtained using a preset second function. The second function includes the functional relationship between the current radiation type index, the historical radiation type index, and the radiation type index decay rate.

[0024] Based on the predicted power generation and theoretical power generation, the predicted power loss is obtained using a preset third function, which includes the functional relationship between the predicted power generation, theoretical power generation, and predicted power loss.

[0025] Based on the radiation type exponential decay rate and the predicted power loss, the random forest algorithm is used to evaluate and obtain the cleaning priority and cleaning probability.

[0026] Based on the cleaning priority and cleaning probability, a preset cleaning task triggering threshold is used for comparison. When the cleaning probability is greater than the cleaning task triggering threshold, the photovoltaic power station equipment with high cleaning priority is cleaned.

[0027] Through the above implementation method, the random forest algorithm is used to evaluate the surface cleaning priority and cleaning probability of photovoltaic power station equipment based on the radiation type exponential decay rate and predicted power loss. When the cleaning probability is greater than the cleaning trigger threshold, the photovoltaic power station equipment with high cleaning priority is cleaned, thereby reducing the defect of reduced power generation efficiency of photovoltaic power station equipment due to dust cover.

[0028] In one optional implementation, the radiation type exponential decay rate satisfies the following:

[0029] ,

[0030] in, Indicates the radiation type index decay rate; express Radiation type index at any given time; express Historical radiation type index at any given time.

[0031] Through the above implementation method, the radiation type index decay rate can be used to easily sense sudden changes in radiation in the external environment, and to provide early warning for cleaning photovoltaic power station equipment based on sudden changes in the radiation type index, thereby avoiding the probability of the power generation efficiency of photovoltaic power station equipment being reduced due to dust obstruction.

[0032] In an optional implementation, the method further includes:

[0033] Based on the predicted power generation and the actual power generation after implementing the optimal equipment operation rules, the random forest prediction model is evaluated using the method of calculating the root mean square error or residual distribution.

[0034] When the evaluation result is greater than a preset threshold, a conservative mode instruction is generated, which is used to adjust the photovoltaic power station equipment to a fixed tilt angle.

[0035] The above implementation method uses the current actual power generation to evaluate the predicted power generation. When there is a large error in the evaluation result, it indicates that the current random forest prediction model deviates significantly from the actual state. By adjusting the photovoltaic power station equipment to a fixed tilt angle, the probability of a decrease in power generation caused by optimizing the operation of the photovoltaic power station can be reduced.

[0036] In an optional implementation, the method further includes:

[0037] The received real-time environmental data is monitored. When data loss occurs in the acquired real-time environmental data, the real-time environmental data is processed using a neighbor array interpolation method.

[0038] The above implementation methods can process real-time environmental data that has been lost, and can also facilitate the processing of significantly abnormal data, thereby ensuring the amount of real-time environmental data acquired and reducing the defect of large error in the prediction results output by the random forest prediction model due to insufficient data.

[0039] Secondly, the present invention provides a photovoltaic power plant equipment operation optimization system, the system comprising:

[0040] The radiation type index generation module, based on real-time acquisition of total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, uses a preset first function to obtain the current radiation type index and working mode type. The first function includes the functional relationship between total horizontal radiation data, direct radiation data of the component surface, and radiation type index.

[0041] The power generation prediction module is used to make predictions based on the current radiation type index, the comprehensive real-time environmental data obtained, and the trained random forest prediction model to obtain the predicted power generation and their respective confidence intervals for multiple preset time intervals.

[0042] The rule base construction module is used to obtain multiple interpretable rules based on the random forest prediction model using the random forest rule extraction method, and integrate them into a rule base;

[0043] The decision acquisition module is used to make decisions based on the rule base, the current radiation type index, the working mode type and the predicted power generation, using an intelligent decision model to obtain the optimal equipment operation rules. The optimal equipment operation rules are used to generate instructions to adjust the operating tilt angle of the photovoltaic power station equipment.

[0044] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the photovoltaic power plant equipment operation optimization method of the first aspect or any corresponding embodiment described above.

[0045] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the photovoltaic power plant equipment operation optimization method of the first aspect or any corresponding embodiment described above.

[0046] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the photovoltaic power plant equipment operation optimization method of the first aspect or any corresponding embodiment described above. Attached Figure Description

[0047] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0048] Figure 1 This is a flowchart illustrating a photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention;

[0049] Figure 2 This is a flowchart illustrating another photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention;

[0050] Figure 3 This is a flowchart illustrating another method for optimizing the operation of photovoltaic power plant equipment according to an embodiment of the present invention;

[0051] Figure 4 These are the feature importance analysis results output by the random forest prediction model in the photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention;

[0052] Figure 5 This is a flowchart illustrating another method for optimizing the operation of photovoltaic power plant equipment according to an embodiment of the present invention;

[0053] Figure 6 This is a scatter plot of predicted power generation and actual power generation constructed by the photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention;

[0054] Figure 7 This is a residual distribution diagram of predicted power and actual power generation constructed by the photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention;

[0055] Figure 8 This is a structural block diagram of a photovoltaic power plant equipment operation optimization system according to an embodiment of the present invention;

[0056] Figure 9 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] First, the application scenarios on which the implementation of the photovoltaic power plant equipment optimization method depends are described here.

[0059] Photovoltaic (PV) power generation systems convert solar energy into electricity using photovoltaic (PV) panels. Their power generation efficiency is affected by various factors, including radiation intensity, temperature, the material properties of the PV modules, and the tilt angle of the equipment. Taking a certain region as an example, this area, with its abundant solar resources, has become a hotspot for PV power plant construction. However, it also faces severe challenges such as strong sandstorms, extreme temperature differences, and high radiation fluctuations. The average number of sandstorm days per year exceeds 60 days, and sandstorm coverage reduces the light transmittance of PV panels by 20%-40%. In summer, the temperature of PV modules can reach over 70°C, leading to significant power degradation and shortened lifespan. These special climatic conditions place higher demands on the power prediction accuracy and array operation optimization of PV power generation systems.

[0060] Currently, the industry's technical solutions mainly focus on two directions: power prediction models and array physical control. However, the related technologies still have significant shortcomings in multi-dimensional collaborative optimization and dynamic environmental adaptability.

[0061] In the field of power prediction, the prediction models in related technologies include physical models and data-driven models. Physical models derive power generation through the physical formula of irradiance-temperature-power, but they rely on idealized assumptions about environmental parameters and are difficult to cope with the dynamic changes in the ratio of direct normal irradiance (DNI) to diffuse horizontal irradiance (DHI) during dust storms. Data-driven models, such as Long Short-Term Memory networks and Support Vector Machines, can capture the temporal characteristics of historical data, but they generally lack specific modeling for abrupt changes in radiation types. Schemes using data-driven models for prediction do not distinguish the differences in contributions between direct and diffuse radiation, leading to prediction errors when diffuse radiation drops sharply during dust storms. Furthermore, the conventional hourly data sampling frequency cannot capture minute-level dust storm events, resulting in significant lag in prediction results and making it difficult to provide a reliable basis for real-time control of photovoltaic power plant operations.

[0062] In the field of array optimization, related technologies focus on solar tracking and regular cleaning, which can improve the efficiency of direct light capture under clear sky conditions. However, these technologies rely on dual-axis motor drives, consuming a significant portion of power generation, and do not optimize tilt strategies for scattered radiation scenarios during sandstorms. When sandstorms cause scattered radiation to exceed 70%, the system still maintains a high tilt angle in direct radiation mode, causing the light incident angle to deviate from the optimal range, resulting in a loss of daily power generation. Furthermore, when using a dust warning system based on camera image recognition and infrared temperature detection for cleaning, the strong winds and sandstorms in the desert environment easily contaminate the camera lenses, leading to a high false detection rate and significantly increasing hardware costs and maintenance complexity.

[0063] A comprehensive analysis of the common shortcomings of related technologies reveals at least three problems: First, these technologies generally rely on the single indicator of total radiation (GHI, Global Horizontal Irradiance) without constructing a quantitative model of the ratio of direct to diffuse radiation, resulting in the inability to accurately capture radiation abrupt changes during sandstorms. Second, power prediction, tilt adjustment, and temperature control strategies operate independently, lacking a control closed loop centered on radiation characteristics and a multi-parameter collaborative optimization mechanism. Third, the high energy consumption of dual-axis tracking systems, the high cost of forced air cooling, and the rigid strategy of fixed-cycle cleaning in these technologies are difficult to adapt to the long-term economic needs of photovoltaic power plants in deserts.

[0064] In order to overcome the deficiencies disclosed above, an embodiment of the present invention provides a method embodiment. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0065] This embodiment provides a method for optimizing the operation of photovoltaic power plant equipment, which can be used in the control server of the aforementioned photovoltaic power plant. Figure 1 This is a flowchart of a photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps:

[0066] S101, based on real-time acquisition of total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, the current radiation type index and working mode type are obtained using a preset first function. The first function includes the functional relationship between total horizontal radiation data, direct radiation data of the component surface, and radiation type index.

[0067] Both the total horizontal radiation data and the direct radiation data of the module surface of the photovoltaic power station equipment can be collected using radiation data collected by dual-probe radiation sensors. For example, the EKO MS-57 solar tracking radiometer can be used to collect the direct radiation data of the module surface, while the Kipp & Zonn CMP6 total radiation meter can be used to collect the total horizontal radiation data of the photovoltaic module.

[0068] By utilizing the first function of the proportional quantification model for direct and diffuse radiation, a radiation type index is obtained to characterize direct and diffuse radiation. This facilitates the provision of data references for subsequent power generation prediction and improves the accuracy of the predicted power generation output by the prediction model.

[0069] For example, in this embodiment, the radiation type index in S101 satisfies the following:

[0070] ,

[0071] in, Indicates the radiation type index; This represents the direct radiation data on the component surface; This represents the total radiation data on the horizontal plane; This represents a constant that prevents division by zero.

[0072] By utilizing the total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, a radiation type index is obtained. The radiation type index is used to construct a quantitative model of the ratio of direct radiation to diffuse radiation, thereby facilitating the capture of radiation abrupt changes under dust conditions during the actual operation of the photovoltaic power station. By introducing the radiation type index, the perception efficiency of the environmental characteristics affecting the power generation power under different environmental conditions is improved during the actual operation of the photovoltaic power station, and the convenience of optimizing the equipment operation status of the photovoltaic power station is improved.

[0073] The operating mode type is used to adjust the actual operating status of photovoltaic power station equipment. It can be obtained based on the radiation type index. For example, when the radiation type index is greater than 0.7, it indicates that the current environment of the photovoltaic power station is in clear sky mode. The photovoltaic power station equipment is operated using a solar tracking algorithm to maximize power generation. When the radiation type index is less than 0.3, it indicates that the current environment of the photovoltaic power station is in dusty mode. Since the proportion of scattered radiation increases in dusty mode, the photovoltaic power station equipment can be optimized by performing a scattering optimization tilt angle to improve the light energy utilization rate of the photovoltaic power station equipment and thus improve the power generation of the photovoltaic power station equipment. When the radiation type index is greater than or equal to 0.3 and less than or equal to 0.7, it indicates that the ratio of direct radiation to scattered radiation is dynamic. The photovoltaic power station equipment needs to be dynamically balanced by tilting to ensure the power generation of the photovoltaic power station.

[0074] S102, based on the current radiation type index, and comprehensively acquiring real-time environmental data, a trained random forest prediction model is used to make predictions, obtaining the predicted power generation power and their respective confidence intervals for multiple preset time intervals.

[0075] Real-time environmental data can be obtained using preset environmental parameter sensors. For example, a patch-type temperature sensor with a monitoring accuracy of ±0.5℃ can be installed on the back panel of the photovoltaic power station equipment. The air temperature, air pressure, temperature and wind speed of the photovoltaic power station can be obtained using real-time monitoring data from the environmental meteorological station. This data can be transmitted via RS485 bus. The timestamp alignment error of the data transmitted via RS485 bus is ≤1 second, which can ensure the timeliness of real-time environmental data acquisition.

[0076] For example, real-time environmental data includes: RTI (Radiation Reduction Level) with a time resolution of 15 minutes, temperature, humidity, air pressure, wind speed, component temperature, total radiation, direct radiation, diffuse radiation, whether it is daytime, month, time, quarter, and dust cover, etc.

[0077] The Random Forest prediction model is a prediction model trained using the Random Forest Algorithm (RFA). The Random Forest algorithm is an ensemble learning algorithm that constructs multiple decision trees and aggregates the prediction results of these trees to arrive at the final decision.

[0078] The current radiation type index and real-time environmental data are input into the trained random forest prediction model. The random forest prediction model is used to predict the power generation and confidence interval. By using the radiation type index as an input variable, the shortcomings of related technologies in lacking targeted modeling of radiation type mutations are overcome. At the same time, it is convenient to distinguish the difference between direct radiation and diffuse radiation, and reduce the prediction error of the power generation efficiency of photovoltaic power plants in deserts.

[0079] In this embodiment, the confidence interval is the estimated interval range obtained based on the OOB (Out-of-Ba) error. The random forest prediction model is used to output the power prediction values ​​and 80% confidence bands for the next four time points, based on the environmental data with a time resolution of 15 minutes.

[0080] S103. Based on the random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base.

[0081] Interpretable rules can be extracted from a random forest prediction model and expressed in if-then form, as shown in the following example:

[0082] If RTI < 0.3 AND temperature > 30℃, then tilt angle = horizontal plane + 12°.

[0083] If RTI < 0.3 AND temperature ≤ 30℃, then tilt angle = horizontal plane + 10°.

[0084] If RTI>0.7 AND temperature>50℃ THEN tilt angle=60°.

[0085] Specifically, the rule "IF RTI>0.7 AND temperature>50℃ THEN tilt angle=60°" means that when the current radiation type index is greater than 0.7 and the temperature is greater than 50 degrees Celsius, this interpretable rule is executed to trigger tilt angle adjustment of the photovoltaic power station equipment. This can adjust the tilt angle of the photovoltaic power station equipment to 60°, utilizing the self-shading effect of the photovoltaic panels for cooling, thereby extending the estimated lifespan of the photovoltaic power station equipment components. This eliminates the drawbacks of traditional forced air cooling or water cooling solutions that lead to increased energy consumption costs, and also avoids the insufficiency of extending the lifespan of photovoltaic power station equipment components to maintain the power generation efficiency of the photovoltaic power station during high-temperature periods. It also overcomes the problem of power generation loss due to shutdown for cooling.

[0086] Since the random forest prediction model constructs multiple decision trees during the prediction process, and the decision path of each decision tree corresponds to an interpretable rule, by using the random forest rule extraction method, the decision path of each decision tree is treated as an interpretable rule, resulting in multiple interpretable rules corresponding to the predicted power generation. These multiple interpretable rules are then integrated into a rule base. Through the rule base, photovoltaic power plants can easily generate operating instructions based on the interpretable rules in the rule base. The interpretable rules can be used to adjust the operating tilt angle of photovoltaic power plant equipment, thereby ensuring the efficient operation of the photovoltaic power plant.

[0087] S104, based on the rule base, the current radiation type index, the working mode type, and the predicted power generation, a smart decision-making model is used to make a decision to obtain the optimal equipment operation rule. The optimal equipment operation rule is used to generate instructions to adjust the tilt angle of the photovoltaic power station equipment.

[0088] After acquiring multiple interpretable rules from the rule base, the intelligent decision-making model is used based on the current radiation type index, operating mode type, and predicted power generation. The intelligent decision-making model can be obtained through definitions from external users, such as the maximum power generation, thus obtaining the optimal equipment operation rules. By inputting the optimal equipment operation rules into the photovoltaic power station equipment, the photovoltaic power station equipment adjusts its tilt angle according to the optimal equipment operation rules, thereby achieving the fusion of radiation type characteristics and improving the power generation efficiency and grid stability of the photovoltaic power generation equipment.

[0089] In the photovoltaic power station equipment operation optimization method provided in this embodiment, a radiation type index is obtained by using the total horizontal radiation data and direct radiation data of the module surface of the photovoltaic power station equipment. This index is then used to construct a quantitative model of the ratio of direct radiation to diffuse radiation, which facilitates the capture of radiation abrupt changes under sandstorm conditions during the actual operation of the photovoltaic power station. Subsequently, by inputting the current radiation type index and real-time environmental data into the trained random forest prediction model, the limitation of traditional power prediction models relying solely on total radiation is overcome, thereby obtaining a more accurate predicted power generation under sandstorm weather. By predicting the power generation, the current radiation type index, and the operating mode type, a smart decision-making model is used to make decisions and obtain the optimal equipment operation rules. The optimal equipment operation rules are then used to optimize the operating status of the photovoltaic power station equipment, thereby ensuring the efficient and reliable operation of the photovoltaic power station equipment.

[0090] In some embodiments, to reduce the possibility of large errors in the predicted power generation due to missing real-time environmental data, this embodiment further includes the following before S101:

[0091] The received real-time environmental data is monitored, and when data loss occurs, the real-time environmental data is processed using a neighbor array interpolation method.

[0092] For example, based on the time and date of the currently received real-time environmental data, environmental data corresponding to the current scene is obtained from historical environmental data, and the obtained data is used to supplement the environmental data through neighbor array interpolation, thereby avoiding the defect of large prediction error in the output of random forest prediction model due to insufficient real-time environmental data.

[0093] In some embodiments, the present invention also provides a method for optimizing the operation of photovoltaic power plant equipment, which can be used in the control server of the aforementioned photovoltaic power plant. Figure 2 This is a flowchart of a photovoltaic power plant equipment operation optimization method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0094] S201, based on real-time acquisition of total horizontal radiation data and direct surface radiation data of photovoltaic power station equipment, uses a preset first function to obtain the current radiation type index and operating mode type. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0095] S202, Based on the historical data of the current photovoltaic power station, the constructed random forest prediction model is trained to obtain the trained random forest prediction model;

[0096] Through the above implementation method, the temporal and spatial extended features of the radiation type index are obtained by utilizing historical environmental data. The random forest prediction model is then trained using the historical environmental data, temporal and spatial extended features. The trained forest prediction model can more accurately fit the key features in the environmental data, thereby making the predicted power generation more accurate and providing a reference for optimizing the operation of photovoltaic power plants.

[0097] S203, based on the current radiation type index and comprehensively acquired real-time environmental data, a trained random forest prediction model is used to predict the power generation at multiple preset time intervals and their respective confidence intervals. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0098] S204. Based on the aforementioned random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base. For details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0099] S205, based on the rule base, the current radiation type index, the operating mode type, and the predicted power generation, an intelligent decision-making model is used to make a decision and obtain the optimal equipment operation rules. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0100] Reference Figure 3 For example, step S202 includes:

[0101] S2021, based on the historical environmental data of the current photovoltaic power station, the temporal extension features and spatial extension features of the radiation type index are obtained using a preset feature extraction method. The temporal extension features include: the mean, variance, temperature change rate, and average irradiance intensity of the radiation type index in 1 hour; the spatial extension features include: the power difference coefficient between adjacent strings and the string-level radiation attenuation rate.

[0102] The preset feature extraction method can be implemented as a spatiotemporal feature extraction method to construct spatiotemporal relationships between historical environmental data. This facilitates the training of the random forest prediction model with historical environmental data after the spatiotemporal relationships are constructed, thereby maximizing the fit between the trained random forest prediction model and the historical environmental data and improving the prediction accuracy of the random forest prediction model.

[0103] S2022, Combining the historical environmental data, temporal extension features, and spatial extension features, input them into the random forest prediction model to be trained to obtain the predicted power generation.

[0104] Based on the acquired historical environmental data, temporal extended features, and spatial extended features, the constructed random forest prediction model is trained using these features. The random forest prediction model is then used to output the predicted power generation, which can be implemented as the power generation at the level of 15 minutes in the next day.

[0105] For example, the parameters of the random forest prediction model are set to 300 trees and 15 depths. Oversampling is performed by applying a 3x weight to data with a radiation type index greater than 0.7. The loss function for the dust storm period is weighted by mean square error, and the weight of the loss function for the dust storm period can be set to 2.5. The prediction results of the random forest prediction model are output in real time using an FPGA module, and the delay of the output prediction results is shortened to within 50ms.

[0106] S2023, based on actual and predicted power generation, uses feature importance ranking to filter the input environmental data and obtains several key feature variables.

[0107] like Figure 4 As shown, when the random forest prediction model predicts power generation, the results also include feature importance analysis results. These results contain important parameters that affect the prediction data. Therefore, by fitting the model with the features that rank highest in importance, the accuracy of the prediction results output by the random forest prediction model can be ensured. For example, the top five environmental data points in terms of importance can be dynamically selected every 6 hours, and these selected environmental data can be used as key feature variables to optimize the training of the random forest prediction model.

[0108] S2024, Based on the multiple key feature variables, the feature input weights of the random forest prediction model are optimized to obtain the optimized random forest prediction model, and the output is the trained random forest prediction model.

[0109] By utilizing key feature variables and optimizing the feature input weights of the random forest prediction model using feature importance ranking, environmental data with different importance rankings can be appropriately adjusted, thereby ensuring that the random forest prediction model can predict the power generation of photovoltaic power station equipment based on real-time environmental data.

[0110] The photovoltaic power station equipment operation optimization method provided in this embodiment utilizes the total horizontal radiation data and direct radiation data of the photovoltaic power station equipment to obtain a radiation type index, which is used to construct a quantitative model of the ratio of direct radiation to diffuse radiation, thereby facilitating the capture of radiation abrupt changes under sandstorm conditions during the actual operation of the photovoltaic power station. Then, by inputting the current radiation type index and real-time environmental data into the trained random forest prediction model, the limitation of traditional power prediction models relying solely on total radiation is overcome, thus obtaining a more accurate predicted power generation under sandstorm weather. By predicting the power generation, the current radiation type index, and the operating mode type, a smart decision-making model is used to make decisions and obtain the optimal equipment operation rules. The optimal equipment operation rules are then used to optimize the operating status of the photovoltaic power station equipment, thereby ensuring the efficient and reliable operation of the photovoltaic power station equipment.

[0111] Reference Figure 5 In another embodiment, the photovoltaic power plant equipment operation optimization method provided by this invention can be used in the control server of the aforementioned photovoltaic power plant, comprising:

[0112] S301, based on real-time acquisition of total horizontal radiation data and direct surface radiation data of photovoltaic power station equipment, uses a preset first function to obtain the current radiation type index and operating mode type. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0113] S302, based on the current radiation type index and comprehensively acquired real-time environmental data, a trained random forest prediction model is used to predict the power generation at multiple preset time intervals and their respective confidence intervals. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0114] S303, Based on the aforementioned random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base. For details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0115] S304, based on the rule base, the current radiation type index, the operating mode type, and the predicted power generation, an intelligent decision-making model is used to make a decision and obtain the optimal equipment operation rules. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0116] S305, Based on the predicted power generation and the actual power generation after implementing the optimal equipment operation rules, the random forest prediction model is evaluated using the method of calculating the root mean square error or residual distribution;

[0117] When the evaluation result is greater than a preset threshold, a conservative mode instruction is generated, which is used to adjust the photovoltaic power station equipment to a fixed tilt angle.

[0118] After optimizing the operating status of photovoltaic power station equipment using optimal equipment operation rules, the accuracy of the random forest prediction model can be determined by obtaining the current actual power generation and the actual power generation after executing the optimal equipment operation strategy. Then, the accuracy of the random forest prediction model is judged by using a preset threshold, and the photovoltaic power station equipment is adjusted to a fixed tilt angle when the random forest prediction model is inaccurate, so as to avoid the photovoltaic power station equipment operating under the intervention of the inaccurate random forest prediction model.

[0119] In addition, to ensure power generation, the conservative mode may also include triggering manual intervention, using staff to manually adjust the photovoltaic power station equipment.

[0120] For example, the following uses measured data from a certain region in November and December, and evaluates the random forest prediction model and the traditional LSTM model by obtaining the root mean square error. The evaluation results are shown in the table below:

[0121]

[0122] The root mean square error of traditional LSTM or random forest prediction models satisfies the following:

[0123] ,

[0124] In the formula, This represents the root mean square error of the traditional LSTM model or random forest prediction model. Indicates the first The actual value corresponding to each environmental data point; This indicates that the LSTM model or random forest prediction model is used to predict the first... The predicted value based on environmental data; Indicates the first Environmental data; This indicates the total number of environmental data.

[0125] like Figure 6 As shown, a scatter plot is constructed based on the predicted power generation and actual power generation in the above embodiments. Each point represents a pairing of model prediction and actual observation. The closer the scatter points in the scatter plot are to the diagonal, the more accurate the model prediction is. According to the information in the scatter plot, it can be seen that the predicted power generation obtained by the random forest prediction model provided in this embodiment of the invention has a high degree of fit with the actual power generation.

[0126] Furthermore, when evaluating the random forest prediction model using the residual distribution calculation method of this invention, based on the predicted power generation and actual power generation in the above embodiments, the residual distribution calculation method can yield the following results: Figure 7 The residual distribution diagram is shown below. Figure 7 The residuals approximately follow a normal distribution with a mean of -0.12% and a standard deviation of 2.3%. Meanwhile, 95% of the residuals are concentrated within the range of ±4.5%, while extreme errors (>±10%) account for 0.7%. Through comprehensive analysis of historical environmental data, it was found that the extreme errors mainly occurred in the early stages of sandstorms, which was caused by the delay when the photovoltaic power station equipment switched working modes according to the radiation type index.

[0127] Since sandstorm conditions reduce the light transmittance of photovoltaic power station equipment surfaces, affecting the power generation capacity of the photovoltaic power generation equipment in this invention, this invention also provides a photovoltaic power station equipment operation optimization method to reduce the occurrence of the above situation. This method can be used in the control server of the aforementioned photovoltaic power station, and includes:

[0128] S401, based on real-time acquisition of total horizontal radiation data and direct surface radiation data of photovoltaic power station equipment, uses a preset first function to obtain the current radiation type index and operating mode type. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0129] S402, based on the current radiation type index and comprehensively acquired real-time environmental data, a trained random forest prediction model is used to predict the power generation at multiple preset time intervals and their respective confidence intervals. For details, please refer to [link to relevant documentation]. Figure 1 Step S102 of the illustrated embodiment will not be described again here.

[0130] S403, Based on the aforementioned random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base. For details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0131] S404, based on the rule base, the current radiation type index, the operating mode type, and the predicted power generation, an intelligent decision-making model is used to make a decision and obtain the optimal equipment operation rules. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0132] S405, based on the current radiation type index and the historical radiation type index, the radiation type index decay rate is obtained using a preset second function, wherein the second function includes the functional relationship between the current radiation type index, the historical radiation type index and the radiation type index decay rate.

[0133] For example, the radiation type index decay rate satisfies:

[0134] ,

[0135] in, Indicates the radiation type index decay rate; express Radiation type index at any given time; express Historical radiation type index at any given time.

[0136] In this embodiment, This indicates the current radiation type index, and the time resolution of the radiation type index is 15 minutes. This is expressed as a time resolution of 4 periods, therefore... This indicates the radiation type index from one hour ago. This system is used to monitor rapid changes in external cloud cover, providing data for subsequent equipment cleaning in photovoltaic power plants. When |ΔRTI|>0.1%, it indicates that the cloud cover is changing rapidly or dust is forming, requiring an alert to remind staff to prepare.

[0137] Prolonged sandstorms can cause photovoltaic power station equipment to become covered in dust, leading to a decrease in power generation. By utilizing the radiation type index decay rate to provide early warning of rapid changes in cloud cover in the external environment, the probability of actual power generation being affected by sudden changes in the external environment can be reduced.

[0138] S406, based on the predicted power generation and theoretical power generation, a preset third function is used to obtain the predicted power loss, wherein the third function includes the functional relationship between the predicted power generation, theoretical power generation and predicted power loss.

[0139] For example, the predicted power loss can be expressed as:

[0140] ,

[0141] in, Indicates the predicted power loss; Indicates theoretical power generation capacity; This indicates the predicted power generation capacity.

[0142] In this embodiment, This represents the power generation forecast curve for the next hour; This represents the theoretical maximum value based on the nominal efficiency of photovoltaic power plant modules and the current real-time radiation.

[0143] When the predicted power loss exceeds a preset threshold, it indicates that the current photovoltaic power station equipment is significantly affected by dust on its surface, requiring timely cleaning. Through the above implementation method, the predicted power loss is obtained using both predicted and theoretical power generation. This predicted power loss is then used to evaluate the power generation efficiency of the photovoltaic power station equipment, facilitating the prioritization of cleaning for equipment with high predicted power losses and improving the overall cleaning efficiency.

[0144] S407. Based on the radiation type exponential decay rate and the predicted power loss, the random forest algorithm is used for evaluation to obtain the cleaning priority and cleaning probability.

[0145] By utilizing the radiation type exponential decay rate and predicted power loss, and evaluating them using the random forest algorithm, we can obtain the cleaning priority and cleaning efficiency of photovoltaic power station equipment. The cleaning priority and cleaning efficiency are used as a reference for the cleaning work of photovoltaic power station equipment, so as to carry out the cleaning work of photovoltaic power station equipment while ensuring power generation.

[0146] S408, based on the cleaning priority and cleaning probability, a preset cleaning task triggering threshold is used for comparison. When the cleaning probability is greater than the cleaning task triggering threshold, the photovoltaic power station equipment with high cleaning priority is cleaned.

[0147] By utilizing power prediction results to drive tilt adjustment and cleaning scheduling in real time, the control closed loop of photovoltaic power station equipment is ensured. Based on the radiation type exponential decay rate and predicted power loss, the random forest algorithm is used for evaluation to obtain the cleaning priority and cleaning probability of the photovoltaic power station equipment surface. When the cleaning probability is greater than the cleaning trigger threshold, the photovoltaic power station equipment with high cleaning priority is cleaned, thereby reducing the defect of reduced power generation efficiency of photovoltaic power station equipment due to dust cover.

[0148] This embodiment also provides a photovoltaic power plant equipment operation optimization system, which is used to implement the above embodiments and preferred embodiments, and will not be repeated for details already described. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0149] This embodiment provides a photovoltaic power plant equipment operation optimization system, the system comprising:

[0150] The radiation type index generation module 510 is used to obtain the current radiation type index and working mode type based on the real-time acquisition of the total horizontal radiation data and the direct radiation data of the component surface of the photovoltaic power station equipment, using a preset first function. The first function includes the functional relationship between the total horizontal radiation data, the direct radiation data of the component surface, and the radiation type index.

[0151] The power generation prediction module 520 is used to predict the power generation at multiple preset time intervals and their respective confidence intervals based on the current radiation type index, the comprehensive real-time environmental data obtained, and the trained random forest prediction model.

[0152] The rule base construction module 530 is used to obtain multiple interpretable rules based on the random forest prediction model and the random forest rule extraction method, and to integrate them into a rule base.

[0153] The decision acquisition module 540 is used to make decisions based on the rule base, the current radiation type index, the working mode type and the predicted power generation, using an intelligent decision model to obtain the optimal equipment operation rules. The optimal equipment operation rules are used to generate instructions to adjust the operating tilt angle of the photovoltaic power station equipment.

[0154] In some optional implementations, the system further includes a prediction model training module 550, which includes:

[0155] The feature extraction unit 5501 is used to obtain the temporal extension features and spatial extension features of the radiation type index based on the historical environmental data of the current photovoltaic power station and using a preset feature extraction method. The temporal extension features include: the mean, variance, temperature change rate, and average irradiance intensity of the radiation type index in 1 hour; the spatial extension features include: the power difference coefficient between adjacent strings and the string-level radiation attenuation rate.

[0156] The model prediction unit 5502 is used to integrate the historical environmental data, temporal extended features and spatial extended features, input them into the random forest prediction model to be trained, and obtain the predicted power generation.

[0157] The variable filtering unit 5503 is used to filter the input environmental data based on the actual power generation and the predicted power generation, and to obtain multiple key feature variables by ranking the feature importance.

[0158] The model optimization unit 5504 is used to optimize the feature input weights of the random forest prediction model based on the multiple key feature variables, so as to obtain the optimized random forest prediction model and output the trained random forest prediction model.

[0159] In some embodiments, the radiation type index output by the radiation type index generation module 510 satisfies the following:

[0160] ,

[0161] in, Indicates the radiation type index; This represents the direct radiation data on the component surface; This represents the total radiation data on the horizontal plane; This represents a constant that prevents division by zero.

[0162] In some optional embodiments, the system further includes a cleaning scheduling module 560, which includes:

[0163] The first parameter acquisition unit 5601 is used to obtain the radiation type index decay rate based on the current radiation type index and the historical radiation type index using a preset second function. The second function includes the functional relationship between the current radiation type index, the historical radiation type index and the radiation type index decay rate.

[0164] The second parameter acquisition unit 5602 is used to obtain the predicted power loss based on the predicted power generation and theoretical power generation using a preset third function. The third function includes the functional relationship between the predicted power generation, theoretical power generation and predicted power loss.

[0165] The cleaning assessment unit 5603 is used to evaluate the cleaning priority and cleaning probability based on the radiation type exponential decay rate and the predicted power loss using a random forest algorithm.

[0166] The cleaning execution unit 5604 is used to compare the cleaning priority and cleaning probability using a preset cleaning task triggering threshold. When the cleaning probability is greater than the cleaning task triggering threshold, the photovoltaic power station equipment with high cleaning priority is cleaned.

[0167] In some embodiments, the radiation type exponential decay rate output by the first parameter acquisition module satisfies the following:

[0168] ,

[0169] in, Indicates the radiation type index decay rate; express Radiation type index at any given time; express Historical radiation type index at any given time.

[0170] In some embodiments, the system provided by the present invention further includes:

[0171] The model evaluation module 570 is used to evaluate the random forest prediction model based on the predicted power generation and the actual power generation after executing the optimal equipment operation rules, using the method of calculating the root mean square error or residual distribution.

[0172] When the evaluation result is greater than a preset threshold, a conservative mode instruction is generated, which is used to adjust the photovoltaic power station equipment to a fixed tilt angle.

[0173] In some embodiments, the system provided by the present invention further includes:

[0174] The data monitoring and processing module 580 is used to monitor the received real-time environmental data. When there is data loss in the acquired real-time environmental data, the real-time environmental data is processed by using the nearest neighbor array interpolation method.

[0175] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0176] In this embodiment, the photovoltaic power plant equipment operation optimization system is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0177] This invention also provides a computer device having the above-described features. Figure 8 The photovoltaic power plant equipment operation optimization system shown.

[0178] Please see Figure 9 , Figure 9 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 9 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 9 Take a processor 10 as an example.

[0179] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0180] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0181] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0182] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0183] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0184] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0185] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0186] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for optimizing the operation of photovoltaic power plant equipment, characterized in that, The method includes: Based on real-time acquisition of total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, the current radiation type index and working mode type are obtained using a preset first function. The first function includes the functional relationship between total horizontal radiation data, direct radiation data of the module surface, and radiation type index. Based on the current radiation type index, and by comprehensively acquiring real-time environmental data, a trained random forest prediction model is used to make predictions, resulting in predicted power generation at multiple preset time intervals and their respective confidence intervals. Based on the aforementioned random forest prediction model, multiple interpretable rules are obtained using the random forest rule extraction method, and then integrated into a rule base. Based on the rule base, the current radiation type index, the operating mode type, and the predicted power generation, an intelligent decision-making model is used to make decisions and obtain the optimal equipment operation rules. The optimal equipment operation rules are used to generate instructions to adjust the tilt angle of the photovoltaic power station equipment.

2. The method according to claim 1, characterized in that, The training of the random forest prediction model includes: Based on the historical environmental data of the current photovoltaic power station, the temporal extension features and spatial extension features of the radiation type index are obtained using a preset feature extraction method. The temporal extension features include: the mean, variance, temperature change rate of the radiation type index over 1 hour, and the average irradiance intensity over the previous 3 hours. The spatial extension features include: the power difference coefficient between adjacent strings and the string-level radiation attenuation rate. By combining the historical environmental data, temporal extension features, and spatial extension features, the data is input into the random forest prediction model to be trained to obtain the predicted power generation. Based on actual and predicted power generation, the input environmental data is filtered by ranking the importance of features to obtain several key feature variables. Based on the aforementioned key feature variables, the feature input weights of the random forest prediction model are optimized to obtain the optimized random forest prediction model, and the output is the trained random forest prediction model.

3. The method according to claim 1 or 2, characterized in that, The radiation type index satisfies the following: ， in, Indicates the radiation type index; This represents the direct radiation data on the component surface; This represents the total radiation data on the horizontal plane; This represents a constant that prevents division by zero.

4. The method according to claim 1 or 2, characterized in that, The method further includes: Based on the current radiation type index and the historical radiation type index, the radiation type index decay rate is obtained using a preset second function. The second function includes the functional relationship between the current radiation type index, the historical radiation type index, and the radiation type index decay rate. Based on the predicted power generation and theoretical power generation, the predicted power loss is obtained using a preset third function, which includes the functional relationship between the predicted power generation, theoretical power generation, and predicted power loss. Based on the radiation type exponential decay rate and the predicted power loss, the random forest algorithm is used to evaluate and obtain the cleaning priority and cleaning probability. Based on the cleaning priority and cleaning probability, a preset cleaning task triggering threshold is used for comparison. When the cleaning probability is greater than the cleaning task triggering threshold, the photovoltaic power station equipment with high cleaning priority is cleaned.

5. The method according to claim 4, characterized in that, The radiation type index attenuation rate satisfies the following: ， in, Indicates the radiation type index decay rate; express Radiation type index at any given time; express Historical radiation type index at any given time.

6. The method according to claim 1 or 2, characterized in that, The method further includes: Based on the predicted power generation and the actual power generation after implementing the optimal equipment operation rules, the random forest prediction model is evaluated using the method of calculating the root mean square error or residual distribution. When the evaluation result is greater than a preset threshold, a conservative mode instruction is generated, which is used to adjust the photovoltaic power station equipment to a fixed tilt angle.

7. The method according to claim 1 or 2, characterized in that, The method further includes: The received real-time environmental data is monitored. When data loss occurs in the acquired real-time environmental data, the real-time environmental data is processed using a neighbor array interpolation method.

8. A photovoltaic power plant equipment operation optimization system, characterized in that, The system includes: The radiation type index generation module, based on real-time acquisition of total horizontal radiation data and direct radiation data of the photovoltaic power station equipment, uses a preset first function to obtain the current radiation type index and working mode type. The first function includes the functional relationship between total horizontal radiation data, direct radiation data of the component surface, and radiation type index. The power generation prediction module is used to make predictions based on the current radiation type index, the comprehensive real-time environmental data obtained, and the trained random forest prediction model to obtain the predicted power generation and their respective confidence intervals for multiple preset time intervals. The rule base construction module is used to obtain multiple interpretable rules based on the random forest prediction model using the random forest rule extraction method, and integrate them into a rule base; The decision acquisition module is used to make decisions based on the rule base, the current radiation type index, the working mode type and the predicted power generation, using an intelligent decision model to obtain the optimal equipment operation rules. The optimal equipment operation rules are used to generate instructions to adjust the operating tilt angle of the photovoltaic power station equipment.

9. A computer device, characterized in that, include: A memory and a processor are interconnected, the memory stores computer instructions, and the processor executes the computer instructions to perform the photovoltaic power plant equipment operation optimization method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the photovoltaic power plant equipment operation optimization method according to any one of claims 1 to 7.

Images