Distributed photovoltaic energy storage system intelligent scheduling method and system based on AI technology

Abstract

The invention relates to the technical field of intelligent scheduling of electric quantity, in particular to an intelligent scheduling method and system for a distributed photovoltaic energy storage system based on an AI technology. The method comprises the following steps: S1, collecting multi-modal data of a distributed photovoltaic system; s2, using a long-short-term memory network model to predict future generating capacity and power consumption based on the multi-modal data; s3, making a preliminary charging and discharging strategy of the energy storage system based on the prediction result; s4, optimizing the preliminary strategy by adopting a reinforcement learning method; and S5, executing the optimized strategy, and carrying out real-time monitoring to detect an abnormal condition. The intelligent scheduling method and system can accurately predict the power generation amount of a photovoltaic system and the power consumption of a user side by integrating a long-short-term memory network model and a reinforcement learning algorithm, and meanwhile, the influence of an aging attenuation model of a photovoltaic panel and external environment factors (such as altitude and cloud layer shielding) on the power generation efficiency is considered.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent dispatching of electricity, and in particular to an intelligent dispatching method and system for a distributed photovoltaic energy storage system based on AI technology. Background Art

[0002] Solar energy is an intermittent renewable energy source. Its power generation is affected by weather conditions (such as sunny days, cloudy days, cloud cover, etc.), seasonal changes, geographical location and other factors. It has great uncertainty and volatility. This characteristic makes it difficult to predict the power generation of photovoltaic systems, which brings challenges to the stable operation of the power grid. In order to smooth the output fluctuations of photovoltaic systems and improve energy utilization, energy storage devices are usually required. However, how to efficiently manage and dispatch these energy storage resources so that they can maximize the economic benefits of the energy storage system and ensure the reliability of power supply is a complex issue. As the photovoltaic panels age, their photoelectric conversion efficiency will gradually decrease, which has a negative impact on the long-term performance and economic benefits of the photovoltaic system. Therefore, an intelligent scheduling method and system for distributed photovoltaic energy storage systems based on AI technology are provided. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent scheduling method and system for a distributed photovoltaic energy storage system based on AI technology to solve the problems of solar energy volatility and uncertainty, low energy utilization, and decreased efficiency of aging photovoltaic panels raised in the above background technology.

[0004] To achieve the above objectives, the present invention aims to provide an intelligent scheduling method for a distributed photovoltaic energy storage system based on AI technology, comprising: S1, collect multimodal data of distributed photovoltaic systems; S2. Based on multimodal data, the long short-term memory network model is used to predict future power generation and power consumption. Since the aging of photovoltaic panels has an impact on power generation efficiency, the photovoltaic panel aging attenuation model is introduced in the prediction process; S3. Based on the prediction results, formulate a preliminary charging and discharging strategy for the energy storage system; S4, using reinforcement learning methods to optimize the initial strategy; S5. Execute the optimized strategy and perform real-time monitoring to detect abnormal situations.

[0005] As a further improvement of the present technical solution, in S1, the multimodal data of the distributed photovoltaic system includes the power generation efficiency of the photovoltaic panels, the status of the energy storage equipment, and the grid load.

[0006] As a further improvement of the technical solution, in S2, predicting future power generation and power consumption using a long short-term memory network model based on multimodal data includes the following steps: S2.1, preprocessing multimodal data; S2.2, extracting features related to power generation and power consumption prediction from multimodal data; S2.3. Select the long short-term memory network as the prediction model, and divide the multimodal data set into a training set, a validation set, and a test set. Construct a power generation model and a power consumption model. Since the aging of photovoltaic panels affects the power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model. The aging of photovoltaic panels is affected by the dust thickness on the photovoltaic panels and the crack length of the photovoltaic panels. The dust thickness and the crack length parameters of the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model. S2.4. Use the power gap function to measure the gap between the model prediction value and the actual value, and use the gradient descent method to optimize the power gap function. According to the influence of altitude, the power gap function is optimized; S2.5, using the training set to train the power model and the power consumption model; S2.6. Receive new input data in real time through the trained power generation model and power consumption model to predict future power generation and power consumption.

[0007] As a further improvement of the technical solution, in S2.3, the power generation model is: ; Since the aging of photovoltaic panels has an impact on power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model: ; The aging of photovoltaic panels is affected by the dust thickness and crack length on the photovoltaic panels. The dust thickness and crack length parameters on the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model: ; The power consumption model is: ; in, Indicates at a point in time Forecasted power generation; Indicates at a point in time The power generation feature vector input to the model; It represents the predicted power generation after introducing the photovoltaic panel aging attenuation model; Indicates at a point in time Forecasted electricity consumption; Indicates at a point in time The electricity consumption feature vector input to the model; Indicates a point in time; represents the function that maps input features and hidden states to predicted power generation; Indicates the previous time point The hidden state of Represents the parameters of the model; Indicates at a point in time Historical power generation; Indicates at a point in time Historical power generation; Indicates the working status of the photovoltaic panel; Indicates at a point in time The aging degree of photovoltaic panels; Indicates at a point in time The dust thickness on the photovoltaic panel; Indicates at a point in time Length of cracks in photovoltaic panels; A function that represents the calculation of the comprehensive aging attenuation degree.

[0008] As a further improvement of the technical solution, in S2.4, the power difference function is: ; In view of the influence of altitude, the power difference function is optimized: ; in, Indicates the gap between the model prediction value and the actual value; Represents the optimized power gap function value; represents the number of samples; Indicates at a point in time Actual electricity generation or consumption; Indicates at a point in time Forecasted electricity generation or consumption; The weight coefficient representing the effect of altitude; represents the influence coefficient of altitude; Indicates The altitude of the samples; Indicates the sample index; Indicates a point in time.

[0009] As a further improvement of the technical solution, in S3, based on the prediction results, a preliminary charging and discharging strategy for the energy storage system is formulated, including the following steps: S3.1. Based on the forecast results, identify the peak and valley periods of photovoltaic power generation; S3.2. Analyze the user electricity demand in the forecast results and identify the peak and off-peak hours; S3.3, during peak power generation periods, excess electricity is stored in energy storage devices. During peak power consumption periods, electricity in the energy storage devices is used to meet user demand. During low power consumption periods, if the energy storage devices are insufficient, electricity is purchased from the power grid to supplement the energy storage devices. S3.4. Set the charging rate according to the specifications and current state of the energy storage device, set the discharging rate according to the specifications and current state of the energy storage device, and set the maximum and minimum allowable power of the energy storage device; S3.5. Forecast the specific situation of electricity supply and demand in the next few hours and adjust the charging and discharging plan.

[0010] As a further improvement of the technical solution, in S4, the preliminary strategy is optimized by using a reinforcement learning method, including the following steps: S4.1. Define the state variables of the system and define the actions that the energy storage system can take; S4.2. Design a reward function that reflects the system goal to minimize transmission losses. Consider the different transmission losses caused by different transmission distances, optimize the reward function, and further optimize the reward function for the shadow of photovoltaic panels caused by obstructions blocking the sun; S4.3. Select a reinforcement learning algorithm and set the algorithm's hyperparameters; S4.4. Use the reinforcement learning algorithm to iteratively train the preliminary strategy. In each training cycle, the algorithm selects an action based on the current state, observes the new state variables and obtains the reward function after executing the action, and then updates and optimizes the preliminary strategy.

[0011] As a further improvement of the technical solution, in S4.2, the reward function is: ; Considering the different transmission losses caused by different transmission distances, the reward function is optimized: ; The reward function is further optimized for the shadow of photovoltaic panels caused by obstructions blocking the sun: ; in, Represents the transmission loss of the system; It represents the transmission loss after considering different transmission distances; represents the transmission loss after optimization for the obstruction of the sun; The coefficient that adjusts the proportion of transmission losses in the total reward; Indicates the length of the time period; Indicates a point in time; Indicates a time interval; Indicates at time transmission distance; Indicates at time The electrical power delivered; The coefficient that adjusts the proportion of zero load in the total reward; Indicates zero load time; Indicates the maximum possible load time; It represents the coefficient related to the transmission distance and is used to adjust the effect of the transmission distance on the transmission loss; Indicates the total length of the transmission line; It represents the coefficient related to the transmitted electric power and is used to adjust the influence of the transmitted electric power on the transmission loss; Indicates the electrical power delivered; Indicates at time Correction factor for illumination due to occluders.

[0012] As a further improvement of the technical solution, in S5, executing the optimized strategy and performing real-time monitoring to detect abnormal situations include the following steps: S5.1. Send the final optimized dispatching strategy to the energy storage system and related equipment; S5.2, adjust the charging and discharging rate and state of the energy storage device in real time through the control system; S5.3. Continuously collect real-time data from photovoltaic panels and energy storage devices, and use the collected real-time data to evaluate the current status of the system; S5.4. Set the safety range and threshold of the collected real-time data based on historical data; S5.5. Compare the real-time data collected in real time with the set threshold to detect whether there is an abnormal situation. Once an abnormal situation is detected, the alarm mechanism is immediately triggered.

[0013] On the other hand, the present invention provides an intelligent scheduling system for a distributed photovoltaic energy storage system based on AI technology, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any one of the above-mentioned intelligent scheduling methods for a distributed photovoltaic energy storage system based on AI technology.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The intelligent dispatching method and system of the distributed photovoltaic energy storage system based on AI technology can accurately predict the power generation of the photovoltaic system and the power consumption on the user side by integrating the long short-term memory network model and the reinforcement learning algorithm, while considering the aging attenuation model of the photovoltaic panels and the influence of external environmental factors (such as altitude, cloud cover, etc.) on the power generation efficiency. This method and system not only helps to reasonably plan the charging and discharging strategy of the energy storage system and reduce unnecessary energy waste, but also effectively reduce the user's electricity bill expenditure and improve the economic benefits of the entire photovoltaic energy storage system.

[0015] 2. The AI-based distributed photovoltaic energy storage system intelligent scheduling method and system can timely detect and handle problems in the system operation process, such as overcharging, over-discharging or equipment failure, by monitoring the working status of the energy storage equipment in real time and combining the preset safety threshold to detect potential abnormalities, thereby ensuring the long-term stable operation of the system. In addition, through the optimized charging and discharging strategy, it is also possible to avoid damage to the battery caused by frequent charging and discharging operations, extend the service life of the energy storage equipment, and improve the overall safety of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The figure is a flow chart of the overall method of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Embodiment 1: See also Figure 1 As shown, this embodiment provides an intelligent scheduling method for a distributed photovoltaic energy storage system based on AI technology, including the following steps: S1, collect multimodal data of distributed photovoltaic systems; In this embodiment, the multimodal data of the distributed photovoltaic system includes the power generation efficiency of the photovoltaic panels, the status of the energy storage device, and the grid load.

[0019] S2, predicting future power generation and consumption using long short-term memory network model based on multimodal data; In this embodiment, the long short-term memory network model is used to predict future power generation and power consumption based on multimodal data, including the following steps: S2.1. Preprocess multimodal data to remove noise and outliers, handle missing values, and ensure data quality. Different cleaning methods are required for different types of data. Data of different scales should be converted to the same scale to facilitate model training. For example, all numerical data should be scaled to between 0 and 1 or Z-score standardized. S2.2, extracting features related to power generation and power consumption prediction from multimodal data, including historical power generation, working status of photovoltaic panels, etc.; S2.3. Select the long short-term memory network as the prediction model, and divide the multimodal data set into a training set, a validation set, and a test set. Construct a power generation model and a power consumption model. Since the aging of photovoltaic panels affects the power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model. The aging of photovoltaic panels is affected by the dust thickness on the photovoltaic panels and the crack length of the photovoltaic panels. The dust thickness and the crack length parameters of the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model. Among them, the long short-term memory network is a recurrent neural network that is used to process and predict data with time series characteristics; the long short-term memory network can effectively capture the long-term dependencies in time series data. In the photovoltaic energy storage system, both power generation and power consumption have obvious time dependence. For example, photovoltaic power generation is affected by sunshine time and usually reaches its peak during the day, especially from noon to afternoon; while power consumption is affected by people's living habits and usually reaches its peak in the morning and evening. The long short-term memory network can model these time patterns and trends well, thereby improving the accuracy of the prediction; in the distributed photovoltaic energy storage system, the data sources are diverse, including the power generation efficiency of photovoltaic panels, the status of energy storage equipment, and the grid load. The long short-term memory network can process multimodal data and improve the comprehensiveness and accuracy of the prediction by fusing different types of input data together for joint modeling; the power generation model is: ; Furthermore, in the long short-term memory network model It is a complex nonlinear function, which consists of multiple gating mechanisms, including input gate, forget gate and output gate. The function of this function is to convert the current input and the hidden state at the previous moment Mapped to the output of the current moment ; The function breaks down into the following steps: The input gate controls which values ​​will update the cell state: ; in, represents the Sigmoid activation function, and are the input weight matrix of the input gate and the hidden state weight matrix, is the bias term; The forget gate decides which information to discard: ; The candidate cell state calculates new information to update the cell state: ; in, is the hyperbolic tangent activation function, which is used to limit the output between -1 and 1; The cell state is updated using the results of the forget gate and the input gate: ; in, represents element-wise multiplication; The output gate determines which parts of the cell state will be output: ; Output at the current moment Determined by the cell state and the output gate: ; The final predicted power generation yes After an additional linear layer or directly equal to ; The aging of photovoltaic panels will gradually reduce their power generation efficiency. If this factor is not taken into account and only historical data is relied upon for power generation prediction, there may be a large deviation between the prediction results and the actual situation. The introduction of the photovoltaic panel aging attenuation model can more accurately reflect the actual power generation capacity of the photovoltaic panels, thereby improving the accuracy of power generation prediction; based on more accurate power generation prediction, a more reasonable energy storage system charging and discharging strategy can be formulated. For example, when the power generation is predicted to be low, electricity can be purchased from the power grid in advance to supplement the energy storage equipment; when the power generation is predicted to be high, the excess electricity can be stored first for emergency use; the photovoltaic panel aging attenuation model can be adjusted according to different photovoltaic panel types and usage environments, and has strong adaptability; since the aging of photovoltaic panels has an impact on power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model: ; Furthermore, To predict at a point in time Power generation , this function not only takes into account the historical power generation and , and also considers the working status of the photovoltaic panels , Aging degree of photovoltaic panels and the hidden state at the previous time point , define the input feature vector ,in, The expanded form of the function is: Input Gate : ; Forget Gate : ; Candidate cell states : ; Cell status update : ; Output Gate : ; The hidden state at the current moment : ; Forecasting power generation : ,in, is a weight matrix, are bias terms, which are part of the model parameters ; Dust covering the surface of photovoltaic panels will block part of the sunlight from reaching the photovoltaic panels, thereby reducing light absorption, lowering the photoelectric conversion efficiency, and further increasing the aging of photovoltaic panels. Dust particles will reflect and scatter part of the light, further reducing the effective light intensity reaching the photovoltaic panels. Dust coverage will increase the surface temperature of photovoltaic panels, and high temperature will reduce the photoelectric conversion efficiency of photovoltaic panels. In the power generation prediction model, the introduction of dust thickness parameters can more accurately reflect the actual power generation capacity of photovoltaic panels. For example, if the dust thickness is predicted to be large within a certain period of time, the power generation prediction value can be appropriately lowered. Cracks will damage the physical structure of photovoltaic panels, affecting their internal electrical The continuity of the crack will reduce the photoelectric conversion efficiency and may also cause the electrical performance of the photovoltaic panel to decline, such as short circuit, open circuit, etc., further affecting the power generation efficiency; the increase in crack length will cause a significant decrease in the power generation efficiency of the photovoltaic panel, and serious cracks may even cause the photovoltaic panel to fail completely; in the power generation prediction model, the introduction of crack length parameters can more accurately reflect the actual power generation capacity of the photovoltaic panel. For example, if the crack length is predicted to be large within a certain period of time, the power generation prediction value can be appropriately lowered; by considering the dust thickness and crack length parameters at the same time, the power generation prediction model can more comprehensively reflect the actual condition of the photovoltaic panel and improve the prediction accuracy; The aging of photovoltaic panels is affected by the dust thickness and crack length on the photovoltaic panels. The dust thickness and crack length parameters on the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model: ; The power consumption model is: ; in, Indicates at a point in time Forecasted power generation; Indicates at a point in time The power generation feature vector input to the model; It represents the predicted power generation after introducing the photovoltaic panel aging attenuation model; Indicates at a point in time Forecasted electricity consumption; Indicates at a point in time The electricity consumption feature vector input to the model; Indicates a point in time; represents the function that maps input features and hidden states to predicted power generation; Indicates the previous time point The hidden state contains the information of all previous time points; Represents the parameters of the model, including the weight matrix and bias vector; Indicates at a point in time Historical power generation; Indicates at a point in time Historical power generation; Indicates the working status of the photovoltaic panel; Indicates at a point in time The aging degree of photovoltaic panels; Indicates at a point in time The dust thickness on the photovoltaic panel; Indicates at a point in time Length of cracks in photovoltaic panels; A function representing the degree of calculated comprehensive aging attenuation; The weight representing the dust thickness; represents the weight of crack length; Indicates the weight of time; is a constant, indicating the aging rate; Furthermore, the electricity consumption model The expansion of is similar to the power generation model. Including historical electricity consumption , weather conditions (such as temperature, humidity, etc.), user behavior patterns ; Therefore, the input feature vector is expressed as: ; Input Gate : ; Forget Gate : ; Candidate cell states : ; Cell status update : ; Output Gate: ; The hidden state at the current moment : ; Forecast electricity consumption : The final predicted electricity consumption is obtained through an additional linear layer: ; S2.4. Use the power gap function to measure the gap between the model prediction value and the actual value, and use the gradient descent method to optimize the power gap function. According to the influence of altitude, the power gap function is optimized; Among them, the power gap function can quantify the difference between the model's predicted value and the actual value, thereby evaluating the model's prediction performance, which helps to determine whether the model accurately captures the patterns and trends in the data; based on more accurate prediction results, a more reasonable energy storage system charging and discharging strategy can be formulated; the power gap function is: ; Altitude has a significant impact on the power generation efficiency of photovoltaic systems. Environmental factors such as atmospheric pressure, temperature and humidity in high altitude areas will affect the photoelectric conversion efficiency of photovoltaic panels. By introducing the influence of altitude in the power gap function, the impact of these environmental factors on power generation and power consumption can be more accurately reflected, thereby improving the prediction accuracy of the model; the altitudes in different regions vary greatly, and the optimized power gap function can better adapt to photovoltaic systems in different geographical environments, which is conducive to the promotion and application of intelligent scheduling methods in different regions and improve the universality and flexibility of the system. For example, in high altitude areas, due to the influence of atmospheric pressure and temperature, the power generation efficiency of photovoltaic panels may be different. The optimized model can more accurately predict these changes, thereby formulating a more reasonable scheduling strategy; the power gap function is optimized in view of the influence of altitude: ; in, Indicates the gap between the model prediction value and the actual value; It represents the optimized power difference function value, taking into account the effect of altitude; represents the number of samples, that is, the number of data points in the time series; Indicates at a point in time Actual electricity generation or consumption; Indicates at a point in time Forecasted electricity generation or consumption; The weight coefficient representing the impact of altitude is used to adjust the importance of altitude in the loss function; Represents the influence coefficient of altitude, which indicates the influence of unit altitude on the loss function; Indicates The altitude of the samples; Indicates the sample index; Indicates a point in time; Furthermore, the relationship model between altitude and power generation or power consumption is established using regression analysis method, and the coefficients of the model are used to determine and If the regression model is: , Indicates the amount of electricity generated or consumed. is the altitude, is the intercept, is the influence coefficient of altitude, is the error term, obtained through regression analysis After the value is set, adjust it according to the performance of the model and actual needs The value of S2.5. Use the training set to train the power model and the power consumption model. During the training process, regularly use the validation set to evaluate the model performance, and adjust the hyperparameters (such as learning rate, batch size, etc.) according to the performance of the validation set to prevent overfitting; S2.6. New input data is received in real time through the trained power generation model and power consumption model to predict future power generation and power consumption, providing a basis for subsequent energy storage system scheduling. These predicted values ​​will be used as input to formulate preliminary charging and discharging strategies for the energy storage system and further optimize energy management. If the power generation is expected to be greater than the power consumption, it indicates that there is a power surplus; conversely, if the power consumption is greater than the power generation, there may be a power shortage.

[0020] S3. Based on the prediction results, formulate a preliminary charging and discharging strategy for the energy storage system; In this embodiment, based on the prediction results, a preliminary charging and discharging strategy for the energy storage system is formulated, including the following steps: S3.1. Based on the prediction results, identify the peak and valley periods of photovoltaic power generation. Usually, the peak period of photovoltaic power generation occurs during the day, especially from noon to afternoon on sunny days; while the valley period occurs at night or on rainy days; S3.2. Analyze the user electricity demand in the forecast results and identify the peak and off-peak periods. Generally speaking, the peak electricity demand of household users is at night, while the peak electricity demand of commercial users may be during the day; S3.3. During peak power generation periods, excess electricity is stored in energy storage devices (if the energy storage device has reached full load, the energy storage device is used for immediate high-energy-consuming devices (such as hot water heating, electric vehicle charging, etc.) to reduce the energy consumption of these devices during peak power consumption periods). During peak power consumption periods, the electricity in the energy storage device is used to meet user needs. During low power consumption periods, if the energy storage device is insufficient, electricity is purchased from the power grid to supplement the energy storage device; S3.4. Set the charging rate according to the specifications and current status of the energy storage device to avoid over-charging that may shorten the battery life or cause safety issues. Set the discharge rate according to the specifications and current status of the energy storage device to avoid over-discharging that may cause battery performance degradation or failure to meet user needs. Set the maximum and minimum allowable power of the energy storage device to ensure that the device operates within a safe range, for example, avoid fully charging or fully discharging the battery to extend the battery life. S3.5. Forecast the specific situation of electricity supply and demand for the next few hours or a day, and adjust the charging and discharging plan to adapt to actual needs. For example, if it is predicted that user demand will suddenly increase within a certain period of time, increase the charging capacity of the energy storage equipment in advance.

[0021] S4, using reinforcement learning methods to optimize the initial strategy; In this embodiment, reinforcement learning is a method for learning the optimal behavior strategy by interacting with the environment. The reward function is the core concept in reinforcement learning and is used to evaluate the quality of each action taken by the intelligent agent. The design of the reward function determines the learning goal of the system. The preliminary strategy is usually generated based on static rules or simple models, which may not fully consider the dynamically changing environmental factors and system states. Through the reinforcement learning method, the scheduling strategy can be dynamically adjusted and optimized to make it more adaptable to actual operating conditions and improve the overall performance of the system. The reinforcement learning method can automatically adjust the scheduling strategy according to the real-time state of the system and environmental changes to achieve adaptive scheduling, which helps the system to maintain efficient operation under different working conditions and improve the flexibility and robustness of the system. Through the reinforcement learning method, the supply and demand balance of energy can be more accurately predicted and managed, and the charging and discharging strategy of energy storage equipment can be optimized. The optimization of the preliminary strategy using the reinforcement learning method includes the following steps: S4.1. Define the state variables of the system, including the current time, the power generation of the photovoltaic panel, the remaining power of the energy storage device, and the grid load, and define the actions that the energy storage system can take, such as charging, discharging, and maintaining the current state; S4.2. Design a reward function that reflects the system goal to minimize transmission losses. Consider the different transmission losses caused by different transmission distances, optimize the reward function, and further optimize the reward function for the shadow of photovoltaic panels caused by obstructions blocking the sun. The reward function considers immediate rewards (such as giving positive rewards when the system charges during peak power generation and discharges during peak power consumption; giving positive rewards when the system avoids unnecessary charging and discharging operations and reduces equipment losses) and long-term rewards (such as giving positive rewards when the system maintains stable operation for a long time and avoids frequent charging and discharging operations); Among them, the reward function is: ; Transmission loss is an inevitable part of the energy transmission process, especially in long-distance transmission. By considering the transmission distance, the energy transmission path can be optimized and the energy loss during the transmission process can be reduced. The system can choose a path with less transmission loss for energy transmission, thereby improving the efficiency of energy transmission. Transmission loss will cause part of the electric energy to be lost during the transmission process, increasing energy waste. By optimizing the transmission path and strategy, unnecessary transmission loss can be reduced, thereby improving energy utilization efficiency. Transmission loss not only wastes energy, but also increases the operating cost of the system. By optimizing the transmission path and strategy, transmission loss can be reduced, thereby reducing the cost of purchasing electricity from the power grid. Considering the different transmission losses caused by different transmission distances, the reward function is optimized: ; When an obstruction blocks the sun, the power generation of the photovoltaic panels will suddenly drop. This instantaneous change places higher demands on the prediction of power generation. By optimizing the reward function, this instantaneous change can be more accurately predicted and managed, improving the accuracy of power generation prediction. The decrease in power generation caused by obstructions blocking the sun will affect the charging and discharging strategy of the energy storage system. The optimized reward function can better cope with this change and ensure the stable operation of the system under different weather conditions. For example, when it is predicted that an obstruction will occur, electricity can be purchased from the power grid in advance to supplement the energy storage equipment, or the discharge strategy of the energy storage equipment can be adjusted to cope with the decrease in power generation. By optimizing the reward function, energy waste caused by obstructions can be reduced, for example, excessive discharge can be avoided when power generation decreases, resulting in the inability of the energy storage equipment to meet subsequent demand. The reward function is further optimized for the shadow of photovoltaic panels caused by obstructions blocking the sun: ; ; in, Represents the transmission loss of the system; It represents the transmission loss after considering different transmission distances; represents the transmission loss after optimization for the obstruction of the sun; The coefficient that adjusts the proportion of transmission losses in the total reward; Indicates the length of the time period; Indicates a point in time; Indicates a time interval; Indicates at time transmission distance; Indicates at time The electrical power delivered; The coefficient that adjusts the proportion of zero load in the total reward; represents zero load time, i.e. the time when the system is not loaded; Indicates the maximum possible load time, that is, the total time the system can theoretically continue to work; It represents the coefficient related to the transmission distance and is used to adjust the effect of the transmission distance on the transmission loss; Indicates the total length of the transmission line; It represents the coefficient related to the transmitted electric power and is used to adjust the influence of the transmitted electric power on the transmission loss; Indicates the electrical power delivered (a constant or average value); Indicates at time Light correction factor due to occlusion; Indicates at time Predicted light intensity; Indicates at time The ratio of occlusion by the occluder. The value range is [0,1], where 0 means no occlusion and 1 means complete occlusion. S4.3. Select a reinforcement learning algorithm (Q-Learning) and set the algorithm's hyperparameters, such as learning rate, discount factor, and exploration strategy; S4.4. Use the reinforcement learning algorithm to iteratively train the preliminary strategy. In each training cycle, the algorithm selects an action based on the current state, observes the new state variables and obtains the reward function after executing the action, and then updates and optimizes the preliminary strategy.

[0022] S5. Execute the optimized strategy and perform real-time monitoring to detect abnormal situations; In this embodiment, executing the optimized strategy and performing real-time monitoring to detect abnormal situations include the following steps: S5.1. Send the final optimized dispatching strategy to the energy storage system and related equipment to ensure that all equipment performs charging and discharging operations according to the optimized strategy; S5.2. Use the control system to adjust the charge and discharge rate and state of the energy storage device in real time to ensure that the system operates in the optimal state; S5.3. Continuously collect real-time data from photovoltaic panels and energy storage devices. The real-time data includes the power generation of photovoltaic panels, the remaining power of energy storage devices, and the load on the power grid. Use the collected real-time data to evaluate the current status of the system, such as determining whether the energy storage device is overloaded or whether the photovoltaic panels are working properly. S5.4. Set the safety range and threshold of the collected real-time data based on historical data; The power generation of photovoltaic panels is set with a safety range based on the average and standard deviation of historical data. For example, assuming that the average power generation of historical data is 1000 kWh and the standard deviation is 100 kWh, the safety range can be set as: , set the alarm threshold, and trigger the alarm when the power generation exceeds the normal range, for example: lower threshold: 750 kWh, upper threshold: 1250 kWh; if the average remaining power of historical data is 50% and the standard deviation is 10%, the remaining power safety range of the energy storage device status can be set as: , the lower limit threshold of the remaining power is 20%, and the upper limit threshold is 80%; if the average grid load of historical data is 500 kWh and the standard deviation is 50 kWh, the safety range can be set as: , grid load lower threshold: 350 kWh, upper threshold: 650 kWh; S5.5. Compare the real-time data collected in real time with the set threshold to detect whether there is an abnormal situation. Once an abnormal situation is detected, the alarm mechanism is immediately triggered.

[0023] Embodiment 2: This embodiment provides an intelligent scheduling system for a distributed photovoltaic energy storage system based on AI technology, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the above-mentioned intelligent scheduling method for a distributed photovoltaic energy storage system based on AI technology.

[0024] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. An intelligent dispatching method for distributed photovoltaic energy storage system based on AI technology, characterized in that: The following steps are involved: S1, collect multimodal data of distributed photovoltaic systems; S2. Use the long short-term memory network model based on multimodal data to predict future power generation and power consumption, and introduce the photovoltaic panel aging attenuation model in the prediction process; S3. Based on the prediction results, formulate a preliminary charging and discharging strategy for the energy storage system; S4, using reinforcement learning methods to optimize the initial strategy; S5. Execute the optimized strategy and perform real-time monitoring to detect abnormal situations.

2. The intelligent scheduling method for distributed photovoltaic energy storage system based on AI technology according to claim 1 is characterized by: In S1, the multimodal data of the distributed photovoltaic system includes the power generation efficiency of the photovoltaic panels, the status of the energy storage equipment, and the grid load.

3. The intelligent scheduling method for distributed photovoltaic energy storage system based on AI technology according to claim 2 is characterized by: In S2, the future power generation and power consumption are predicted using a long short-term memory network model based on multimodal data, including the following steps: S2.1, preprocessing multimodal data; S2.2, extracting features related to power generation and power consumption prediction from multimodal data; S2.

3. Select the long short-term memory network as the prediction model, and divide the multimodal data set into a training set, a validation set, and a test set. Construct a power generation model and a power consumption model. Since the aging of photovoltaic panels affects the power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model. The aging of photovoltaic panels is affected by the dust thickness on the photovoltaic panels and the crack length of the photovoltaic panels. The dust thickness and the crack length parameters of the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model. S2.

4. Use the power gap function to measure the gap between the model prediction value and the actual value, and use the gradient descent method to optimize the power gap function. According to the influence of altitude, the power gap function is optimized; S2.5, using the training set to train the power model and the power consumption model; S2.

6. Receive new input data in real time through the trained power generation model and power consumption model to predict future power generation and power consumption.

4. The intelligent scheduling method for distributed photovoltaic energy storage system based on AI technology according to claim 3 is characterized by: In S2.3, the power generation model is: ； Since the aging of photovoltaic panels has an impact on power generation efficiency, the photovoltaic panel aging attenuation model is introduced into the power generation model: ； The aging of photovoltaic panels is affected by the dust thickness and crack length on the photovoltaic panels. The dust thickness and crack length parameters on the photovoltaic panels are introduced into the photovoltaic panel aging attenuation model: ； The power consumption model is: ； in, Indicates at a point in time Forecasted power generation; Indicates at a point in time The power generation feature vector input to the model; It represents the predicted power generation after introducing the photovoltaic panel aging attenuation model; Indicates at a point in time Forecasted electricity consumption; Indicates at a point in time The electricity consumption feature vector input to the model; Indicates a point in time; represents the function that maps input features and hidden states to predicted power generation; Indicates the previous time point The hidden state of Represents the parameters of the model; Indicates at a point in time Historical power generation; Indicates at a point in time Historical power generation; Indicates the working status of the photovoltaic panel; Indicates at a point in time The aging degree of photovoltaic panels; Indicates at a point in time The dust thickness on the photovoltaic panel; Indicates at a point in time Length of cracks in photovoltaic panels; A function that represents the calculation of the comprehensive aging attenuation degree.

5. The intelligent dispatching method for distributed photovoltaic energy storage system based on AI technology according to claim 4 is characterized in that: In S2.4, the power difference function is: ； In view of the influence of altitude, the power difference function is optimized: ； in, Indicates the gap between the model prediction value and the actual value; Represents the optimized power gap function value; represents the number of samples; Indicates at a point in time Actual electricity generation or consumption; Indicates at a point in time Forecasted electricity generation or consumption; The weight coefficient representing the effect of altitude; represents the influence coefficient of altitude; Indicates The altitude of the samples; Indicates the sample index; Indicates a point in time.

6. The intelligent dispatching method for distributed photovoltaic energy storage system based on AI technology according to claim 5 is characterized by: In S3, based on the prediction results, a preliminary charging and discharging strategy for the energy storage system is formulated, including the following steps: S3.

1. Based on the forecast results, identify the peak and valley periods of photovoltaic power generation; S3.

2. Analyze the user electricity demand in the forecast results and identify the peak and off-peak hours; S3.3, during peak power generation periods, excess electricity is stored in energy storage devices. During peak power consumption periods, electricity in the energy storage devices is used to meet user demand. During low power consumption periods, if the energy storage devices are insufficient, electricity is purchased from the power grid to supplement the energy storage devices. S3.

4. Set the charging rate according to the specifications and current state of the energy storage device, set the discharging rate according to the specifications and current state of the energy storage device, and set the maximum and minimum allowable power of the energy storage device; S3.

5. Forecast the specific situation of electricity supply and demand in the next few hours and adjust the charging and discharging plan.

7. The intelligent dispatching method for distributed photovoltaic energy storage system based on AI technology according to claim 6 is characterized by: In S4, the preliminary strategy is optimized using a reinforcement learning method, including the following steps: S4.

1. Define the state variables of the system and define the actions that the energy storage system can take; S4.

2. Design a reward function that reflects the system goal to minimize transmission losses. Consider the different transmission losses caused by different transmission distances, optimize the reward function, and further optimize the reward function for the shadow of photovoltaic panels caused by obstructions blocking the sun; S4.

3. Select a reinforcement learning algorithm and set the algorithm's hyperparameters; S4.

4. Use the reinforcement learning algorithm to iteratively train the preliminary strategy. In each training cycle, the algorithm selects an action based on the current state, observes the new state variables and obtains the reward function after executing the action, and then updates and optimizes the preliminary strategy.

8. The intelligent dispatching method for distributed photovoltaic energy storage system based on AI technology according to claim 7 is characterized by: In S4.2, the reward function is: ； Considering the different transmission losses caused by different transmission distances, the reward function is optimized: ； The reward function is further optimized for the shadow of photovoltaic panels caused by obstructions blocking the sun: ； in, Represents the transmission loss of the system; It represents the transmission loss after considering different transmission distances; represents the transmission loss after optimization for the obstruction of the sun; The coefficient that adjusts the proportion of transmission losses in the total reward; Indicates the length of the time period; Indicates a point in time; Indicates a time interval; Indicates at time transmission distance; Indicates at time The electrical power delivered; The coefficient that adjusts the proportion of zero load in the total reward; Indicates zero load time; Indicates the maximum possible load time; It represents the coefficient related to the transmission distance and is used to adjust the effect of the transmission distance on the transmission loss; Indicates the total length of the transmission line; It represents the coefficient related to the transmitted electric power and is used to adjust the influence of the transmitted electric power on the transmission loss; Indicates the electrical power delivered; Indicates at time Correction factor for illumination due to occluders.

9. The intelligent scheduling method for distributed photovoltaic energy storage system based on AI technology according to claim 8 is characterized in that: In S5, the optimized strategy is executed, and real-time monitoring is performed to detect abnormal situations, including the following steps: S5.

1. Send the final optimized dispatching strategy to the energy storage system and related equipment; S5.2, adjust the charging and discharging rate and state of the energy storage device in real time through the control system; S5.

3. Continuously collect real-time data from photovoltaic panels and energy storage devices, and use the collected real-time data to evaluate the current status of the system; S5.

4. Set the safety range and threshold of the collected real-time data based on historical data; S5.

5. Compare the real-time data collected in real time with the set threshold to detect whether there is an abnormal situation. Once an abnormal situation is detected, the alarm mechanism is immediately triggered.

10. An intelligent dispatching system for a distributed photovoltaic energy storage system based on AI technology, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes a computer program to implement the intelligent scheduling method for a distributed photovoltaic energy storage system based on AI technology as described in any one of claims 1 to 9.

Images