Multi-source fusion new energy power generation control method

Abstract

The invention provides a multi-source fusion new energy power generation control method, which comprises the following steps: acquiring historical fault data and a preset fault evaluation rule, and evaluating the influence range and severity of a current fault to obtain a fault evaluation result; according to the fault evaluation result and a pre-established restart switching rule, obtaining the starting characteristics and the mutual dependency relationship of the energy subsystems, and generating a restart switching scheme; according to the generated restart switching scheme, transient response of the restart process is simulated and predicted, and if a prediction result exceeds a preset threshold value, the restart switching scheme is optimized and adjusted; the optimized restart switching scheme is issued to a control unit and executed, and operation parameters of all the energy subsystems are obtained and analyzed in real time; when it is detected that parameter abnormity exists in the restart process, early warning is triggered, a restart adjustment scheme is obtained and issued again to be executed, and the restart process is dynamically adjusted through an intelligent execution mechanism till all the energy subsystems are stably started and recovered to the normal operation state.

Description

Technical Field

[0001] The present invention relates to the field of renewable energy power generation technology, and in particular to a multi-source integrated renewable energy power generation control method. Background Art

[0002] New energy generally refers to renewable energy that is developed and utilized on the basis of new technologies, including solar energy, biomass energy, wind energy, geothermal energy, wave energy, ocean current energy and tidal energy to achieve power generation. New energy power generation has the characteristics of intermittency and volatility, and is more prone to failure than conventional energy power generation systems. Therefore, after the large-scale access to the power grid, new energy power generation systems need to be coordinated and cooperated, requiring them to continuously improve their adaptability and safe and stable control capabilities. For example, they need to be restarted as soon as possible after a failure occurs, in order to reduce the safety and stability risks brought by the connection of new energy power generation to the grid, and ultimately ensure the safe and stable operation of the power grid. The intelligent restart strategy in the control of multi-source fusion new energy power generation faces many technical challenges. First, the new energy power generation control system in the prior art cannot quickly diagnose the fault type and location after the fault occurs, nor can it assess the severity and impact range of the fault, and thus it is difficult to quickly generate an effective restart plan and execute the restart to minimize the impact and loss. Secondly, the new energy power generation control system in the prior art does not fully consider the differences in the startup characteristics and interdependence of different types of new energy when generating the restart plan, so the restart efficiency is not high and the stability is also poor. In addition, the startup time and load lifting rate of different types of power generation equipment are different, and the optimization of the restart sequence and time interval faces the problem of combinatorial explosion. At the same time, due to the intermittent and volatile nature of renewable energy output, the transient response during the restart process is difficult to accurately predict, which may cause secondary problems such as voltage and frequency fluctuations and power imbalance. Summary of the invention

[0003] In order to solve the restart control optimization problem existing in different types of renewable energy power generation, the present invention provides a multi-source fusion renewable energy power generation control method, which is applicable to the power generation fusion of at least two different energy subsystems and the self-control of energy subsystem restart due to fault, and mainly includes the following steps:

[0004] S1. Obtain historical fault data and preset fault assessment rules of the energy subsystem corresponding to the new energy fusion power generation system, conduct an assessment based on the impact scope and severity of the fault, and obtain a fault assessment result;

[0005] S2. Preset the restart switching rules, combine the fault assessment results, obtain the startup characteristics and interdependencies of the energy subsystems, and generate a restart switching scheme through a genetic algorithm, wherein the restart switching scheme includes a restart sequence and a time interval;

[0006] S3. For the generated restart switching scheme, simulate or emulate the transient response of the restart process to predict the voltage fluctuation and power imbalance problems that may occur during the restart process. If the prediction result exceeds the preset threshold, optimize and adjust the restart switching scheme to obtain an optimized restart switching scheme.

[0007] S4. Send the optimized restart switching plan to the relevant control unit for execution, obtain the operating parameters of each energy subsystem through the sensor network, and use the anomaly detection algorithm to analyze the operating parameters in real time to determine whether there is an anomaly; when it is detected that there is an abnormality in the restart process, trigger an early warning, and according to the type and severity of the abnormal parameters, derive a restart adjustment plan through rule-based reasoning, send the restart adjustment plan to the relevant control unit again, and dynamically adjust the restart process through the intelligent execution mechanism until each energy subsystem starts smoothly and returns to normal operation;

[0008] S5. After the restart is completed, various data of the restart process are obtained, and support vector machines are used to analyze and mine the data to extract key feature parameters. The support vector machine identifies the key influencing factors in the restart process by establishing a high-dimensional feature space, thereby optimizing and improving the fault assessment rules and restart switching rules, and improving the system's intelligent restart capability.

[0009] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0010] The present invention discloses a multi-source fusion new energy power generation control method. The method evaluates the impact of faults through a decision tree algorithm, generates a restart switching plan using a genetic algorithm, and simulates and predicts the restart process using dynamic simulation technology. During the restart process, the present invention monitors the operating parameters in real time, and can dynamically adjust the restart strategy when an abnormality is detected. After the restart is completed, the present invention also uses a support vector machine to analyze the restart data and extract key features to optimize fault assessment and restart rules. The present invention realizes intelligent evaluation, automatic restart and dynamic adjustment after an energy system failure, thereby improving the reliability and recovery efficiency of the system. Through data mining and machine learning, the present invention also has self-optimization capabilities, which can continuously improve the level of intelligent restart, providing a strong guarantee for the safe and stable operation of the energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flow chart of an embodiment of a multi-source fusion new energy power generation control method of the present invention.

[0012] Figure 2 It is a schematic diagram of an S1 implementation example in a multi-source fusion new energy power generation control method of the present invention.

[0013] Figure 3It is a schematic diagram of an S2 implementation example in a multi-source fusion new energy power generation control method of the present invention. DETAILED DESCRIPTION

[0014] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0015] like Figure 1-3 In this embodiment, a multi-source integrated new energy power generation control method may specifically include the following steps:

[0016] S1. Obtain the historical fault data and preset fault assessment rules of the energy subsystem corresponding to the new energy fusion power generation system, conduct an assessment based on the impact scope and severity of the fault, and obtain the fault assessment result.

[0017] According to the preset fault assessment rules, historical fault data is obtained, and relevant data of the current fault is extracted and preprocessed; data cleaning and feature engineering techniques are used to denoise and normalize the extracted fault data to obtain standardized fault feature data; through feature selection algorithms, such as information gain and chi-square test, key features that have a greater impact on the fault assessment results are screened out; based on the screened key features, a decision tree model is constructed, and appropriate parameters such as split nodes and leaf nodes are set; the preprocessed current fault data is input into the decision tree model, and the leaf node category to which the fault belongs is determined by recursively traversing the nodes of the decision tree; based on the classification results of the decision tree model and the preset fault severity threshold, the severity level of the current fault is determined; based on the fault severity level and the scope of impact, a final fault assessment report is generated as a decision-making basis for fault handling.

[0018] Specifically, according to the preset energy subsystem fault assessment rules, 500 fault record data in the past year are obtained from the historical fault database. For the current energy subsystem fault, relevant operating parameters such as temperature, pressure, flow, voltage, etc. are extracted, totaling 20 parameters. The extracted fault data is denoised using the median filter algorithm to remove outliers and noise interference. Then the data is normalized using the maximum and minimum normalization method, and the data is mapped to the [0,1] interval to obtain standardized fault feature data. The influence of each feature on the fault type is calculated using the information gain algorithm, and 10 features with information gain values ​​greater than 2 are selected as key features. The decision tree model is constructed using the ID3 algorithm, and the information gain threshold is set to 1 and the maximum tree depth is 5. The standardized feature data of the current fault is input into the decision tree model, and the nodes of the decision tree are recursively traversed to finally determine that the fault belongs to the leaf node of a severe fault. Combined with the preset fault severity threshold, a fault that lasts for more than 30 minutes and affects more than two subsystems is a severe fault, and the current fault is assessed as a severe fault. Based on the fault severity and impact scope analysis, a fault assessment report is generated, detailing the fault time, fault location, fault type, severity, impact scope and recommended handling measures, providing a decision-making basis for fault handling.

[0019] S2. According to the fault assessment results and the pre-established restart switching rules, the startup characteristics and interdependencies of the energy subsystems are obtained, and a restart switching scheme is generated through a genetic algorithm. The restart switching scheme includes the restart sequence and time interval. The pre-established restart switching rules include parameters such as the priority, load capacity and startup time of each subsystem. Specifically, the following steps are included:

[0020] S201. Obtain a list of affected energy subsystems based on the fault assessment results and determine the scope of subsystems that need to be restarted; S202. Obtain parameters such as priority, load capacity, and startup time of each subsystem from the preset restart switching rules; S203. Construct a subsystem dependency graph by analyzing the interdependencies between energy subsystems;

[0021] S204, using a genetic algorithm, taking the subsystem as a basic unit, and taking priority, load capacity and startup time as constraints, to generate a population of initial restart sequence and time interval;

[0022] S205, evaluating the restart switching schemes in the population, and calculating the fitness value of each scheme based on indicators such as fault recovery efficiency and system stability;

[0023] S206. Through genetic operations such as selection, crossover and mutation, the restart switching scheme is continuously optimized to obtain a new generation of population with a higher fitness value;

[0024] S207, repeat S205 and S206 until the termination condition is met, and obtain the optimal restart switching plan, including the restart sequence and time interval, which is used to guide the restart process of the energy subsystem.

[0025] Specifically, according to the fault assessment results, the list of affected energy subsystems is obtained, and the scope of subsystems that need to be restarted is determined. For example, the assessment results show that the power supply system, refrigeration system and lighting system are affected and need to be restarted. From the preset restart switching rules, the priority, load capacity and start-up time of each subsystem are obtained. The power supply system priority is 1, the load capacity is 500kW, and the start-up time is 5 minutes; the refrigeration system priority is 2, the load capacity is 200kW, and the start-up time is 10 minutes; the lighting system priority is 3, the load capacity is 50kW, and the start-up time is 2 minutes. By analyzing the interdependence between energy subsystems, a subsystem dependency graph is constructed. The power supply system is the root node, and the refrigeration system and lighting system depend on the power supply system. A genetic algorithm is used to generate a population of initial restart order and time interval with subsystems as the basic unit and priority, load capacity and start-up time as constraints. The population size is set to 50, the encoding method is integer encoding, and the number of gene bits is 3, representing the power supply, refrigeration and lighting systems respectively. The restart switching schemes in the population are evaluated, and the fitness value of each scheme is calculated based on indicators such as fault recovery efficiency and system stability. The fault recovery efficiency is measured by the sum of the restart time, and the system stability is measured by the standard deviation of the load capacity. The fitness function is designed as the weighted sum of the recovery efficiency and stability, with weights of 6 and 4 respectively. Through genetic operations such as selection, crossover and mutation, the restart switching scheme is continuously optimized to obtain a new generation of populations with higher fitness values. The selection operation uses roulette selection, the crossover operation uses single-point crossover with a crossover probability of 8, and the mutation operation uses basic bit mutation with a mutation probability of 1. Repeat the evaluation and genetic operations until the average fitness value of the population changes by less than 1 for 10 consecutive generations, and obtain the optimal restart switching scheme, including the restart order of the power supply system, the cooling system, and the lighting system, with time intervals of 0 minutes, 5 minutes, and 15 minutes, respectively, which is used to guide the restart process of the energy subsystem.

[0026] S3. For the generated restart switching scheme, dynamic simulation technology is used to simulate the transient response of the restart process to predict the voltage fluctuation and power imbalance problems that may occur during the restart process. If the prediction result exceeds the preset threshold, the restart switching scheme is optimized and adjusted to obtain an optimized restart switching scheme. The preset threshold is determined according to the stability requirements and safe operation standards of the energy subsystem, including indicators such as voltage fluctuation range and power balance. Specifically, the following steps are included:

[0027] S301, obtaining the restart switching plan and the stability requirements and safe operation standards of the energy subsystem, and determining the voltage fluctuation range and power balance threshold;

[0028] S302, constructing a dynamic simulation model of the energy subsystem according to the restart switching plan and setting simulation parameters;

[0029] S303, running a dynamic simulation model to simulate the transient response of the restart process and obtain voltage and power data;

[0030] S304, determine whether the voltage fluctuation and power balance exceed the preset threshold, if so, execute S305, otherwise execute S306; S305, use the particle swarm optimization algorithm, with the minimum voltage fluctuation and the optimal power balance as the optimization goal, to adjust the control parameters in the restart switching scheme;

[0031] Repeat S303-S304 until the voltage fluctuation and power balance meet the requirements;

[0032] S306: Output the optimized restart switching plan as the final restart strategy.

[0033] Specifically, we first need to obtain the stability requirements and safe operation standards of the restart switching scheme and the energy subsystem, determine that the voltage fluctuation range is ±5%, and the power balance threshold is 95%. Then, according to the restart switching scheme, MATLAB / Simulink is used to build a dynamic simulation model of the energy subsystem, and the simulation step is set to 1s and the simulation time is 10s. Then, the dynamic simulation model is run to simulate the transient response of the restart process and obtain voltage and power data. By statistically analyzing the voltage and power data, it is determined that the voltage fluctuation is ±6% and the power balance is 93%, which exceeds the preset threshold. Therefore, the particle swarm optimization algorithm is adopted, with the minimum voltage fluctuation and the optimal power balance as the optimization goals. By adjusting the control parameters in the restart switching scheme, such as adjusting the load switching time and adjusting the generator excitation voltage, the optimization is continuously iterated until the voltage fluctuation converges to ±5% and the power balance reaches 96%, meeting the stability requirements. Finally, the optimized restart switching scheme is output as the final restart strategy to ensure the safe and stable operation of the energy subsystem.

[0034] S4. Send the optimized restart switching plan to the relevant control unit, and make it execute the optimized restart switching plan. Obtain the operating parameters of each energy subsystem through the sensor network, and use the anomaly detection algorithm to analyze the operating parameters in real time to determine whether there is an abnormality. When it is detected that there is a parameter abnormality in the restart process, an early warning is triggered. According to the type and severity of the abnormal parameters, a restart adjustment plan is derived through rule-based reasoning. The restart adjustment plan is sent to the relevant control unit again, and the restart process is dynamically adjusted through the intelligent execution mechanism until each energy subsystem starts smoothly and returns to normal operation. The restart adjustment plan specifically includes the following steps:

[0035] S401. According to the optimized restart switching scheme, a restart instruction is issued to the control unit corresponding to each energy subsystem;

[0036] S402, after receiving the restart instruction, each control unit executes a corresponding restart operation to start each energy subsystem;

[0037] S403, through the sensor network deployed in each energy subsystem, the operating parameters of each key device and component are collected in real time; S404, the collected real-time operating parameters are transmitted to the anomaly detection module, and the parameters are analyzed by using an anomaly detection algorithm such as a class of SVM to determine whether there is an operating anomaly in each energy subsystem;

[0038] S405, when it is detected that a key parameter of a certain energy subsystem exceeds a preset threshold range, an early warning is triggered, and information such as the type and severity of the abnormal parameter is sent to the fault diagnosis module;

[0039] S406, after receiving the abnormal information, the fault diagnosis module uses a rule reasoning engine built based on the expert knowledge base to comprehensively analyze the abnormal parameters and operating conditions, and intelligently generates a targeted restart adjustment plan;

[0040] Then, the generated restart adjustment plan is sent to the corresponding control unit, and the control parameters during the restart process are dynamically adjusted through the intelligent execution mechanism to optimize the restart strategy until each energy subsystem starts smoothly and returns to normal operating state.

[0041] Specifically, according to the optimized restart switching scheme, the restart command is sent to the control unit corresponding to each energy subsystem through the CAN bus protocol. After receiving the restart command, each control unit executes the preset restart operation sequence, such as starting the photovoltaic array, wind turbine generator set and energy storage battery group in sequence, and the start time interval is set to 5 seconds. Through the ZigBee wireless sensor network deployed in each energy subsystem, the key parameters such as inverter output voltage, current and active power are collected in real time at a sampling frequency of 100ms. The collected real-time operating parameters are transmitted to the anomaly detection module through the MQTT protocol, and the parameters are analyzed by a class of SVM algorithm based on radial basis kernel function. By comparing with historical data, it is determined whether each energy subsystem has an operating anomaly. When it is detected that the key parameters of a certain energy subsystem exceed the preset threshold range, such as the inverter output voltage fluctuation exceeds 5% of the rated value, an early warning is triggered, and the type and severity of the abnormal parameters are encapsulated in JSON format and sent to the fault diagnosis module through the RESTful API interface. After receiving the abnormal information, the fault diagnosis module uses the expert knowledge base built based on the Drools rule engine to comprehensively analyze the abnormal parameters and the current environmental factors such as wind speed and light intensity, and uses a combination of forward reasoning and reverse reasoning to intelligently generate a targeted restart adjustment plan within 500ms. The generated restart adjustment plan is sent to the corresponding control unit through the OPC UA protocol, and the control parameters in the restart process are adaptively optimized through the PID control algorithm, such as dynamically adjusting the angle of the fan blade according to the actual wind speed, and finally achieving smooth startup of each energy subsystem and recovery to the rated working point within 30 seconds. S5. After the restart is completed, obtain various data of this restart process, use support vector machine to analyze and mine the data, and extract key feature parameters. The support vector machine establishes a high-dimensional feature space to identify the key influencing factors in the restart process, thereby optimizing and improving the fault assessment rules and restart switching rules, and improving the system's intelligent restart capability. Optimizing and improving fault assessment rules and restart switching rules includes the following steps:

[0042] S501, obtaining various monitoring data and log information during the system restart process, and preprocessing the data, including data cleaning, feature extraction, and data standardization, to obtain a standardized restart process data set;

[0043] S502: construct a high-dimensional feature space according to preset support vector machine model parameters, map the restart process data set to the feature space, perform classification and regression analysis on the data through the support vector machine algorithm, and identify key feature parameters that have a greater impact on the system restart process;

[0044] S503, analyzing the association rules between the identified key characteristic parameters and the system fault type and fault severity, and obtaining an optimized fault assessment rule base through an association rule mining algorithm;

[0045] S504, applying the optimized fault assessment rule base to the intelligent restart decision module, when a system failure occurs, automatically selecting the optimal restart switching strategy according to the fault assessment result, and dynamically adjusting the restart switching parameters;

[0046] S505, online analysis of the restart process data is performed, and the support vector machine model is dynamically updated through incremental learning, and the model parameters are continuously optimized to improve the accuracy of fault identification and restart decision-making;

[0047] S506. Establish a restart evaluation feedback mechanism to score the restart effect, use the scoring results as new training samples, and feed them back to the support vector machine model. Through continuous iterative optimization, the success rate and efficiency of the system's intelligent restart will be continuously improved. S507. Regularly test and verify the intelligent restart module, optimize the algorithm and adjust the rules for the problems found, ensure that the intelligent restart strategy is adapted to the actual system operating environment, and improve the system's self-healing ability and reliability.

[0048] Specifically, during the restart process of each energy subsystem, various monitoring data and log information were collected, and the data cleaning algorithm was used to remove noise and outliers, and 32 characteristic parameters related to the system restart process were extracted. The Min-Max normalization method was used to normalize the data to obtain a standardized restart process data set. A high-dimensional feature space was constructed based on the RBF kernel function, and the support vector machine model parameters were optimized by grid search and cross-validation methods. The data set was mapped to the feature space for classification and regression analysis, and 8 key characteristic parameters that had a greater impact on the restart process were identified, such as voltage fluctuation rate and current peak value. The Apriori association rule mining algorithm was used to analyze the association rules between these key parameters and the fault type and fault severity of each energy subsystem, and an optimized fault assessment rule base containing 120 rules was obtained. The optimized rule base was applied to the intelligent restart decision module. When a system fault occurs, the restart switching strategy is dynamically adjusted according to the fault assessment results, such as adjusting the restart timing and optimizing load distribution. During the restart switching process, the system operation status is monitored in real time with a sampling frequency of 200ms, and the support vector machine model is updated online using the incremental learning algorithm to continuously optimize the model parameters. At the same time, a restart evaluation feedback mechanism is established to score the effect of each restart with a score of 1-5 points. The scoring results are fed back to the model as new training samples for iterative optimization to improve the success rate of intelligent restart. The intelligent restart module is tested regularly using evaluation methods such as confusion matrix and ROC curve, and algorithm optimization and rule adjustment are performed for the problems found, such as introducing integrated learning methods to improve the accuracy of fault identification, optimizing the restart timing strategy to improve the system recovery efficiency, ensuring that the intelligent restart strategy is adapted to the actual operating environment, and maximizing the self-healing ability and reliability of the system.

[0049] What is disclosed above is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. A person skilled in the art can understand that all or part of the processes of the above embodiments and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A multi-source fusion new energy power generation control method, characterized in that: The method is applicable to power generation integration of at least two different energy subsystems and self-control of restarting of energy subsystems due to failure, and comprises the following steps: S1. Obtain historical fault data and preset fault assessment rules of the energy subsystem corresponding to the new energy fusion power generation system, conduct an assessment based on the impact scope and severity of the fault, and obtain a fault assessment result; S2. Preset the restart switching rules, combine the fault assessment results, obtain the startup characteristics and interdependencies of the energy subsystems, and generate a restart switching scheme through a genetic algorithm, wherein the restart switching scheme includes a restart sequence and a time interval; S3. For the generated restart switching scheme, simulate or emulate the transient response of the restart process to predict the voltage fluctuation and power imbalance problems that may occur during the restart process. If the prediction result exceeds the preset threshold, optimize and adjust the restart switching scheme to obtain an optimized restart switching scheme. S4. Send the optimized restart switching plan to the relevant control unit for execution, obtain the operating parameters of each energy subsystem through the sensor network, and use the anomaly detection algorithm to analyze the operating parameters in real time to determine whether there is an anomaly; when it is detected that there is an abnormality in the restart process, trigger an early warning, and according to the type and severity of the abnormal parameters, derive a restart adjustment plan through rule-based reasoning, send the restart adjustment plan to the relevant control unit again, and dynamically adjust the restart process through the intelligent execution mechanism until each energy subsystem starts smoothly and returns to normal operation; S5. After the restart is completed, various data of the restart process are obtained, and support vector machines are used to analyze and mine the data to extract key feature parameters. The support vector machine identifies the key influencing factors in the restart process by establishing a high-dimensional feature space, thereby optimizing and improving the fault assessment rules and restart switching rules, and improving the system's intelligent restart capability.

2. A multi-source fusion new energy power generation control method according to claim 1, characterized in that: The fault assessment result is automatically obtained based on a decision tree algorithm.

3. A multi-source fusion new energy power generation control method according to claim 2, characterized in that: The following steps are involved in obtaining fault assessment results using the decision tree algorithm: De-noising and normalizing the extracted historical fault data to obtain standardized fault feature data; Through the feature selection algorithm, the key features that have a greater impact on the fault assessment results are screened out; According to the selected key features, a decision tree model is constructed and appropriate split nodes and leaf nodes are set; Input the preprocessed current fault data into the decision tree model, and recursively traverse the nodes of the decision tree to determine the leaf node category to which the fault belongs; According to the classification results of the decision tree model and the preset fault severity threshold, the severity level of the current fault is determined; According to the fault severity level and impact scope, a final fault assessment report is generated as a decision-making basis for fault handling.

4. The multi-source fusion new energy power generation control method according to claim 1 is characterized in that: In step S2, the following steps are also included: S201. According to the fault assessment result, obtain a list of affected energy subsystems and determine the scope of subsystems that need to be restarted; S202, obtaining the priority, load capacity and startup time of each subsystem from the preset restart switching rules; S203, constructing a subsystem dependency diagram by analyzing the interdependencies between energy subsystems; S204, using a genetic algorithm, taking the subsystem as a basic unit, and taking priority, load capacity and startup time as constraints, to generate a population of initial restart sequence and time interval; S205, evaluating the restart switching schemes in the population, and calculating the fitness value of each scheme according to the fault recovery efficiency and system stability; S206, continuously optimizing the restart switching scheme through selection, crossover and mutation, and obtaining a new generation population with a higher fitness value; S207, repeat S205 and S206 until the termination condition is met, and obtain the optimal restart switching plan, including the restart sequence and time interval, which is used to guide the restart process of the energy subsystem.

5. The multi-source fusion new energy power generation control method according to claim 1 is characterized in that: Step S3 comprises the following steps: S301, obtaining the restart switching plan and the stability requirements and safe operation standards of the energy subsystem, and determining the voltage fluctuation range and power balance threshold; S302, constructing a dynamic simulation model of the energy subsystem according to the restart switching plan and setting simulation parameters; S303, running a dynamic simulation model to simulate the transient response of the restart process and obtain voltage and power data; S304, determine whether the voltage fluctuation and power balance exceed the preset threshold, if so, execute S305, otherwise execute S306; S305, using a particle swarm optimization algorithm, with minimum voltage fluctuation and optimal power balance as optimization goals, to adjust control parameters in the restart switching scheme; Repeat S303-S304 until the voltage fluctuation and power balance meet the requirements; S306: Output the optimized restart switching plan as the final restart strategy.

6. A multi-source fusion new energy power generation control method according to claim 1, characterized in that: In step S4, obtaining a restart adjustment plan includes the following steps: S401. According to the optimized restart switching scheme, a restart instruction is issued to the control unit corresponding to each energy subsystem; S402, after receiving the restart instruction, each control unit executes a corresponding restart operation and starts the corresponding energy subsystem; S403, collecting operating parameters of key equipment and components in real time through the sensor network deployed in each energy subsystem; S404, transmitting the collected real-time operating parameters to the anomaly detection module, using an anomaly detection algorithm to analyze the parameters, and determining whether each energy subsystem has an operating anomaly; S405, when it is detected that a key parameter of a certain energy subsystem exceeds a preset threshold range, an early warning is triggered, and information about the abnormal parameter is sent to a fault diagnosis module; S406. After receiving the abnormal information, the fault diagnosis module uses a rule reasoning engine built based on the expert knowledge base to comprehensively analyze the abnormal parameters and operating conditions, and intelligently generates a targeted restart adjustment plan.

7. The multi-source fusion new energy power generation control method according to claim 1 is characterized in that: In step S5, optimizing and improving the fault assessment rules and the restart switching rules includes the following steps: S501, obtaining various monitoring data and log information during the current system restart process, preprocessing the data, and obtaining a standardized restart process data set; S502: construct a high-dimensional feature space according to preset support vector machine model parameters, map the restart process data set to the feature space, perform classification and regression analysis on the data through the support vector machine algorithm, and identify key feature parameters that have a greater impact on the system restart process; S503, analyzing the association rules between the identified key characteristic parameters and the system fault type and fault severity, and obtaining an optimized fault assessment rule base through an association rule mining algorithm; S504: Apply the optimized fault assessment rule base to the intelligent restart decision module. When a system failure occurs, automatically select the optimal restart switching strategy and dynamically adjust the restart switching parameters according to the fault assessment result.

8. A multi-source fusion new energy power generation control method according to claim 7, characterized in that: Also includes: S505, online analysis of the restart process data is performed, and the support vector machine model is dynamically updated through incremental learning, and the model parameters are continuously optimized to improve the accuracy of fault identification and restart decision-making; S506. Establish a restart evaluation feedback mechanism to score the restart effect. Use the score results as new training samples and feed them back to the support vector machine model. Through continuous iterative optimization, the success rate and efficiency of the system intelligent restart are continuously improved. S507. Regularly test and verify the intelligent restart module, optimize the algorithm and adjust the rules according to the problems found, ensure that the intelligent restart strategy is compatible with the actual system operating environment, and improve the self-healing ability and reliability of the system.

Images