Multi-source power system scheduling optimization method and system based on parrot algorithm model

Abstract

The invention relates to the technical field of multi-energy power system scheduling optimization, in particular to a multi-source power system scheduling optimization method and system based on a parrot algorithm model. The method comprises the steps of constructing a running state data set, performing priority ranking, generating an initial scheduling scheme, extracting abnormal fluctuation characteristics, performing pattern recognition processing, and dynamically adjusting and generating an optimal scheduling scheme. The scheduling efficiency is improved through the multi-source energy collaborative framework, the fluctuation mode distribution is analyzed in combination with the deep learning model to dynamically adjust the scheduling scheme, and the adaptability and economical efficiency of the system are remarkably enhanced. The demand of a multi-energy power system for efficient and intelligent scheduling can be met, and the environmental protection property and stability of system operation are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-energy power system dispatch optimization, and specifically relates to a multi-source power system dispatch optimization method and system based on a Parrot algorithm model. Background Art

[0002] With the rapid development of multi-energy power systems, the importance of optimized dispatching methods in improving system economics and environmental performance has become increasingly prominent. In modern power systems, the coordinated optimization of multiple energy sources has become key, especially in complex multi-source scenarios. How to achieve efficient and intelligent dispatching has become a research hotspot. However, existing multi-energy power system optimization dispatching methods mostly rely on traditional algorithms or single intelligent algorithms. These methods have shortcomings in adaptability and efficiency, making it difficult to meet the high-precision and fast-response requirements of modern power systems.

[0003] After searching, it was found that the patent with publication number CN112580897B discloses a multi-energy power system optimization scheduling method based on the Parrot algorithm. By establishing a mathematical model that takes into account economic and environmental factors and using the Parrot algorithm for optimization scheduling, it has the advantages of high efficiency and accuracy. However, this technical solution still has certain limitations: first, the application of the Parrot algorithm is limited to the verification of three specific scheduling strategies, and does not fully consider the real-time response capability of the multi-source power system under dynamic conditions, which may lead to insufficient adaptability to emergencies in actual operation; second, the method simplifies the modeling of the coupling relationship of multiple energy sources, which may affect the global optimality of the scheduling results. The above problems show that the existing multi-energy power system optimization scheduling method based on the Parrot algorithm still needs to be improved in terms of dynamic condition adaptability, multi-source coupling modeling accuracy, and real-time response capability.

[0004] Therefore, in view of the shortcomings of the existing technology, there is an urgent need to provide an optimization scheduling method and system that can further improve the adaptability of dynamic working conditions, optimize the accuracy of multi-source coupling modeling and enhance real-time response capabilities, so as to better meet the needs of multi-energy power systems for efficient and intelligent scheduling. Summary of the Invention

[0005] The present invention provides a multi-source power system scheduling optimization method and system based on the Parrot algorithm model, the main purpose of which is to improve the adaptability of the multi-energy power system under dynamic conditions, optimize the accuracy of multi-source coupling modeling and enhance the real-time response capability.

[0006] To achieve the above objectives, the present invention provides a multi-source power system dispatch optimization method based on the Parrot algorithm model, comprising: Construct an operating status dataset of a multi-source power system and extract key operating parameters; Prioritizing the key operating parameters according to preset weights to obtain a priority parameter sequence; Utilizing a multi-source energy collaborative framework to optimize the allocation of the priority parameter sequence and generate an initial scheduling plan; Acquiring operational feedback data of the initial scheduling plan, and extracting abnormal fluctuation characteristics from the operational feedback data; Performing pattern recognition processing on the abnormal fluctuation characteristics to obtain a fluctuation pattern distribution; The initial scheduling plan is dynamically adjusted according to the fluctuation pattern distribution to generate an optimized scheduling plan.

[0007] Optionally, the prioritizing the key operating parameters according to preset weights to obtain a priority parameter sequence includes: Obtain contribution evaluation indicators of each energy type in a multi-energy power system; Performing weighted calculation on the key operating parameters according to the contribution evaluation index to obtain weighted parameter values; The key operating parameters are arranged in descending order according to the weighted parameter values ​​to generate a priority parameter sequence.

[0008] Optionally, the optimizing allocation of the priority parameter sequence using a multi-source energy coordination framework to generate an initial scheduling plan includes: Input the priority parameter sequence into the multi-source energy coordination framework and perform preliminary allocation through the resource allocation module within the framework; Conduct energy balance verification on the preliminary allocation results; If the preliminary allocation result fails to pass the energy balance check, the allocation strategy is adjusted and the preliminary allocation steps are performed again; If the preliminary allocation result passes the energy consumption balance check, an initial scheduling plan is generated.

[0009] Optionally, obtaining operation feedback data of the initial scheduling plan and extracting abnormal fluctuation characteristics in the operation feedback data include: Starting a real-time monitoring module of the multi-source power system to collect operating data of the initial scheduling plan; Performing time series analysis on the operating data to identify abnormal fluctuation points therein; Corresponding fluctuation features are extracted according to the abnormal fluctuation points to obtain an abnormal fluctuation feature set.

[0010] Optionally, performing pattern recognition processing on the abnormal fluctuation characteristics to obtain a fluctuation pattern distribution includes: Performing cluster analysis on the abnormal fluctuation feature set to divide the abnormal fluctuation pattern into several categories; The frequency of occurrence of each category in the abnormal fluctuation feature set is counted to generate a fluctuation pattern distribution graph.

[0011] Optionally, before dynamically adjusting the initial scheduling solution according to the fluctuation pattern distribution to generate an optimized scheduling solution, the method further includes: Collect a large number of scheduling cases with optimization annotations, and extract the key operating parameters and corresponding optimization results in the scheduling cases; Dividing the key operating parameters and optimization results in the scheduling case into training samples and test samples; Using a deep learning model to perform pattern fitting on the training samples to generate a training optimization plan; Evaluating the consistency of the training optimization scheme with a preset standard optimization scheme for the training sample; If the consistency is lower than a preset threshold, the pattern fitting step is re-executed after optimizing the structure of the deep learning model; If the consistency reaches or exceeds a preset threshold, the performance of the deep learning model is verified using the test sample; If the deep learning model fails the performance verification, then the step of extracting the key operating parameters and corresponding optimization results from the scheduling case is re-executed after increasing the data volume of the scheduling case; If the deep learning model passes the performance verification, the model training is confirmed to be complete and an optimized scheduling model is generated.

[0012] Optionally, the evaluating the consistency of the training optimization scheme with a standard optimization scheme preset for the training sample includes: Comparing the degree of difference between the training optimization scheme and the standard optimization scheme in key operating parameters; The consistency is quantified according to the degree of difference to generate a consistency score.

[0013] Optionally, dynamically adjusting the initial scheduling plan according to the fluctuation pattern distribution to generate an optimized scheduling plan includes: Analyzing the fluctuation pattern distribution using the optimization scheduling model to generate an adjustment strategy; Applying the adjustment strategy to the initial scheduling plan to generate an intermediate scheduling plan; Performing a security check on the intermediate scheduling plan; If the intermediate scheduling solution fails the security check, revising the adjustment strategy and then re-performing the step of applying the adjustment strategy to the initial scheduling solution to generate the intermediate scheduling solution; If the intermediate scheduling solution passes the security check, an optimized scheduling solution is generated.

[0014] In order to solve the above problems, the present invention also provides a multi-source power system dispatch optimization system based on the Parrot algorithm model, the system comprising: Data acquisition module, used to construct the operating status data set of multi-source power systems and extract key operating parameters; A priority sorting module is used to prioritize the key operating parameters according to preset weights to obtain a priority parameter sequence; a scheduling generation module, configured to optimize the allocation of the priority parameter sequence using a multi-source energy collaborative framework and generate an initial scheduling plan; a feature extraction module, configured to obtain operational feedback data of the initial scheduling plan and extract abnormal fluctuation features from the operational feedback data; A pattern recognition module is used to perform pattern recognition processing on the abnormal fluctuation characteristics to obtain a fluctuation pattern distribution; A dynamic adjustment module is used to dynamically adjust the initial scheduling plan according to the fluctuation pattern distribution to generate an optimized scheduling plan.

[0015] The embodiment of the present invention improves the pertinence of the scheduling process by constructing an operating status data set of a multi-source power system, extracting key operating parameters for constructing the operating status data set of the multi-source power system and prioritizing them; generates an initial scheduling plan using a multi-source energy collaborative framework, thereby enhancing the collaborative efficiency among multi-source energies; improves the dynamic adaptability of the scheduling plan by performing pattern recognition processing on abnormal fluctuation characteristics in operating feedback data; and finally generates an optimized scheduling plan in combination with an optimized scheduling model, thereby significantly improving the economy and environmental protection of the multi-energy power system. Therefore, the multi-source power system scheduling optimization method and system based on the Parrot algorithm model proposed in the present invention can meet the needs of the multi-energy power system for efficient and intelligent scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a multi-source power system dispatch optimization method based on the Parrot algorithm model provided in one embodiment of the present invention; Figure 2 A schematic diagram of a process for extracting abnormal fluctuation characteristics from operation feedback data provided by one embodiment of the present invention; Figure 3 This is a functional module diagram of a multi-source power system scheduling optimization system based on the Parrot algorithm model provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] The present invention provides a multi-source power system scheduling optimization method and system based on the Parrot algorithm model. The core of the method and system is to construct an operating status data set of the multi-source power system, extract key operating parameters of the operating status data set of the multi-source power system and prioritize them, generate an initial scheduling plan in combination with a multi-energy collaborative framework, and finally generate an optimized scheduling plan by performing pattern recognition processing on abnormal fluctuation characteristics in the operating feedback data. Figure 1 To the attached Figure 3 Specific embodiments of the present invention are described in detail.

[0018] First, based on Figure 1 As shown in the flowchart, the specific implementation process of the present invention begins with the data acquisition module, which is responsible for constructing a multi-source power system operating status dataset and extracting key operating parameters. In actual operation, a multi-source power system includes multiple energy sources such as wind, solar, hydropower, and thermal power. The operating status data of each energy source is collected in real time via a sensor network. This data includes, but is not limited to, generated power, load demand, energy storage status, and environmental conditions (such as wind speed and light intensity). To ensure the comprehensiveness and accuracy of the data, sensors must be deployed across all energy nodes, and time synchronization technology must be used to ensure temporal consistency of the data. Subsequently, a data preprocessing algorithm is used to clean and normalize the collected raw data to eliminate noise and dimensional differences. Based on this, key operating parameters reflecting the system's operating status are extracted, such as the real-time output power of each energy source, the charge and discharge status of the energy storage device, and the distribution of grid load. These parameters form the basis for subsequent optimized scheduling.

[0019] Next, enter the priority sorting module, which prioritizes the above-mentioned key operating parameters according to the preset weights to obtain a priority parameter sequence. Specifically, the core of priority sorting lies in calculating the contribution evaluation index of each energy type. The contribution evaluation index is obtained by weighted summation of multiple sub-indicators, mainly including three dimensions: energy utilization rate, environmental protection benefits and economic costs. For example, energy utilization rate can be measured by the ratio of the power generation of a certain energy type to the rated capacity per unit time; environmental protection benefits are evaluated by the degree of reduction in carbon emissions or pollutant emissions; economic costs take into account the direct costs (such as fuel costs) and indirect costs (such as maintenance costs) in the energy production process. During the calculation process, the following formula is used to perform weighted calculations on each key operating parameter: in, represents the weighted value of the i-th key operating parameter, is the weight of the jth sub-indicator, is the score of the i-th key operating parameter on the j-th sub-index, and n is the total number of sub-indexes. All key operating parameters are arranged from high to low by their magnitude, generating a priority parameter sequence. This process ensures that the scheduling strategy prioritizes the parameters that have the greatest impact on the overall system performance, improving the relevance and efficiency of scheduling.

[0020] Subsequently, the scheduling generation module uses the multi-source energy coordination framework to optimize the priority parameter sequence and generate an initial scheduling plan. Figure 1 As shown, the workflow of this module is divided into three steps: preliminary allocation, energy balance verification and scheme generation. First, the priority parameter sequence is input into the multi-source energy coordination framework, and preliminary allocation is performed through the resource allocation module within the framework. The resource allocation module uses the parrot algorithm model to optimize the allocation based on the priority parameter sequence. The parrot algorithm is an intelligent optimization algorithm inspired by the foraging behavior of parrots in nature. Its core idea is to achieve the search for the global optimal solution by simulating the cooperation and competition between individuals. In the present invention, the specific implementation of the parrot algorithm model includes the following steps: initialize the population, set the population size to N, and each individual represents a possible resource allocation scheme; define the fitness function to evaluate the pros and cons of each allocation scheme. The form of the fitness function is: Among them, F(x) represents the fitness value, E(x) is the energy utilization rate, C(x) is the economic cost, and G(x) is the environmental benefit. 、 and are the weight coefficients for each objective. By iteratively updating the positions of individuals in the population, the optimal solution is gradually approached. After the initial allocation is completed, the allocation results are verified for energy balance. Verification criteria include whether the total power generation meets the load demand, whether the charge and discharge status of the energy storage equipment is reasonable, and whether the output power of each energy type is within its rated range. If the initial allocation result fails the verification, the allocation strategy is adjusted and the above steps are repeated. If the verification is passed, an initial scheduling plan is generated.

[0021] After generating the initial dispatching plan, the feature extraction module starts the real-time monitoring module of the multi-source power system, collects the operating data of the initial dispatching plan, and extracts the abnormal fluctuation characteristics. Figure 2As shown in Figure 1, the module's workflow includes three steps: data acquisition, time series analysis, and feature extraction. First, the real-time monitoring module collects operational data through a sensor network deployed at each energy node. The sampling frequency is typically set to once per second to ensure sufficient temporal resolution. Subsequently, the collected data is subjected to time series analysis to identify abnormal fluctuation points. This identification utilizes a sliding window approach, segmenting the data into fixed-length time windows and calculating the standard deviation of each segment. If the standard deviation of a data segment exceeds a preset threshold, the segment is considered to have abnormal fluctuations. Finally, the corresponding fluctuation features are extracted based on the abnormal fluctuation points to form a set of abnormal fluctuation features. Fluctuation features include amplitude, duration, and frequency of occurrence, providing the foundational data for subsequent pattern recognition.

[0022] The pattern recognition module performs cluster analysis on the abnormal fluctuation feature set, divides it into several fluctuation pattern categories, and counts the frequency of occurrence of each category to generate a fluctuation pattern distribution map. Cluster analysis is implemented using the K-means algorithm, and its basic principle is to divide the data set by minimizing the sum of squares of the distances from the sample points to the cluster centers. In the present invention, the input of the clustering is the abnormal fluctuation feature set, and each sample point consists of three features: fluctuation amplitude, duration, and frequency of occurrence. The number of clusters K is determined by the elbow rule, that is, the error square sum curve under different K values ​​is plotted, and the K value at the inflection point of the curve is selected as the optimal number of clusters. After clustering is completed, the frequency of occurrence of each category in the abnormal fluctuation feature set is counted, and a fluctuation pattern distribution map is drawn. The fluctuation pattern distribution map intuitively shows the probability of occurrence of different fluctuation patterns, providing a basis for subsequent dynamic adjustments.

[0023] After generating the fluctuation pattern distribution, the dynamic adjustment module dynamically adjusts the initial scheduling plan according to the fluctuation pattern distribution to generate an optimized scheduling plan. The adjustment process is divided into two stages: training the optimized scheduling model and applying the adjustment strategy. In the first stage, a large number of scheduling cases with optimization annotations are collected, the key operating parameters and corresponding optimization results in the cases are extracted, and these data are divided into training samples and test samples. Subsequently, the deep learning model is used to perform pattern fitting on the training samples to generate a training optimization plan. The deep learning model adopts a convolutional neural network (CNN) structure, whose input is the key operating parameters and the output is the optimization results. During the model training process, the cross entropy loss function is used to evaluate the difference between the predicted results and the actual results, and the model parameters are updated through the back propagation algorithm. After the training is completed, the consistency of the training optimization plan and the standard optimization plan preset for the training sample is evaluated. The mean square error (MSE) indicator is used for consistency evaluation, and the calculation formula is: Where m is the number of samples, is the output of the standard optimization solution, This is the output of the training optimization plan. If the consistency is lower than the preset threshold, the model structure is optimized and retrained; if it reaches or exceeds the threshold, the model performance is verified using test samples. After verification, the model training is confirmed to be complete and the optimized scheduling model is generated. In the second stage, the optimized scheduling model is used to analyze the fluctuation pattern distribution and generate an adjustment strategy. The adjustment strategy includes adjusting the output power of each energy type, changing the charge and discharge status of energy storage equipment, and optimizing load distribution. After applying the adjustment strategy to the initial scheduling plan, an intermediate scheduling plan is generated and its safety is verified. The safety verification criteria include grid stability, equipment operation safety, and load satisfaction rate. If the intermediate scheduling plan fails the verification, the adjustment strategy is corrected and the steps of applying the adjustment strategy to the initial scheduling plan and generating the intermediate scheduling plan are re-executed; if the verification is passed, the optimized scheduling plan is generated.

[0024] In addition, if Figure 3 As shown, the present invention also provides a multi-source power system scheduling optimization system based on the Parrot algorithm model, which includes a data acquisition module, a priority sorting module, a scheduling generation module, a feature extraction module, a pattern recognition module and a dynamic adjustment module. The modules interact with each other through a data interface to form a complete closed-loop optimization system. In actual application scenarios, the system can be deployed in a regional power grid dispatching center, and by connecting with a distributed energy management system and a load management system, efficient and intelligent scheduling of multi-source power systems can be achieved. For example, in a certain regional power grid, the system discovers through real-time monitoring that the output power of photovoltaic power generation fluctuates abnormally due to cloud cover, and then triggers the dynamic adjustment module to generate an optimized scheduling plan, increase the output power of wind power and hydropower to make up for the shortfall of photovoltaic power, and adjust the charge and discharge status of the energy storage device to smooth out the fluctuations. This process significantly improves the economy and environmental protection of the system, verifying the technical advantages of the present invention.

[0025] In summary, the present invention constructs an operating status dataset for a multi-source power system, extracts key operating parameters for constructing the operating status dataset for the multi-source power system and prioritizes them, generates an initial scheduling plan in combination with a multi-energy collaborative framework, and finally generates an optimized scheduling plan by performing pattern recognition processing on abnormal fluctuation characteristics in the operating feedback data. This method not only improves the adaptability of the multi-energy power system under dynamic working conditions, but also significantly enhances the economy and environmental protection of the system, providing strong support for the efficient and intelligent scheduling of the multi-energy power system.

Claims

1. A multi-source power system dispatch optimization method based on the Parrot algorithm model, characterized in that: The method comprises: Construct an operating status dataset of a multi-source power system and extract key operating parameters; Prioritizing the key operating parameters according to preset weights to obtain a priority parameter sequence; Utilizing a multi-source energy collaborative framework to optimize the allocation of the priority parameter sequence and generate an initial scheduling plan; Acquiring operational feedback data of the initial scheduling plan, and extracting abnormal fluctuation characteristics from the operational feedback data; Performing pattern recognition processing on the abnormal fluctuation characteristics to obtain a fluctuation pattern distribution; The initial scheduling plan is dynamically adjusted according to the fluctuation pattern distribution to generate an optimized scheduling plan.

2. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 1, characterized in that: Prioritizing the key operating parameters according to preset weights to obtain a priority parameter sequence includes: Obtain contribution evaluation indicators of each energy type in a multi-energy power system; Performing weighted calculation on the key operating parameters according to the contribution evaluation index to obtain weighted parameter values; The key operating parameters are arranged in descending order according to the weighted parameter values ​​to generate a priority parameter sequence.

3. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 1, characterized in that: The method of utilizing a multi-source energy collaborative framework to optimize and allocate the priority parameter sequence to generate an initial scheduling plan includes: Input the priority parameter sequence into the multi-source energy coordination framework and perform preliminary allocation through the resource allocation module within the framework; Conduct energy balance verification on the preliminary allocation results; If the preliminary allocation result fails to pass the energy balance check, the allocation strategy is adjusted and the preliminary allocation steps are performed again; If the preliminary allocation result passes the energy consumption balance check, an initial scheduling plan is generated.

4. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 1, characterized in that: The obtaining of the operation feedback data of the initial scheduling plan and extracting abnormal fluctuation characteristics in the operation feedback data include: Starting a real-time monitoring module of the multi-source power system to collect operating data of the initial scheduling plan; Performing time series analysis on the operating data to identify abnormal fluctuation points therein; Corresponding fluctuation features are extracted according to the abnormal fluctuation points to obtain an abnormal fluctuation feature set.

5. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 1, characterized in that: The performing pattern recognition processing on the abnormal fluctuation characteristics to obtain the fluctuation pattern distribution includes: Performing cluster analysis on the abnormal fluctuation feature set to divide the abnormal fluctuation pattern into several categories; The frequency of occurrence of each category in the abnormal fluctuation feature set is counted to generate a fluctuation pattern distribution graph.

6. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 1, characterized in that: Before dynamically adjusting the initial scheduling solution according to the fluctuation pattern distribution to generate an optimized scheduling solution, the method further includes: Collect a large number of scheduling cases with optimization annotations, and extract the key operating parameters and corresponding optimization results in the scheduling cases; Dividing the key operating parameters and optimization results in the scheduling case into training samples and test samples; Using a deep learning model to perform pattern fitting on the training samples to generate a training optimization plan; Evaluating the consistency of the training optimization scheme with a preset standard optimization scheme for the training sample; If the consistency is lower than a preset threshold, the pattern fitting step is re-executed after optimizing the structure of the deep learning model; If the consistency reaches or exceeds a preset threshold, the performance of the deep learning model is verified using the test sample; If the deep learning model fails the performance verification, then the step of extracting the key operating parameters and corresponding optimization results from the scheduling case is re-executed after increasing the data volume of the scheduling case; If the deep learning model passes the performance verification, the model training is confirmed to be complete and an optimized scheduling model is generated.

7. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 6, characterized in that: The evaluating the consistency of the training optimization scheme with the standard optimization scheme preset for the training sample includes: Comparing the degree of difference between the training optimization scheme and the standard optimization scheme in key operating parameters; The consistency is quantified according to the degree of difference to generate a consistency score.

8. The multi-source power system dispatch optimization method based on the Parrot algorithm model according to claim 6 is characterized in that: The dynamically adjusting the initial scheduling plan according to the fluctuation pattern distribution to generate an optimized scheduling plan includes: Analyzing the fluctuation pattern distribution using the optimization scheduling model to generate an adjustment strategy; Applying the adjustment strategy to the initial scheduling plan to generate an intermediate scheduling plan; Performing a security check on the intermediate scheduling plan; If the intermediate scheduling solution fails the security check, revising the adjustment strategy and then re-performing the step of applying the adjustment strategy to the initial scheduling solution to generate the intermediate scheduling solution; If the intermediate scheduling solution passes the security check, an optimized scheduling solution is generated.

9. A multi-source power system dispatch optimization system based on the Parrot algorithm model, characterized in that: The system comprises: Data acquisition module, used to construct the operating status data set of multi-source power systems and extract key operating parameters; A priority sorting module is used to prioritize the key operating parameters according to preset weights to obtain a priority parameter sequence; a scheduling generation module, configured to optimize the allocation of the priority parameter sequence using a multi-source energy collaborative framework and generate an initial scheduling plan; a feature extraction module, configured to obtain operational feedback data of the initial scheduling plan and extract abnormal fluctuation features from the operational feedback data; A pattern recognition module is used to perform pattern recognition processing on the abnormal fluctuation characteristics to obtain a fluctuation pattern distribution; A dynamic adjustment module is used to dynamically adjust the initial scheduling plan according to the fluctuation pattern distribution to generate an optimized scheduling plan.

10. The multi-source power system dispatch optimization system based on the Parrot algorithm model according to claim 9, characterized in that: The system further comprises a real-time monitoring module for collecting operation data of the multi-source power system and providing the data to the feature extraction module for extracting abnormal fluctuation features.

Images