A solar photovoltaic power generation monitoring and scheduling method

Abstract

This invention discloses a method for monitoring and scheduling solar photovoltaic power generation, comprising the following steps: S1, collecting real-time operating data of the solar photovoltaic power generation system; S2, modeling each component of the system using a hybrid graph optimization algorithm to generate an optimal resource allocation scheme; S3, dynamically adjusting the scheduling strategy based on real-time data using adaptive neural networks, long short-term memory networks, and graph convolutional networks; S4, optimizing power generation and energy storage strategies using a time-varying adaptive optimization algorithm; S5, monitoring health status and predicting faults using an improved variational Bayesian prediction algorithm; S6, optimizing the energy storage management scheme based on fault early warning information; S7, refining system scheduling and adjusting various operating parameters using a simulated annealing algorithm; S8, evaluating and optimizing the scheduling scheme in real time using an intelligent scheduling system and a multivariate adaptive control algorithm. This invention can improve the scheduling efficiency and operational stability of photovoltaic power generation systems, optimize resource allocation, and provide fault early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of solar photovoltaic power generation, and in particular to a solar photovoltaic power generation monitoring and scheduling method. Background Art

[0002] As the global climate change problem intensifies, solar photovoltaic power generation, as a green and renewable energy source, has become one of the important ways for countries to promote energy transformation. Photovoltaic power generation systems can significantly reduce greenhouse gas emissions and reduce dependence on traditional fossil energy by converting solar energy into electrical energy. However, despite its significant environmental advantages, photovoltaic power generation still faces many challenges in its large-scale application, especially in power prediction, scheduling optimization and system management.

[0003] The output power of photovoltaic power generation systems is affected by many factors, including sunshine intensity, meteorological conditions, seasonal changes, etc. Due to the large power fluctuations of photovoltaic power generation, traditional power generation prediction methods often rely only on historical data and use simple statistical analysis models for prediction. These methods fail to fully consider the dynamic changes of environmental factors. Therefore, the accuracy of the prediction results is poor, which in turn affects the scheduling of the system and the matching of grid loads. With the increase in the proportion of photovoltaic power generation, how to accurately predict the power generation of photovoltaic systems has become an important issue to ensure the stability of the grid and improve energy efficiency.

[0004] In addition, photovoltaic power generation systems are usually composed of multiple components. How to coordinate the work of these components to achieve optimal system scheduling and resource allocation remains a difficult problem. Traditional scheduling methods are mostly based on preset rules and control by setting fixed scheduling parameters. However, this static scheduling method is not flexible enough when facing time-varying environments and cannot effectively respond to changes in different time and environmental conditions. Therefore, traditional photovoltaic system scheduling methods often have problems such as slow response and low energy efficiency.

[0005] In order to optimize the operating efficiency of photovoltaic power generation systems, many researchers have tried to use emerging technologies such as artificial intelligence and machine learning to solve the prediction and scheduling problems of photovoltaic power generation. Common technologies include neural networks, genetic algorithms, fuzzy control, etc. Although these methods can improve prediction accuracy and scheduling efficiency to a certain extent, most methods still have limitations such as over-reliance on historical data, high computational complexity, and inability to respond to sudden changes in real time. In addition, existing technologies often ignore the health management and fault prediction of photovoltaic system equipment, which makes it difficult to detect potential problems in time during system operation, affecting the reliability and safety of the system.

[0006] Existing energy storage system dispatching methods are usually based on simple rules or optimization models. These methods are difficult to respond quickly to rapidly changing grid demand and photovoltaic power generation fluctuations, resulting in low charging and discharging efficiency of energy storage systems and low grid dispatching efficiency. At the same time, changes in grid load and fluctuations in photovoltaic power generation are also factors that affect system stability. Traditional load forecasting and dispatching methods cannot make flexible adjustments in real time according to the actual needs of the grid, thus affecting the overall operating efficiency of the photovoltaic system.

[0007] Therefore, how to provide a solar photovoltaic power generation monitoring and scheduling method is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0008] One object of the present invention is to propose a solar photovoltaic power generation monitoring and scheduling method, which aims to solve the deficiencies of existing photovoltaic power generation systems in scheduling, prediction, system management and equipment maintenance. At present, photovoltaic power generation systems usually rely on traditional scheduling methods and lack real-time response to environmental changes and system dynamics, resulting in inaccurate power generation prediction, low system scheduling efficiency, and delayed equipment fault prediction. The present invention realizes power generation prediction and scheduling optimization based on real-time data by combining hybrid graph optimization algorithm, adaptive neural network, enhanced long short-term memory network and graph convolution network, which not only improves the accuracy of prediction, but also can dynamically adjust the system's scheduling strategy and resource allocation, and improve the operating efficiency of photovoltaic power generation system. In addition, an improved variational Bayes prediction algorithm is used to monitor the health status of photovoltaic system equipment and predict faults to ensure the long-term stable operation of the equipment. Through intelligent scheduling and optimization algorithms, the present invention can improve the coordination ability of photovoltaic system and power grid, and ensure the stable operation of the system under fluctuations in grid demand and photovoltaic power generation.

[0009] A solar photovoltaic power generation monitoring and scheduling method according to an embodiment of the present invention comprises the following steps:

[0010] S1. Collect real-time operation data of solar photovoltaic power generation system;

[0011] S2. Based on real-time operation data, the hybrid graph optimization algorithm is used to model the components of the photovoltaic power generation system, construct the graph structure of the photovoltaic system components, perform multi-objective scheduling optimization, and generate scheduling strategies and resource allocation plans among the components;

[0012] S3. Taking the dispatch strategy as input, applying adaptive neural network, combined with enhanced long short-term memory network and graph convolution network, iteratively optimizes the dispatch strategy, predicts and analyzes the changes in the operating status of the photovoltaic system based on real-time data, dynamically adjusts the dispatch strategy, and outputs the optimized dispatch plan;

[0013] S4. Based on the optimized scheduling scheme, the time-varying adaptive optimization algorithm is used to dynamically optimize the power generation and energy storage strategies of the photovoltaic system by combining the real-time load forecast results with the principle of maximizing marginal utility, output the optimized power generation power and energy storage strategy parameters, and update the relevant resource allocation scheme;

[0014] S5. Based on the optimized power generation and energy storage strategy parameters, the improved variational Bayesian prediction algorithm is applied to monitor the health status of each component of the photovoltaic power generation system in real time and predict faults, output fault warning information, and provide equipment maintenance suggestions in combination with system operation data;

[0015] S6. Utilize fault warning information and health status data, adopt a multi-objective nonlinear optimization algorithm to optimize the charging and discharging strategy of the energy storage device, and output an optimized energy storage management plan;

[0016] S7. Based on the optimized energy storage management solution, the simulated annealing algorithm is used to finely optimize the system scheduling, and various operating parameters are adjusted in combination with the changes in grid demand and the fluctuations in photovoltaic power generation, and the final optimized scheduling strategy is output;

[0017] S8. The intelligent dispatching system monitors the dispatching process in real time, combines the improved multivariate adaptive control algorithm to evaluate and optimize the system dispatching results in real time, and further adjusts the dispatching plan according to the system feedback signal and environmental changes.

[0018] Optionally, the S2 specifically includes:

[0019] S21. Based on real-time operation data, each component of the system is defined as a node in a graph structure. The node attributes include real-time operation status and performance parameters. The nodes are connected by edges, and the weight of the edges represents the influence relationship between the components.

[0020] S22. In the graph structure, the node status is dynamically updated according to the operation data and performance characteristics of the node, and the weight of the edge is determined by the degree of data association between components, power transmission efficiency, and the degree of mutual synergy;

[0021] S23. Construct objective function to achieve multi-objective scheduling optimization of photovoltaic system:

[0022]

[0023] Among them, f(x) is the objective function, x is the scheduling parameter, and w i is the target weight, c i (x) is the i-th objective function, n is the number of optimization objectives;

[0024] S24. Based on the constructed graph structure and objective function, a hybrid graph optimization algorithm is used to iteratively optimize the graph. The hybrid graph optimization algorithm dynamically updates the attributes of nodes and edges, evaluates the fitness of the current scheduling scheme, generates a candidate set of scheduling strategies, and selects the scheduling scheme with the highest fitness.

[0025] S25. In iterative optimization, multiple scheduling schemes are initialized by introducing the particle swarm optimization method. Each scheme is represented as a particle, and its fitness function is determined by the objective function value and the penalty term of the constraint condition:

[0026]

[0027] in, To update the rules, is the position of the ith particle in the tth iteration, is the particle velocity, pbest i is the historical optimal solution of the particle, gbest is the global optimal solution, c1 and c2 are learning factors, r1 and r2 are random variables, and w is the inertia weight;

[0028] S26. After the iterative optimization is completed, the optimal scheduling strategy and resource allocation plan among the components are generated. The scheduling strategy includes the power output optimization of the photovoltaic panels, the charging and discharging plan of the energy storage equipment, and the power balance allocation plan of the grid load; the resource allocation plan includes the energy transfer ratio between photovoltaic power generation and the energy storage system, the component operation schedule, and the backup plan for dealing with power fluctuations.

[0029] Optionally, the S3 specifically includes:

[0030] S31, taking the scheduling strategy as input, constructing an adaptive neural network model, the neural network includes multiple layers, each layer contains multiple neurons, and the layers are connected by activation functions;

[0031] S32. Take the real-time operation data of the current photovoltaic system as input, use the enhanced long short-term memory network to process the time series data, capture the dynamic characteristics of the system state changing over time, apply the graph convolutional network to analyze the relationship between the components in the photovoltaic system, explore the potential correlation between the components and their impact on the overall system operation state, and generate a spatiotemporal feature vector;

[0032] S33, inputting the generated spatiotemporal feature vector into an adaptive neural network, the neural network is trained by a back propagation algorithm, and the scheduling strategy is continuously optimized. Each layer in the neural network adjusts the weight according to the output information of the previous layer, and the optimization objective function is to minimize the scheduling error;

[0033]

[0034] Among them, the scheduling error function is L, is the scheduling result predicted by the neural network, y i is the actual scheduling demand, N is the number of samples;

[0035] S34, through the trained adaptive neural network, combined with the real-time feedback operation data, the parameters of the photovoltaic system's power generation, the charging and discharging scheduling of the energy storage equipment, and the grid load are optimized, and a new scheduling strategy is generated, which is compared with the current photovoltaic system operation status, and the strategy is further adjusted to adapt to the changing environmental conditions;

[0036] S35. Through repeated iterative training and real-time data feedback, the neural network can dynamically adjust its structure and weights and output the final optimized scheduling plan;

[0037] S36. Input the optimized dispatching plan into the photovoltaic power generation system, implement the optimized dispatching strategy through the real-time monitoring and control system, and continuously optimize the dispatching strategy in real time according to environmental changes and feedback signals.

[0038] Optionally, the S4 specifically includes:

[0039] S41. Based on the generated optimal dispatching scheme, the power generation and energy storage strategies of the system are optimized by using a time-varying adaptive optimization algorithm, combined with real-time load forecasting data and the principle of maximizing marginal utility;

[0040] S42. Setting multi-objective optimization of photovoltaic power generation system:

[0041] F=α1feff+α2fcharge+α3fload+α4fhealth;

[0042] Where F is the objective function, α1, α2, α3, α4 are the weight coefficients of power generation efficiency, energy storage efficiency, load balance and equipment health, respectively, and f eff is the power generation efficiency, f charge is the energy storage efficiency, f load For load balancing, f health An indicator of the health status of the device;

[0043] S43, adjusting the photovoltaic power generation and energy storage power based on historical load data, real-time meteorological data, light intensity and grid load demand, combined with the real-time load forecasting results based on the time series forecasting model;

[0044] S44. Optimize the configuration of power generation and energy storage according to load forecast results and resource constraints:

[0045]

[0046] Where U(Ppv ,E storage ) represents the marginal utility of power generation and energy storage, is the partial derivative of the generated power with respect to the objective function, is the partial derivative of the energy storage capacity with respect to the objective function;

[0047] S45. Based on the optimized power generation and energy storage solutions, a dynamic programming algorithm is used to further optimize the scheduling strategy. The algorithm optimizes the resource allocation of the system at multiple time points by making decisions in stages. By discretizing the time window, the decision problem in each time period is transformed into a local optimal problem, and finally the global optimal scheduling solution is solved by backtracking method.

[0048] S46. Based on real-time operation data and environmental changes, the time-varying adaptive optimization algorithm continuously adjusts the power generation and energy storage strategies, and updates the scheduling plan in real time;

[0049] S47: Output the optimized power generation power and energy storage strategy parameters, and use them as input to provide support for the optimization of the scheduling plan.

[0050] Optionally, the S5 specifically includes:

[0051] S51. According to the optimized power generation and energy storage strategy parameters, combined with the real-time operating status of each component of the photovoltaic system, a health status monitoring model of each component of the photovoltaic system is constructed, and the health status probability distribution of each component is initialized;

[0052] S52. The improved variational Bayes prediction algorithm is used to dynamically estimate the health status of each component of the photovoltaic system, and the posterior probability of the health status of each component is calculated by combining the working environment, load condition and historical data information of each component:

[0053]

[0054] Among them, P(H|D) is the posterior probability of the health state H under the given data D, P(D|H) is the likelihood function under the health state, P(H) is the prior probability of the health state, and P(D) is the marginal probability of the data;

[0055] S53. Based on the posterior probability of the health status of each component, the health status information of the component is updated using Bayesian inference, and the failure mode is predicted in combination with historical data and system status:

[0056]

[0057] Among them, P fault (C i ) is component C i The failure probability, P(Hj |D) is the health status H j The posterior probability, P fault (H j ) is the health status H j The corresponding failure probability, m is the number of failure modes;

[0058] S54. Combined with the failure mode prediction results, the remaining life of each component and the time point of failure occurrence are calculated, early warning information of the component is generated, and the maintenance strategy of the system is adjusted according to the failure probability of the component:

[0059] Pmaintenance(Ci)=Pfault(Ci)·Tremaining(Ci);

[0060] Among them, P maintenance (C i ) is component C i Maintenance recommendations, P fault (C i ) is the probability of component failure, T remaining (C i ) is the remaining useful life of the component;

[0061] S55. Update the resource allocation plan of the photovoltaic system according to the fault warning information, remaining life and maintenance suggestions, adjust the power generation and energy storage strategies of the system, and output a new resource allocation and scheduling plan in combination with load changes and system requirements;

[0062] S56. By continuously tracking and monitoring the health status, operating data and environmental changes of each component of the photovoltaic system, the fault warning information is updated in real time, and the system scheduling strategy is further optimized based on the feedback information.

[0063] Optionally, the S6 specifically includes:

[0064] S61. Using the fault warning information and health status data, evaluate the health status of the battery and converter components of the energy storage device to obtain a health score of the energy storage device;

[0065] S62. Based on the health status assessment, dynamically adjust the charging power upper limit and discharging power lower limit parameters according to the charging and discharging capacity and current health status of the energy storage device;

[0066] S63. Establish a multi-objective optimization model for charging and discharging strategies based on the load demand, power generation capacity and health status of the photovoltaic system;

[0067] S64, using a multi-objective optimization algorithm, combined with the load demand of the photovoltaic power generation system and the health status of the energy storage device, calculate the optimal energy storage charging and discharging power in each time period, and generate an energy storage charging and discharging control plan based on the calculation results;

[0068] S65. According to the multi-objective optimization calculation results, the charging and discharging strategies of the energy storage device are adjusted, and the key operating parameters of the battery voltage, current and temperature of the energy storage device are monitored in real time;

[0069] S66. Continuously monitor the charging and discharging process of the energy storage device and the load demand of the photovoltaic power generation system, collect real-time feedback data, further adjust the charging and discharging strategy, and dynamically adjust the multi-objective optimization model based on the feedback data.

[0070] Optionally, the S7 specifically includes:

[0071] S71. Formulate a charging and discharging strategy for the energy storage device based on the optimized energy storage management plan;

[0072] S72. Establish a dispatch optimization model based on the real-time photovoltaic power generation, the charging and discharging status of the energy storage device, and the grid load demand;

[0073] S73. Use simulated annealing algorithm to search for the optimal scheduling solution through iterative optimization:

[0074]

[0075] Among them, f(x) is the objective function, which represents the overall optimization degree of the system, P pv (t) is the photovoltaic power generation at time t, P load (t) is the grid load power at time t, P storage (t) is the charging / discharging power of the energy storage device at time t, is the maximum power limit of the energy storage device, α and β are weight coefficients, which control the matching priority between photovoltaic power generation and energy storage power and load respectively;

[0076] S74, adjusting the scheduling parameters of the photovoltaic power generation system and the energy storage device according to the calculation results of the simulated annealing algorithm;

[0077] S75. Adjust the charging and discharging operation of the energy storage device according to the optimization scheduling result;

[0078] S76. Monitor the dispatching plan in real time and make automatic adjustments to cope with changes in grid load.

[0079] The beneficial effects of the present invention are:

[0080] First of all, the solar photovoltaic power generation monitoring and scheduling method based on deep learning and optimization algorithm provided by the present invention significantly improves the intelligence level and operation efficiency of photovoltaic power generation system in many aspects. First of all, by adopting hybrid graph optimization algorithm and adaptive neural network, the present invention can accurately model the various components of photovoltaic power generation system and realize multi-objective scheduling optimization based on real-time data. This optimization method can comprehensively consider multiple factors such as power generation efficiency, battery health, grid load, etc., so as to dynamically adjust the resource allocation plan of photovoltaic system, avoiding the deficiency of traditional scheduling method that cannot respond to environmental changes in real time.

[0081] Secondly, the present invention effectively solves the common dynamic scheduling problems in photovoltaic power generation systems by introducing enhanced long short-term memory networks and graph convolutional networks. It can predict the operating status of photovoltaic systems in real time and make adaptive adjustments to environmental changes. Compared with traditional static scheduling methods, the present invention can make quick decisions based on real-time data, greatly improving the accuracy and response speed of scheduling.

[0082] In addition, the present invention uses an improved variational Bayesian prediction algorithm to monitor the health and predict faults of each component of the photovoltaic system, thereby improving the reliability of the system. By real-time monitoring and predicting equipment failures, early warning and maintenance can be carried out before problems occur, significantly reducing the system failure rate and extending the service life of the equipment.

[0083] Finally, the present invention optimizes the charging and discharging strategy of the energy storage device through a multi-objective optimization algorithm and a time-varying adaptive optimization strategy, ensuring the best match between the photovoltaic system's power generation and the grid load. This not only improves the system's energy utilization efficiency, but also enhances the grid's stability and anti-interference ability. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0085] Figure 1 A flow chart of a solar photovoltaic power generation monitoring and scheduling method proposed by the present invention;

[0086] Figure 2 A schematic diagram of a hybrid graph optimization algorithm for a solar photovoltaic power generation monitoring and scheduling method proposed in the present invention. DETAILED DESCRIPTION

[0087] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0088] refer to Figure 1and Figure 2 , a solar photovoltaic power generation monitoring and scheduling method, comprising the following steps:

[0089] S1. Collect real-time operation data of solar photovoltaic power generation system;

[0090] S2. Based on real-time operation data, the hybrid graph optimization algorithm is used to model the components of the photovoltaic power generation system, construct the graph structure of the photovoltaic system components, perform multi-objective scheduling optimization, and generate scheduling strategies and resource allocation plans among the components;

[0091] S3. Taking the dispatch strategy as input, applying adaptive neural network, combined with enhanced long short-term memory network and graph convolution network, iteratively optimizes the dispatch strategy, predicts and analyzes the changes in the operating status of the photovoltaic system based on real-time data, dynamically adjusts the dispatch strategy, and outputs the optimized dispatch plan;

[0092] S4. Based on the optimized scheduling scheme, the time-varying adaptive optimization algorithm is used to dynamically optimize the power generation and energy storage strategies of the photovoltaic system by combining the real-time load forecast results with the principle of maximizing marginal utility, output the optimized power generation power and energy storage strategy parameters, and update the relevant resource allocation scheme;

[0093] S5. Based on the optimized power generation and energy storage strategy parameters, the improved variational Bayesian prediction algorithm is applied to monitor the health status of each component of the photovoltaic power generation system in real time and predict faults, output fault warning information, and provide equipment maintenance suggestions in combination with system operation data;

[0094] S6. Utilize fault warning information and health status data, adopt a multi-objective nonlinear optimization algorithm to optimize the charging and discharging strategy of the energy storage device, and output an optimized energy storage management plan;

[0095] S7. Based on the optimized energy storage management solution, the simulated annealing algorithm is used to finely optimize the system scheduling, and various operating parameters are adjusted in combination with the changes in grid demand and the fluctuations in photovoltaic power generation, and the final optimized scheduling strategy is output;

[0096] S8. The intelligent dispatching system monitors the dispatching process in real time, combines the improved multivariate adaptive control algorithm to evaluate and optimize the system dispatching results in real time, and further adjusts the dispatching plan according to the system feedback signal and environmental changes.

[0097] In this implementation, S2 specifically includes:

[0098] S21. Based on real-time operation data, each component of the system is defined as a node in a graph structure. The node attributes include real-time operation status and performance parameters. The nodes are connected by edges, and the weight of the edges represents the influence relationship between the components.

[0099] S22. In the graph structure, the node status is dynamically updated according to the operation data and performance characteristics of the node, and the weight of the edge is determined by the degree of data association between components, power transmission efficiency, and the degree of mutual synergy;

[0100] S23. Construct objective function to achieve multi-objective scheduling optimization of photovoltaic system:

[0101]

[0102] Among them, f(x) is the objective function, x is the scheduling parameter, and w i is the target weight, c i (x) is the i-th objective function, n is the number of optimization objectives;

[0103] S24. Based on the constructed graph structure and objective function, a hybrid graph optimization algorithm is used to iteratively optimize the graph. The hybrid graph optimization algorithm dynamically updates the attributes of nodes and edges, evaluates the fitness of the current scheduling scheme, generates a candidate set of scheduling strategies, and selects the scheduling scheme with the highest fitness.

[0104] S25. In iterative optimization, multiple scheduling schemes are initialized by introducing the particle swarm optimization method. Each scheme is represented as a particle, and its fitness function is determined by the objective function value and the penalty term of the constraint condition:

[0105]

[0106] in, To update the rules, is the position of the ith particle in the tth iteration, is the particle velocity, pbest i is the historical optimal solution of the particle, gbest is the global optimal solution, c1 and c2 are learning factors, r1 and r2 are random variables, and w is the inertia weight;

[0107] S26. After the iterative optimization is completed, the optimal scheduling strategy and resource allocation plan among the components are generated. The scheduling strategy includes the power output optimization of the photovoltaic panels, the charging and discharging plan of the energy storage equipment, and the power balance allocation plan of the grid load; the resource allocation plan includes the energy transfer ratio between photovoltaic power generation and the energy storage system, the component operation schedule, and the backup plan for dealing with power fluctuations.

[0108] In this implementation, S3 specifically includes:

[0109] S31, taking the scheduling strategy as input, constructing an adaptive neural network model, the neural network includes multiple layers, each layer contains multiple neurons, and the layers are connected by activation functions;

[0110] S32. Take the real-time operation data of the current photovoltaic system as input, use the enhanced long short-term memory network to process the time series data, capture the dynamic characteristics of the system state changing over time, apply the graph convolutional network to analyze the relationship between the components in the photovoltaic system, explore the potential correlation between the components and their impact on the overall system operation state, and generate a spatiotemporal feature vector;

[0111] S33, inputting the generated spatiotemporal feature vector into an adaptive neural network, the neural network is trained by a back propagation algorithm, and the scheduling strategy is continuously optimized. Each layer in the neural network adjusts the weight according to the output information of the previous layer, and the optimization objective function is to minimize the scheduling error;

[0112]

[0113] Among them, the scheduling error function is L, is the scheduling result predicted by the neural network, y i is the actual scheduling demand, N is the number of samples;

[0114] S34, through the trained adaptive neural network, combined with the real-time feedback operation data, the parameters of the photovoltaic system's power generation, the charging and discharging scheduling of the energy storage equipment, and the grid load are optimized, and a new scheduling strategy is generated, which is compared with the current photovoltaic system operation status, and the strategy is further adjusted to adapt to the changing environmental conditions;

[0115] S35. Through repeated iterative training and real-time data feedback, the neural network can dynamically adjust its structure and weights and output the final optimized scheduling plan;

[0116] S36. Input the optimized dispatching plan into the photovoltaic power generation system, implement the optimized dispatching strategy through the real-time monitoring and control system, and continuously optimize the dispatching strategy in real time according to environmental changes and feedback signals.

[0117] In this implementation, S4 specifically includes:

[0118] S41. Based on the generated optimal dispatching scheme, the power generation and energy storage strategies of the system are optimized by using a time-varying adaptive optimization algorithm, combined with real-time load forecasting data and the principle of maximizing marginal utility;

[0119] S42. Setting multi-objective optimization of photovoltaic power generation system:

[0120] F=α1feff+α2fcharge+α3fload+α4fhealth;

[0121] Where F is the objective function, α1, α2, α3, α4 are the weight coefficients of power generation efficiency, energy storage efficiency, load balance and equipment health, respectively, and feff is the power generation efficiency, f charge is the energy storage efficiency, f load For load balancing, f health An indicator of the health status of the device;

[0122] S43, adjusting the photovoltaic power generation power and energy storage power based on historical load data, real-time meteorological data, light intensity and grid load demand, combined with the real-time load forecasting results based on the time series forecasting model;

[0123] S44. Optimize the configuration of power generation and energy storage according to load forecast results and resource constraints:

[0124]

[0125] Where U(P pv ,E storage ) represents the marginal utility of power generation and energy storage, is the partial derivative of the generated power with respect to the objective function, is the partial derivative of the energy storage capacity with respect to the objective function;

[0126] S45. Based on the optimized power generation and energy storage solutions, a dynamic programming algorithm is used to further optimize the scheduling strategy. The algorithm optimizes the resource allocation of the system at multiple time points by making decisions in stages. By discretizing the time window, the decision problem in each time period is transformed into a local optimal problem, and finally the global optimal scheduling solution is solved by backtracking method.

[0127] S46. Based on real-time operation data and environmental changes, the time-varying adaptive optimization algorithm continuously adjusts the power generation and energy storage strategies, and updates the scheduling plan in real time;

[0128] S47: Output the optimized power generation power and energy storage strategy parameters, and use them as input to provide support for the optimization of the scheduling plan.

[0129] In this implementation manner, S5 specifically includes:

[0130] S51. According to the optimized power generation and energy storage strategy parameters, combined with the real-time operating status of each component of the photovoltaic system, a health status monitoring model of each component of the photovoltaic system is constructed, and the health status probability distribution of each component is initialized;

[0131] S52. The improved variational Bayes prediction algorithm is used to dynamically estimate the health status of each component of the photovoltaic system, and the posterior probability of the health status of each component is calculated by combining the working environment, load condition and historical data information of each component:

[0132]

[0133] Among them, P(H|D) is the posterior probability of the health state H under the given data D, P(D|H) is the likelihood function under the health state, P(H) is the prior probability of the health state, and P(D) is the marginal probability of the data;

[0134] S53. Based on the posterior probability of the health status of each component, the health status information of the component is updated using Bayesian inference, and the failure mode is predicted in combination with historical data and system status:

[0135]

[0136] Among them, P fault (C i ) is component C i The failure probability, P(H j |D) is the health status H j The posterior probability, P fault (H j ) is the health status H j The corresponding failure probability, m is the number of failure modes;

[0137] S54. Combined with the failure mode prediction results, the remaining life of each component and the time point of failure occurrence are calculated, early warning information of the component is generated, and the maintenance strategy of the system is adjusted according to the failure probability of the component:

[0138] Pmaintenance(Ci)=Pfault(Ci)·Tremaining(Ci);

[0139] Among them, P maintenance (C i ) is component C i Maintenance recommendations, P fault (C i ) is the probability of component failure, T remaining (C i ) is the remaining useful life of the component;

[0140] S55. Update the resource allocation plan of the photovoltaic system according to the fault warning information, remaining life and maintenance suggestions, adjust the power generation and energy storage strategies of the system, and output a new resource allocation and scheduling plan in combination with load changes and system requirements;

[0141] S56. By continuously tracking and monitoring the health status, operating data and environmental changes of each component of the photovoltaic system, the fault warning information is updated in real time, and the system scheduling strategy is further optimized based on the feedback information.

[0142] In this implementation manner, S6 specifically includes:

[0143] S61. Using the fault warning information and health status data, evaluate the health status of the battery and converter components of the energy storage device to obtain a health score of the energy storage device;

[0144] S62. Based on the health status assessment, dynamically adjust the charging power upper limit and discharging power lower limit parameters according to the charging and discharging capacity and current health status of the energy storage device;

[0145] S63. Establish a multi-objective optimization model for charging and discharging strategies based on the load demand, power generation capacity and health status of the photovoltaic system;

[0146] S64, using a multi-objective optimization algorithm, combined with the load demand of the photovoltaic power generation system and the health status of the energy storage device, calculate the optimal energy storage charging and discharging power in each time period, and generate an energy storage charging and discharging control plan based on the calculation results;

[0147] S65. According to the multi-objective optimization calculation results, the charging and discharging strategies of the energy storage device are adjusted, and the key operating parameters of the battery voltage, current and temperature of the energy storage device are monitored in real time;

[0148] S66. Continuously monitor the charging and discharging process of the energy storage device and the load demand of the photovoltaic power generation system, collect real-time feedback data, further adjust the charging and discharging strategy, and dynamically adjust the multi-objective optimization model based on the feedback data.

[0149] In this implementation manner, the S7 specifically includes:

[0150] S71. Formulate a charging and discharging strategy for the energy storage device based on the optimized energy storage management plan;

[0151] S72. Establish a dispatch optimization model based on the real-time photovoltaic power generation, the charging and discharging status of the energy storage device, and the grid load demand;

[0152] S73. Use simulated annealing algorithm to search for the optimal scheduling solution through iterative optimization:

[0153]

[0154] Among them, f(x) is the objective function, which represents the overall optimization degree of the system, P pv (t) is the photovoltaic power generation at time t, P load (t) is the grid load power at time t, P storage (t) is the charging / discharging power of the energy storage device at time t, is the maximum power limit of the energy storage device, α and β are weight coefficients, which control the matching priority between photovoltaic power generation and energy storage power and load respectively;

[0155] S74, adjusting the scheduling parameters of the photovoltaic power generation system and the energy storage device according to the calculation results of the simulated annealing algorithm;

[0156] S75. Adjust the charging and discharging operation of the energy storage device according to the optimization scheduling result;

[0157] S76. Monitor the dispatching plan in real time and make automatic adjustments to cope with changes in grid load.

[0158] Embodiment 1:

[0159] In order to verify the feasibility of the present invention in implementation, the present invention is applied to a large photovoltaic power station located in Inner Mongolia Autonomous Region, with a photovoltaic power generation system with a total installed capacity of 500MW. The power station includes multiple submodules distributed in different geographical areas. To ensure the efficient operation of the power station under different weather conditions, the system needs to monitor and optimize the power generation capacity of each component, the charging and discharging strategy of the energy storage equipment, and the load demand of the entire power grid in real time.

[0160] In the power station, the real-time operation data of the photovoltaic power generation system includes the power generation of the components, meteorological data, the health status of the equipment, etc. These data are collected in real time through sensors and monitoring equipment. Through the method of the present invention, the system can dynamically dispatch and optimize the photovoltaic power generation equipment, thereby improving the power generation efficiency and economic benefits of the power station.

[0161] Specifically, the system monitors various operating data of the photovoltaic system in real time, including the working status of photovoltaic modules, ambient temperature and humidity, solar radiation intensity, etc., by integrating sensors and data acquisition equipment. These data are input into the hybrid graph optimization algorithm of the present invention. The algorithm combines various components of the photovoltaic system, uses graph structure for modeling, and performs multi-objective scheduling optimization on the power generation system. Through the construction of the graph structure, the relationship and mutual influence between different components can be accurately reflected, thereby generating the optimal scheduling strategy and resource allocation plan.

[0162] Then, the system applies adaptive neural networks, combined with enhanced long short-term memory networks and graph convolutional networks, to analyze and predict real-time data and predict changes in the operating status of each component in the future. Based on these prediction results, the system dynamically adjusts the scheduling strategy. For example, if it is predicted that photovoltaic components in certain areas will be affected by clouds and reduce their power generation capacity, the system will adjust the resource allocation plan in time and transfer the load to other areas with more stable power generation. In addition, the system can also dynamically optimize the energy storage strategy based on real-time weather changes, load demand and other factors to maximize the output power of the photovoltaic system.

[0163] In the comparative experiment, we used the traditional scheduling method based on empirical rules to compare with the optimized scheduling method of the present invention. The data collection time was from January 2024 to December 2024.

[0164] Table 1 Comparison data between the scheduling scheme of the present invention and the traditional scheduling scheme

[0165]

[0166] It can be seen from the data that after adopting the optimized scheduling scheme of the present invention, the power generation efficiency of the photovoltaic power generation system is generally higher than that of the traditional scheme. For example, in January 2023, the power generation efficiency of the traditional scheduling scheme was 82.5%, while the scheme of the present invention was increased to 90.1%. This trend has been continued in all months. In June 2023, the power generation efficiency of the traditional scheme was 88.1%, while the scheme of the present invention was increased to 94.3%, indicating that under high light intensity, the present invention can give full play to the power generation potential of the photovoltaic system and timely dispatch the energy storage system to provide a stable power supply. As can be seen from the table, the scheduling scheme of the present invention is significantly higher than the traditional scheme in terms of power generation efficiency in each month, and this improvement is stable, showing good continuity and reliability. This shows that the optimization method can maintain a high system efficiency under various weather changes and load fluctuations.

[0167] The power generation efficiency of the optimized scheduling scheme of the present invention is generally improved by 7-12 percentage points, which can bring significant benefits to the photovoltaic power station in the long run. The improvement in power generation efficiency is particularly obvious in high-power generation months (such as June and September), further demonstrating that this method can effectively play its advantages in different seasons and different climate conditions.

[0168] In summary, under the same grid load demand, the photovoltaic power station using the optimized scheduling scheme of the present invention has significantly higher power generation efficiency in each month than the traditional scheduling scheme. This shows that the scheduling optimization method of the present invention can better cope with the dynamic changes of the photovoltaic power generation system, improve power generation efficiency, and effectively optimize the charging and discharging strategy of the energy storage system to reduce energy waste.

[0169] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A solar photovoltaic power generation monitoring and scheduling method, characterized in that: The steps include: S1. Collect real-time operation data of solar photovoltaic power generation system; S2. Based on real-time operation data, the hybrid graph optimization algorithm is used to model the components of the photovoltaic power generation system, construct the graph structure of the photovoltaic system components, perform multi-objective scheduling optimization, and generate scheduling strategies and resource allocation plans among the components; S3. Taking the dispatch strategy as input, applying adaptive neural network, combined with enhanced long short-term memory network and graph convolution network, iteratively optimizes the dispatch strategy, predicts and analyzes the changes in the operating status of the photovoltaic system based on real-time data, dynamically adjusts the dispatch strategy, and outputs the optimized dispatch plan; S4. Based on the optimized scheduling scheme, the time-varying adaptive optimization algorithm is used to dynamically optimize the power generation and energy storage strategies of the photovoltaic system by combining the real-time load forecast results with the principle of maximizing marginal utility, output the optimized power generation power and energy storage strategy parameters, and update the relevant resource allocation scheme; S5. Based on the optimized power generation and energy storage strategy parameters, the improved variational Bayesian prediction algorithm is applied to monitor the health status of each component of the photovoltaic power generation system in real time and predict faults, output fault warning information, and provide equipment maintenance suggestions in combination with system operation data; S6. Utilize fault warning information and health status data, adopt a multi-objective nonlinear optimization algorithm to optimize the charging and discharging strategy of the energy storage device, and output an optimized energy storage management plan; S7. Based on the optimized energy storage management solution, the simulated annealing algorithm is used to finely optimize the system scheduling, and various operating parameters are adjusted in combination with the changes in grid demand and the fluctuations in photovoltaic power generation, and the final optimized scheduling strategy is output; S8. The intelligent dispatching system monitors the dispatching process in real time, combines the improved multivariate adaptive control algorithm to evaluate and optimize the system dispatching results in real time, and further adjusts the dispatching plan according to the system feedback signal and environmental changes.

2. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S2 specifically includes: S21. Based on real-time operation data, each component of the system is defined as a node in a graph structure. The node attributes include real-time operation status and performance parameters. The nodes are connected by edges, and the weight of the edges represents the influence relationship between the components. S22. In the graph structure, the node status is dynamically updated according to the operation data and performance characteristics of the node, and the weight of the edge is determined by the degree of data association between components, power transmission efficiency, and the degree of mutual synergy; S23. Construct objective function to achieve multi-objective scheduling optimization of photovoltaic system: Among them, f(x) is the objective function, x is the scheduling parameter, and w i is the target weight, c i (x) is the i-th objective function, n is the number of optimization objectives; S24. Based on the constructed graph structure and objective function, a hybrid graph optimization algorithm is used to iteratively optimize the graph. The hybrid graph optimization algorithm dynamically updates the attributes of nodes and edges, evaluates the fitness of the current scheduling scheme, generates a candidate set of scheduling strategies, and selects the scheduling scheme with the highest fitness. S25. In iterative optimization, multiple scheduling schemes are initialized by introducing the particle swarm optimization method. Each scheme is represented as a particle, and its fitness function is determined by the objective function value and the penalty term of the constraint condition: in, To update the rules, is the position of the ith particle in the tth iteration, is the particle velocity, pbest i is the historical optimal solution of the particle, gbest is the global optimal solution, c1 and c2 are learning factors, r1 and r2 are random variables, and w is the inertia weight; S26. After the iterative optimization is completed, the optimal scheduling strategy and resource allocation plan among the components are generated. The scheduling strategy includes the power output optimization of the photovoltaic panels, the charging and discharging plan of the energy storage equipment, and the power balance allocation plan of the grid load; the resource allocation plan includes the energy transfer ratio between photovoltaic power generation and the energy storage system, the component operation schedule, and the backup plan for dealing with power fluctuations.

3. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S3 specifically includes: S31, taking the scheduling strategy as input, constructing an adaptive neural network model, the neural network includes multiple layers, each layer contains multiple neurons, and the layers are connected by activation functions; S32. Take the real-time operation data of the current photovoltaic system as input, use the enhanced long short-term memory network to process the time series data, capture the dynamic characteristics of the system state changing over time, apply the graph convolutional network to analyze the relationship between the components in the photovoltaic system, explore the potential correlation between the components and their impact on the overall system operation state, and generate a spatiotemporal feature vector; S33, inputting the generated spatiotemporal feature vector into an adaptive neural network, the neural network is trained by a back propagation algorithm, and the scheduling strategy is continuously optimized. Each layer in the neural network adjusts the weight according to the output information of the previous layer, and the optimization objective function is to minimize the scheduling error; Among them, the scheduling error function is L, is the scheduling result predicted by the neural network, y i is the actual scheduling demand, N is the number of samples; S34, through the trained adaptive neural network, combined with the real-time feedback operation data, the parameters of the photovoltaic system's power generation, the charging and discharging scheduling of the energy storage equipment, and the grid load are optimized, and a new scheduling strategy is generated, which is compared with the current photovoltaic system operation status, and the strategy is further adjusted to adapt to the changing environmental conditions; S35. Through repeated iterative training and real-time data feedback, the neural network can dynamically adjust its structure and weights and output the final optimized scheduling plan; S36. Input the optimized dispatching plan into the photovoltaic power generation system, implement the optimized dispatching strategy through the real-time monitoring and control system, and continuously optimize the dispatching strategy in real time according to environmental changes and feedback signals.

4. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S4 specifically includes: S41. Based on the generated optimal dispatching scheme, the power generation and energy storage strategies of the system are optimized by using a time-varying adaptive optimization algorithm, combined with real-time load forecasting data and the principle of maximizing marginal utility; S42. Setting multi-objective optimization of photovoltaic power generation system: F=α1feff+α2fcharge+α3fload+α4fhealth; Where F is the objective function, α1, α2, α3, α4 are the weight coefficients of power generation efficiency, energy storage efficiency, load balance and equipment health, respectively, and f eff is the power generation efficiency, f charge is the energy storage efficiency, f load For load balancing, f health An indicator of the health status of the device; S43, adjusting the photovoltaic power generation and energy storage power based on historical load data, real-time meteorological data, light intensity and grid load demand, combined with the real-time load forecasting results based on the time series forecasting model; S44. Optimize the configuration of power generation and energy storage according to load forecast results and resource constraints: Where U(P pv ,E storage ) represents the marginal utility of power generation and energy storage, is the partial derivative of the generated power with respect to the objective function, is the partial derivative of the energy storage capacity with respect to the objective function; S45. Based on the optimized power generation and energy storage solutions, a dynamic programming algorithm is used to further optimize the scheduling strategy. The algorithm optimizes the resource allocation of the system at multiple time points by making decisions in stages. By discretizing the time window, the decision problem in each time period is transformed into a local optimal problem, and finally the global optimal scheduling solution is solved by backtracking method. S46. Based on real-time operation data and environmental changes, the time-varying adaptive optimization algorithm continuously adjusts the power generation and energy storage strategies, and updates the scheduling plan in real time; S47: Output the optimized power generation power and energy storage strategy parameters, and use them as input to provide support for the optimization of the scheduling plan.

5. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S5 specifically includes: S51. According to the optimized power generation and energy storage strategy parameters, combined with the real-time operating status of each component of the photovoltaic system, a health status monitoring model of each component of the photovoltaic system is constructed, and the health status probability distribution of each component is initialized; S52. The improved variational Bayes prediction algorithm is used to dynamically estimate the health status of each component of the photovoltaic system, and the posterior probability of the health status of each component is calculated by combining the working environment, load condition and historical data information of each component: Among them, P(H|D) is the posterior probability of the health state H under the given data D, P(D|H) is the likelihood function under the health state, P(H) is the prior probability of the health state, and P(D) is the marginal probability of the data; S53. Based on the posterior probability of the health status of each component, the health status information of the component is updated using Bayesian inference, and the failure mode is predicted in combination with historical data and system status: Among them, P fault (C i ) is component C i The failure probability, P(H j |D) is the health status H j The posterior probability, P fault (H j ) is the health status H j The corresponding failure probability, m is the number of failure modes; S54. Combined with the failure mode prediction results, the remaining life of each component and the time point of failure occurrence are calculated, early warning information of the component is generated, and the maintenance strategy of the system is adjusted according to the failure probability of the component: Pmaintenance(Ci)=Pfault(Ci)·Tremaining(Ci); Among them, P maintenance (C i ) is component C i Maintenance recommendations, P fault (C i ) is the probability of component failure, T remaining (C i ) is the remaining useful life of the component; S55. Update the resource allocation plan of the photovoltaic system according to the fault warning information, remaining life and maintenance suggestions, adjust the power generation and energy storage strategies of the system, and output a new resource allocation and scheduling plan in combination with load changes and system requirements; S56. By continuously tracking and monitoring the health status, operating data and environmental changes of each component of the photovoltaic system, the fault warning information is updated in real time, and the system scheduling strategy is further optimized based on the feedback information.

6. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S6 specifically includes: S61. Using the fault warning information and health status data, evaluate the health status of the battery and converter components of the energy storage device to obtain a health score of the energy storage device; S62. Based on the health status assessment, dynamically adjust the charging power upper limit and discharging power lower limit parameters according to the charging and discharging capacity and current health status of the energy storage device; S63. Establish a multi-objective optimization model for charging and discharging strategies based on the load demand, power generation capacity and health status of the photovoltaic system; S64, using a multi-objective optimization algorithm, combined with the load demand of the photovoltaic power generation system and the health status of the energy storage device, calculate the optimal energy storage charging and discharging power in each time period, and generate an energy storage charging and discharging control plan based on the calculation results; S65. According to the multi-objective optimization calculation results, the charging and discharging strategies of the energy storage device are adjusted, and the key operating parameters of the battery voltage, current and temperature of the energy storage device are monitored in real time; S66. Continuously monitor the charging and discharging process of the energy storage device and the load demand of the photovoltaic power generation system, collect real-time feedback data, further adjust the charging and discharging strategy, and dynamically adjust the multi-objective optimization model based on the feedback data.

7. A solar photovoltaic power generation monitoring and scheduling method according to claim 1, characterized in that: The S7 specifically includes: S71. Formulate a charging and discharging strategy for the energy storage device based on the optimized energy storage management plan; S72. Establish a dispatch optimization model based on the real-time photovoltaic power generation, the charging and discharging status of the energy storage device, and the grid load demand; S73. Use simulated annealing algorithm to search for the optimal scheduling solution through iterative optimization: Among them, f(x) is the objective function, which represents the overall optimization degree of the system, P pv (t) is the photovoltaic power generation at time t, P load (t) is the grid load power at time t, P storage (t) is the charging / discharging power of the energy storage device at time t, is the maximum power limit of the energy storage device, α and β are weight coefficients, which control the matching priority between photovoltaic power generation and energy storage power and load respectively; S74, adjusting the scheduling parameters of the photovoltaic power generation system and the energy storage device according to the calculation results of the simulated annealing algorithm; S75. Adjust the charging and discharging operation of the energy storage device according to the optimization scheduling result; S76. Monitor the dispatching plan in real time and make automatic adjustments to cope with changes in grid load.

Images