Photovoltaic power station asset assessment method and system based on AI large model application

Abstract

The invention relates to the technical field of photovoltaic power station asset assessment, and relates to a photovoltaic power station asset assessment method and system based on AI large model application, and the method comprises the steps: obtaining first information, second information and third information, the first information comprises photovoltaic power station operation data, the second information comprises a photovoltaic power station equipment image, and the third information comprises photovoltaic power station operation data; the third information comprises an equipment fault log and real-time meteorological data; training an AI model of the local end by using the first information and the second information to obtain a trained AI model; performing federated learning on the AI model trained at the local end by using a preset federated learning framework to generate a global AI model; optimizing the global AI model by using the third information to obtain an optimized global AI model; according to the photovoltaic power station asset evaluation method and system, assets of a plurality of photovoltaic power stations can be evaluated at the same time, meanwhile, the dynamic asset evaluation report is updated and generated in real time, and the evaluation precision and the evaluation efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station asset evaluation, and in particular to a photovoltaic power station asset evaluation method and system based on AI large model application. Background Art

[0002] The asset evaluation of photovoltaic power stations in the industry has long relied on traditional evaluation methods, which deeply integrate industry experience, financial models and physical models. However, these traditional evaluation methods have significant drawbacks. They require a large amount of historical data and market data, and the data quality directly affects the evaluation results. There is a problem of strong data dependence, but in actual operation, the data quality is uneven, and data missing, errors or lack of timeliness are common. These defects will directly transmit to the evaluation results, seriously affecting their accuracy and reliability. Summary of the invention

[0003] The purpose of the present invention is to provide a photovoltaic power station asset evaluation method and system based on the application of AI large model to improve the above problems.

[0004] In order to achieve the above objectives, the present application provides the following technical solutions: On the one hand, an embodiment of the present application provides a photovoltaic power station asset evaluation method based on the application of an AI large model, the method comprising: Acquire first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; Using the first information and the second information to train a local AI model to obtain a trained AI model, wherein the trained AI model is used to evaluate the health of photovoltaic power station equipment; Use the preset federated learning framework to conduct federated learning on the AI ​​model trained locally to generate a global AI model; Optimizing the global AI model using the third information to obtain an optimized global AI model; An asset evaluation report for a photovoltaic power station is obtained based on the optimized global AI model.

[0005] In a second aspect, an embodiment of the present application provides a photovoltaic power station asset assessment system based on an AI large model application, the system comprising: an acquisition module, configured to acquire first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; A first processing module, configured to train a local AI model using the first information and the second information to obtain a trained AI model, wherein the trained AI model is used to evaluate the health of photovoltaic power station equipment; The second processing module is used to perform federated learning on the AI ​​model trained on the local end using a preset federated learning framework to generate a global AI model; A third processing module, configured to optimize the global AI model using the third information to obtain an optimized global AI model; An evaluation module is used to obtain a photovoltaic power station asset evaluation report based on the optimized global AI model.

[0006] In a third aspect, an embodiment of the present application provides a photovoltaic power station asset assessment device based on the application of an AI large model, the device comprising a memory and a processor. The memory is used to store a computer program; the processor is used to implement the steps of the photovoltaic power station asset assessment method based on the application of an AI large model when executing the computer program.

[0007] In a fourth aspect, an embodiment of the present application provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned photovoltaic power station asset assessment method based on the application of the AI ​​large model are implemented.

[0008] The beneficial effects of the present invention are: The present invention deploys a lightweight AI model locally in the photovoltaic power station to reduce dependence on cloud computing power, and then uses the locally deployed lightweight AI model for federated learning to obtain a global AI model, which solves the data island problem and improves the model generalization capability. The global AI model is optimized using third-party information to improve the accuracy of the model prediction results, and then a photovoltaic power station asset evaluation report is obtained based on the optimized global AI model, thereby improving the evaluation efficiency of photovoltaic power station assets.

[0009] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 It is a flow chart of the photovoltaic power station asset assessment method based on the application of the AI ​​big model described in an embodiment of the present invention.

[0012] Figure 2 It is a schematic diagram of the structure of the photovoltaic power station asset assessment system based on the AI ​​large model application described in an embodiment of the present invention.

[0013] Figure 3 It is a schematic diagram of the structure of the photovoltaic power station asset assessment equipment based on the AI ​​large model application described in an embodiment of the present invention.

[0014] Labels in the figure: 800, photovoltaic power station asset assessment equipment based on AI large model application; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component; 901, acquisition module; 902, first processing module; 903, second processing module; 904, third processing module; 905, assessment module. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0016] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance. Embodiment 1:

[0017] This embodiment provides a photovoltaic power station asset assessment method based on the application of an AI large model. It can be understood that in this embodiment, a scenario can be laid out, for example: using sensors to collect photovoltaic power station operation data and using drones to collect photovoltaic power station equipment images to conduct asset assessment of photovoltaic power stations.

[0018] See also Figure 1, the figure shows that the method includes step S1, step S2, step S3, step S4 and step S5, which specifically include: Step S1, obtaining first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; In this step, the operation data of the photovoltaic power station includes power generation, voltage, current and temperature data of the equipment, and the images of the photovoltaic power station equipment include photos taken by drones of bracket rust, component hot spots, inverter appearance, etc.

[0019] The step S1 also includes step S11, step S12, step S13, step S14 and step S15, which specifically include: Step S11, acquiring modeling information, wherein the modeling information includes location information of the photovoltaic power station, location information of equipment to be inspected in the photovoltaic power station, and terrain data; In this step, the modeling information is obtained by using a UAV equipped with a laser radar or a multi-view camera.

[0020] Step S12: establishing a three-dimensional model of a photovoltaic power station according to the modeling information; In this step, the location and height of key equipment (PV panels, inverters, combiner boxes, brackets, etc.) are marked, and the PV power station location information, equipment distribution, terrain data, etc. are integrated into the GIS platform to establish a 3D model of the PV power station.

[0021] Step S13, obtaining the type of equipment to be inspected; Step S14, planning an inspection route in the three-dimensional model of the photovoltaic power station according to the type of equipment to be inspected; In this step, for the inverter, take off from a nearby take-off and landing point, fly around the inverter for a week, and maintain the safe distance threshold for appearance inspection; for the box-type transformer, take a picture of the overall appearance, and then lower the height safety distance threshold to inspect key components. When inspecting the collector line, fly along the line direction, and use a high-definition camera and laser rangefinder to inspect the line components and measure parameters such as sag at intervals of the safety distance threshold. It should be noted that the safety distance threshold is preset manually.

[0022] Step S15: Acquire second information according to the inspection route.

[0023] In this step, the inspection route is determined according to the different types of equipment, so that the captured second information can better reflect the corresponding characteristics of the equipment and improve the data quality.

[0024] Step S2: using the first information and the second information to train the local AI model to obtain a trained AI model, wherein the trained AI model is used to evaluate the health of photovoltaic power station equipment; In this step, a lightweight AI model is deployed locally in the PV power station to perform preliminary inference on equipment health assessment, reduce reliance on cloud computing power, and reduce data latency to improve the accuracy of the generated dynamic asset assessment report.

[0025] The step S2 also includes step S21, step S22, step S23, step S24 and step S25, which specifically include: Step S21, preprocessing the first information to obtain preprocessed first information; In this step, the collected first information is denoised by using a sliding window mean filter to eliminate instantaneous noise, and then the data is normalized to the interval [0,1].

[0026] Step S22, performing standardization processing on the second information to obtain standardized second information; In this step, Gaussian blur is applied to the second information to reduce the impact of shooting shake, the pixel values ​​are scaled to [-1, 1] for standardization, and each device image included in the second information is cropped to the same size.

[0027] Step S23, obtaining a preset first neural network model and a second neural network model; In this step, the preset first neural network model selects the multi-layer perceptron (MLP) model, which can process numerical equipment operating parameters and map the complex relationship between learning parameters to the health status of the equipment. The second neural network model selects the lightweight neural network MobileNetV3 as the basic model for image feature extraction.

[0028] Step S24, fusing the first neural network model and the second neural network model to obtain a fused neural network model; In this step, the MLP model that processes the device operating parameters and the MobileNetV3 model that processes the image data are fused. The features of the two can be concatenated in the later stage of the model, and then further feature fusion and processing are performed through the fully connected layer to output the device health score.

[0029] Step S25: Use the preprocessed first information and the standardized second information to train the fused neural network model to obtain a locally trained AI model.

[0030] In this step, using data to train a neural network is a technical solution well known to those skilled in the art, so it will not be described here in detail.

[0031] Step S3: Use a preset federated learning framework to perform federated learning on the AI ​​model trained on the local end to generate a global AI model; The step S3 also includes step S31, step S32, step S33 and step S34, which specifically include: Step S31, obtaining the gradient parameters generated by local training of each node; In this step, a federated learning network is established to connect the local AI models of each photovoltaic power station. In this federated learning network, there is a coordination server to coordinate the communication between each node and the interaction of model parameters. The local AI model of each photovoltaic power station uses its own lightweight model for training on local data. During the training process, the model will calculate the gradient parameters, which can reflect the learning status of the model on the local data.

[0032] Step S32, encrypting the gradient parameter using an encryption algorithm to obtain an encrypted gradient parameter; In this step, the encryption algorithm is not limited. Through encryption, even if the gradient parameters are obtained by a third party during the transmission process, the original data cannot be directly restored from these ciphertexts.

[0033] Step S33: sending the encrypted gradient parameters to a coordination server in a federated learning framework to aggregate global model parameters; Step S34: Use differential privacy technology to add noise when updating the global model to obtain an updated global AI model.

[0034] In this step, during the global model update process, differential privacy technology is introduced to add a certain amount of noise when the model parameters are updated. Even if the attacker attempts to obtain sensitive information from the model parameters through reverse engineering and other means, it is difficult to accurately restore the original data due to the presence of noise, thereby further preventing the leakage of sensitive information. After the updated global AI model is obtained, it will be sent to each node.

[0035] In this embodiment, federated learning is performed on the AI ​​model trained locally, which effectively solves the problem of data islands in multiple power stations, improves the generalization ability of the model, and complies with data privacy regulations.

[0036] Step S4: optimizing the global AI model using the third information to obtain an optimized global AI model; The step S4 also includes step S41, step S42, step S43, step S44 and step S45, which specifically include: Step S41, obtaining a historical data set, wherein the historical data set includes a large amount of operating data under normal conditions collected in a historical operation database; In this step, a large amount of data under normal working conditions is collected from the historical operation database of the photovoltaic power station. These data cover various parameters related to the operation of the power station, such as light intensity, temperature, photovoltaic panel output voltage and current, inverter working status parameters, etc. in different time periods. Ensure the integrity and accuracy of the data, and clean up and correct missing values ​​and erroneous values.

[0037] Step S42: training the autoencoder model using the historical data set to obtain a trained autoencoder model; In this step, the network structure of the autoencoder is designed, which usually consists of two parts: the encoder and the decoder. The encoder compresses the input high-dimensional data into a low-dimensional representation (encoded vector), and the decoder restores the encoded vector to high-dimensional data. A specific implementation method is to use a multi-layer perceptron (MLP) structure as the basic architecture of the autoencoder. The encoder part can be composed of multiple fully connected layers, and the number of neurons in each fully connected layer gradually decreases. For example, the number of neurons in the input layer is the same as the number of input data features. After several hidden layers, the number of neurons in the output layer is the dimension of the encoded vector. The decoder is a mirror structure of the encoder. The number of neurons gradually increases from the dimension of the encoded vector, and finally the number of neurons in the output layer is consistent with the number of input data features.

[0038] Step S43, sending the third information to the trained autoencoder model to obtain an abnormal data recognition result; In this step, the real-time data is input into the trained autoencoder model, which encodes the data to obtain a low-dimensional representation, and then decodes and restores it. The reconstruction error between the original input data and the decoded data is calculated. The reconstruction error is compared with a pre-set threshold, and the data with a reconstruction error greater than the pre-set threshold is determined as abnormal data. It can be understood that the threshold is obtained through experiments and statistical analysis on a large amount of historical normal data.

[0039] Step S44, filtering the third information according to the abnormal data identification result to obtain a filtered data set; In this step, due to the influence of sensor failure or communication interference, the collected data contains abnormal data, resulting in low quality of the data set. Therefore, the abnormal data in the third information is filtered out through the abnormal data identification result to obtain a filtered data set, which effectively improves the quality of the data.

[0040] Step S45: After sending the filtered data set to the global AI model, the global AI model is optimized using an incremental learning algorithm.

[0041] In this step, a variant of the online gradient descent algorithm is used to calculate the gradient for the filtered data set. According to the preset learning rate, the calculated gradient is applied to the existing model parameters to gradually update the model. The learning rate can be adjusted using an adaptive strategy, such as dynamic adjustment based on the changing trend of the data or the convergence of the model to balance the speed and stability of model learning.

[0042] In this embodiment, the input is a global AI model that has undergone federated learning, as well as new data such as real-time weather changes and equipment failure events. Through the incremental learning algorithm, the model can be dynamically updated when receiving new data, and low-quality data or noise interference can be filtered out in combination with the abnormal data recognition results, and the output is an optimized model that can adapt to changes in real-time data streams. During the processing, the model is continuously updated according to new data, reducing dependence on historical data and improving robustness to rare scenarios.

[0043] Step S5: obtaining a photovoltaic power station asset assessment report based on the optimized global AI model.

[0044] The step S5 also includes step S51, step S52, step S53 and step S54, which specifically include: Step S51: sending the first information, the second information and the third information to the optimized global AI model to obtain a prediction result, wherein the prediction result includes the health of the photovoltaic power station equipment and the power generation; Step S52: constructing a scatter plot according to the prediction result, wherein the scatter plot includes a correlation distribution between equipment health and power generation efficiency; In this step, a two-dimensional coordinate system is established, with the equipment health as the horizontal coordinate of the scatter plot and the power generation efficiency as the vertical coordinate of the scatter plot. According to the values ​​of equipment health and power generation efficiency, a point is drawn for each device in the two-dimensional coordinate system. The position of each point is determined by its corresponding equipment health and power generation efficiency values.

[0045] Step S53: determining the correlation between the equipment health and the power generation efficiency according to the scatter plot; In this step, the scatter plot shows that the equipment health is positively correlated with the power generation efficiency. That is, the higher the equipment health, the higher the power generation efficiency; the lower the equipment health, the lower the power generation efficiency. For example, when the key components of the equipment are less aged and well maintained, the equipment can operate more efficiently, thereby improving the power generation efficiency.

[0046] Step S54: Generate a photovoltaic power station asset assessment report based on the correlation between the equipment health and power generation efficiency.

[0047] The step S54 also includes step S541, step S542, step S543 and step S544, which specifically include: Step S541, determining a calculation formula for power generation efficiency according to the correlation between the device health and power generation efficiency of each device; In this step, the impact of the equipment health of each device on the power generation efficiency can be determined based on the correlation between the equipment health and the power generation efficiency of each device. The hot spots of photovoltaic modules in a power station account for 5%, resulting in a 4% loss in power generation efficiency. Therefore, the calculation formula for power generation efficiency is:

[0048] In the above formula, E represents the power generation efficiency, A, B, and C represent the equipment health of the photovoltaic module equipment, the equipment health of the inverter equipment, and the equipment health of the bracket, respectively. , and They represent the corresponding weight parameters respectively, b is the error term. It should be noted that the weight parameters of photovoltaic module equipment account for a relatively large proportion, usually 0.8-0.85, the weight of inverter equipment accounts for 0.05-0.1, and the weight parameters of bracket equipment account for 0.05-0.03.

[0049] Step S542, determining the corrected power generation according to the calculation formula; In this step, the loss efficiency can be determined according to the calculation formula, and the lost power generation can be determined according to the loss efficiency, so that the power generation can be corrected according to the lost power generation.

[0050] Step S543, obtaining regional electricity price information; Step S544: Evaluate the photovoltaic power station assets according to the regional electricity price information and the corrected power generation to obtain an asset evaluation report.

[0051] It should be noted that the asset appraisal report also includes real-time operating data, real-time power generation, equipment health score and other indicators; equipment health analysis: a list of high-risk equipment and a heat map showing the distribution of equipment defects; power generation fluctuation attribution analysis: quantifying the contribution of different factors to power generation deviations (such as "shadowing caused an 8% decrease in power generation this month").

[0052] After step S544, the method further includes step S5441, step S5442, step S5443 and step S5444, which specifically include: Step S5441: Generate a maintenance work order according to the equipment health corresponding to each equipment; In this step, when the health of the device is less than the preset health threshold, it is screened and added to the maintenance work order. The maintenance work order includes the maintenance type, priority, recommended execution time, and required resources. It should be noted that the priority is determined based on the risk score. The specific calculation process of the risk score is as follows: ;

[0053] In the above formula, S represents the risk score, and They represent the health weight and fault frequency weight respectively, and A and D represent the equipment health and fault frequency respectively.

[0054] Step S5442: After the equipment is maintained according to the maintenance work order, the photovoltaic power station operation data after the equipment maintenance is obtained; In this step, the operation and maintenance personnel scan the equipment QR code through the mobile APP, record the maintenance operation, upload the comparison photos before and after the maintenance is completed, and then collect real-time operation data.

[0055] Step S5443, determining the asset valuation after maintenance according to the photovoltaic power station operation data after the equipment maintenance; In this step, the health of the equipment can be re-evaluated based on the operating data of the PV power station after equipment maintenance and the photos after maintenance uploaded by the operation and maintenance personnel, and the power generation efficiency can be corrected based on the re-evaluated equipment health, thereby re-valuing the assets of the PV power station after maintenance.

[0056] Step S5444: Generate an asset disposal strategy based on the maintained asset valuation and external market data.

[0057] In this step, the collected asset valuation, electricity price and carbon trading data are first standardized to have a unified dimension and comparable basis. For example, asset valuation, electricity price and carbon trading income are converted into monetary values ​​in annual units. Then, the correlation between the data is established to clarify the comprehensive impact path of asset valuation changes, electricity price fluctuations and carbon trading policy adjustments on power plant operating income.

[0058] Based on the integrated data, we reconstruct the asset disposal strategy income simulation model. For the technology transformation strategy, we consider the investment cost of technology transformation, the improvement of power generation efficiency brought by the asset appreciation after technology transformation, and the changes in electricity price income and carbon trading income corresponding to the increase in power generation. For the sales strategy, we predict the sales income that can be obtained at different sales time points combined with the asset valuation, market electricity price and carbon trading policy environment at that time. For the long-term holding strategy, we simulate the power generation income and carbon trading income generated by the continuous operation of the power station in the future as the electricity price fluctuates and the carbon trading policy changes, and consider the impact of asset maintenance costs on income.

[0059] Finally, the Monte Carlo model was used to simulate and generate 10,000 market scenarios (including a variety of different market scenarios, such as electricity price fluctuations, policy changes, and abnormal weather), and the net present value and internal rate of return of assets in each scenario were calculated. The performance of different asset disposal strategies in each simulation scenario was compared and analyzed based on the net present value and internal rate of return of assets. By comprehensively weighing the advantages and disadvantages of each strategy, the optimal asset disposal strategy under the current market environment and the actual situation of the power plant was selected. A specific implementation method is: in the current market, electricity prices fluctuate frequently and carbon trading policies have large subsidies for efficient power generation. After evaluation, it was found that the technical transformation strategy can achieve the highest net present value of assets and a better internal rate of return in the long term, while having little impact on operational stability. The technical transformation strategy is determined as the optimal strategy.

[0060] In this embodiment, the stability of power station operation is maintained by a closed-loop processing flow of assessment-maintenance-reassessment. At the same time, by simulating the changes in asset value under different maintenance measures and asset disposal strategies, the solution that is most conducive to improving asset value under the current market environment and power station operating conditions can be accurately identified, thereby achieving the purpose of dynamically improving asset value, providing comprehensive and accurate data support and decision-making basis for asset management, and realizing refined asset management of photovoltaic power stations. Embodiment 2:

[0061] like Figure 2 As shown, this embodiment provides a photovoltaic power station asset evaluation system based on AI large model application, the system includes an acquisition module 901, a first processing module 902, a second processing module 903, a third processing module 904 and an evaluation module 905, which specifically include: An acquisition module 901 is used to acquire first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; A first processing module 902 is used to train a local AI model using the first information and the second information to obtain a trained AI model, where the trained AI model is used to evaluate the health of photovoltaic power station equipment; The second processing module 903 is used to perform federated learning on the AI ​​model trained on the local end by using a preset federated learning framework to generate a global AI model; A third processing module 904 is used to optimize the global AI model using the third information to obtain an optimized global AI model; The evaluation module 905 is used to obtain a photovoltaic power station asset evaluation report based on the optimized global AI model.

[0062] In a specific implementation of the present disclosure, the acquisition module includes a first acquisition unit, a first processing unit, a second acquisition unit, a second processing unit, and a third processing unit, which specifically include: A first acquisition unit is used to acquire modeling information, wherein the modeling information includes location information of the photovoltaic power station, location information of equipment to be inspected in the photovoltaic power station, and terrain data; A first processing unit, configured to establish a three-dimensional model of a photovoltaic power station according to the modeling information; The second acquisition unit is used to acquire the type of equipment to be inspected; A second processing unit, configured to plan an inspection route in the three-dimensional model of the photovoltaic power station according to the type of equipment to be inspected; The third processing unit is used to obtain second information according to the inspection route.

[0063] In a specific embodiment of the present disclosure, the first processing module includes a fourth processing unit, a fifth processing unit, a third acquisition unit, a sixth processing unit and a seventh processing unit, which specifically include: a fourth processing unit, configured to preprocess the first information to obtain preprocessed first information; a fifth processing unit, configured to perform standardization processing on the second information to obtain standardized second information; A third acquisition unit, used to acquire a preset first neural network model and a second neural network model; a sixth processing unit, configured to fuse the first neural network model and the second neural network model to obtain a fused neural network model; The seventh processing unit is used to train the fused neural network model using the preprocessed first information and the standardized second information to obtain the locally trained AI model.

[0064] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here. Embodiment 3:

[0065] Corresponding to the above method embodiment, this embodiment also provides a photovoltaic power station asset assessment device based on the application of AI large model. The photovoltaic power station asset assessment device based on the application of AI large model described below and the photovoltaic power station asset assessment method based on the application of AI large model described above can be referenced to each other.

[0066] Figure 38 is a block diagram of a photovoltaic power station asset evaluation device 800 based on an AI large model application according to an exemplary embodiment. Figure 3 As shown, the photovoltaic power station asset assessment device 800 based on the AI ​​large model application may include: a processor 801, a memory 802. The photovoltaic power station asset assessment device 800 based on the AI ​​large model application may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0067] Among them, the processor 801 is used to control the overall operation of the photovoltaic power station asset assessment device 800 based on the AI ​​large model application to complete all or part of the steps in the above-mentioned photovoltaic power station asset assessment method based on the AI ​​large model application. The memory 802 is used to store various types of data to support the operation of the photovoltaic power station asset assessment device 800 based on the AI ​​large model application. These data may include, for example, instructions for any application or method operating on the photovoltaic power station asset assessment device 800 based on the AI ​​large model application, as well as application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules can be keyboards, mice, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the photovoltaic power station asset assessment device 800 based on the AI ​​large model application and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include: Wi-Fi module, Bluetooth module, NFC module.

[0068] In an exemplary embodiment, the photovoltaic power station asset assessment device 800 based on the AI ​​large model application can be implemented by one or more application-specific integrated circuits (Application Specific Integrated Circuit, referred to as ASIC), digital signal processors (Digital Signal Processor, referred to as DSP), digital signal processing devices (Digital Signal Processing Device, referred to as DSPD), programmable logic devices (Programmable Logic Device, referred to as PLD), field programmable gate arrays (Field Programmable Gate Array, referred to as FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned photovoltaic power station asset assessment method based on the AI ​​large model application.

[0069] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the photovoltaic power station asset evaluation method based on the AI ​​large model application are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the photovoltaic power station asset evaluation device 800 based on the AI ​​large model application to complete the above-mentioned photovoltaic power station asset evaluation method based on the AI ​​large model application. Embodiment 4:

[0070] Corresponding to the above method embodiment, a readable storage medium is also provided in this embodiment. The readable storage medium described below and the photovoltaic power station asset assessment method based on the AI ​​large model application described above can be referenced to each other.

[0071] A readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the photovoltaic power station asset evaluation method based on the application of an AI large model in the above-mentioned method embodiment.

[0072] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0073] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0074] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A photovoltaic power station asset evaluation method based on AI large model application, characterized in that: include: Acquire first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; Using the first information and the second information to train a local AI model to obtain a trained AI model, wherein the trained AI model is used to evaluate the health of photovoltaic power station equipment; Use the preset federated learning framework to conduct federated learning on the AI ​​model trained locally to generate a global AI model; Optimizing the global AI model using the third information to obtain an optimized global AI model; An asset evaluation report for a photovoltaic power station is obtained based on the optimized global AI model.

2. The photovoltaic power station asset evaluation method based on the AI ​​large model application according to claim 1 is characterized in that: The step of obtaining the second information includes: Acquire modeling information, wherein the modeling information includes location information of the photovoltaic power station, location information of equipment to be inspected in the photovoltaic power station, and terrain data; Establishing a three-dimensional model of a photovoltaic power station according to the modeling information; Get the type of equipment to be inspected; Planning an inspection route in the three-dimensional model of the photovoltaic power station according to the type of equipment to be inspected; The second information is acquired according to the inspection route.

3. The photovoltaic power station asset evaluation method based on the AI ​​large model application according to claim 1 is characterized in that: Training the local AI model using the first information and the second information includes: Preprocessing the first information to obtain preprocessed first information; Performing standardization on the second information to obtain standardized second information; Acquire a preset first neural network model and a second neural network model; Fusing the first neural network model and the second neural network model to obtain a fused neural network model; The fused neural network model is trained using the preprocessed first information and the standardized second information to obtain a locally trained AI model.

4. The photovoltaic power station asset evaluation method based on the AI ​​large model application according to claim 1 is characterized in that: Use the preset federated learning framework to perform federated learning on the AI ​​model trained locally, including: Get the gradient parameters generated by local training for each node; Encrypting the gradient parameter using an encryption algorithm to obtain an encrypted gradient parameter; Sending the encrypted gradient parameters to a coordination server in a federated learning framework to aggregate global model parameters; Differential privacy technology is used to add noise when the global model is updated to obtain an updated global AI model.

5. The photovoltaic power station asset evaluation method based on AI large model application according to claim 1 is characterized in that: Optimizing the global AI model using the third information includes: Acquire a historical data set, wherein the historical data set includes a large amount of operation data under normal conditions collected in a historical operation database; Using the historical data set to train the autoencoder model to obtain a trained autoencoder model; Sending the third information to the trained autoencoder model to obtain an abnormal data recognition result; Filtering the third information according to the abnormal data identification result to obtain a filtered data set; After the filtered data set is sent to the global AI model, the global AI model is optimized using an incremental learning algorithm.

6. The photovoltaic power station asset evaluation method based on AI large model application according to claim 1 is characterized in that: The photovoltaic power station asset evaluation report is obtained according to the optimized global AI model, including: Sending the first information, the second information, and the third information to the optimized global AI model to obtain a prediction result, wherein the prediction result includes the health of the photovoltaic power station equipment and the power generation; Constructing a scatter plot according to the prediction result, wherein the scatter plot includes a correlation distribution between equipment health and power generation efficiency; Determining the correlation between the equipment health and the power generation efficiency according to the scatter plot; A photovoltaic power station asset evaluation report is generated based on the correlation between the equipment health and power generation efficiency.

7. The photovoltaic power station asset evaluation method based on AI large model application according to claim 6 is characterized in that: Generate a photovoltaic power station asset assessment report based on the correlation between the equipment health and power generation efficiency, including: Determine a calculation formula for power generation efficiency according to the correlation between the equipment health and power generation efficiency of each equipment; Determine the corrected power generation according to the calculation formula; Obtain regional electricity price information; The photovoltaic power station assets are evaluated according to the regional electricity price information and the corrected power generation to obtain an asset evaluation report.

8. A photovoltaic power station asset evaluation system based on AI large model application, characterized in that: include: an acquisition module, configured to acquire first information, second information and third information, wherein the first information includes photovoltaic power station operation data, the second information includes photovoltaic power station equipment images, and the third information includes equipment fault logs and real-time meteorological data; A first processing module, configured to train a local AI model using the first information and the second information to obtain a trained AI model, wherein the trained AI model is used to evaluate the health of photovoltaic power station equipment; The second processing module is used to perform federated learning on the AI ​​model trained on the local end using a preset federated learning framework to generate a global AI model; A third processing module, configured to optimize the global AI model using the third information to obtain an optimized global AI model; An evaluation module is used to obtain a photovoltaic power station asset evaluation report based on the optimized global AI model.

9. The photovoltaic power station asset evaluation system based on AI large model application according to claim 8 is characterized in that: The acquisition module comprises: A first acquisition unit is used to acquire modeling information, wherein the modeling information includes location information of the photovoltaic power station, location information of equipment to be inspected in the photovoltaic power station, and terrain data; A first processing unit, configured to establish a three-dimensional model of a photovoltaic power station according to the modeling information; The second acquisition unit is used to acquire the type of equipment to be inspected; A second processing unit, configured to plan an inspection route in the three-dimensional model of the photovoltaic power station according to the type of equipment to be inspected; The third processing unit is used to obtain second information according to the inspection route.

10. The photovoltaic power station asset evaluation system based on AI large model application according to claim 8 is characterized in that: The first processing module comprises: a fourth processing unit, configured to preprocess the first information to obtain preprocessed first information; a fifth processing unit, configured to perform standardization processing on the second information to obtain standardized second information; A third acquisition unit, used to acquire a preset first neural network model and a second neural network model; a sixth processing unit, configured to fuse the first neural network model and the second neural network model to obtain a fused neural network model; The seventh processing unit is used to train the fused neural network model using the preprocessed first information and the standardized second information to obtain the locally trained AI model.

Images