CN118052152B

CN118052152B - Power performance data based simulation method

Info

Publication number: CN118052152B
Application number: CN202410445075.4A
Authority: CN
Inventors: 李映雪; 戴奇奇; 宫嘉炜; 张雪婷; 王敏; 吴浩; 王伟; 陈日欢
Original assignee: Jiangxi Tengda Electric Power Design Institute Co ltd; State Grid Corp of China SGCC; Economic and Technological Research Institute of State Grid Jiangxi Electric Power Co Ltd
Current assignee: Jiangxi Tengda Electric Power Design Institute Co ltd; State Grid Corp of China SGCC; Economic and Technological Research Institute of State Grid Jiangxi Electric Power Co Ltd
Priority date: 2024-04-15
Filing date: 2024-04-15
Publication date: 2024-08-13
Anticipated expiration: 2044-04-15
Also published as: CN118052152A

Abstract

The invention relates to the technical field of power simulation, in particular to a power performance data-based simulation method. The method comprises the following steps: manifold learning evolution processing is carried out on the acquired power generation equipment group data to obtain electric power performance manifold evolution data; carrying out cooperative vector estimation and causal relation inspection according to the power performance manifold evolution data to obtain power performance variable relation data; constructing a power generation performance simulation model; performing optimization strategy integration on the power generation performance simulation model through the electric power performance variable relation data based on the artificial bee colony algorithm to generate an optimized power generation performance simulation model; performing dynamic event response simulation by using an optimized power generation performance simulation model, and performing power network reconstruction processing to generate power network reconstruction data; and carrying out supply side elasticity assessment on the power network reconstruction data to generate supply side elasticity assessment data. According to the invention, the power generation performance is optimized and the resources are utilized efficiently through the electric power performance simulation.

Description

Power performance data based simulation method

Technical Field

The invention relates to the technical field of power simulation, in particular to a power performance data-based simulation method.

Background

With the continuous expansion of the scale and the improvement of the intelligent degree of the power system, the complexity of the power system is also continuously increasing. Such complexity is manifested in the number of components of the system, the interaction of the various devices, and the existence of various operating states. Meanwhile, large data and artificial intelligence technology are developed, and more power system data are collected, stored and analyzed, including power load data, generator operation data, power grid state data and the like. However, the continuous appearance and development of novel power systems such as distributed energy sources and micro-grids, etc., and the traditional power system simulation methods have great limitations when facing to the complex power network structures, and the novel power systems have the characteristics of distributed energy sources, a plurality of energy nodes, flexible energy flow paths, etc., so that the traditional static simulation methods are difficult to accurately capture the actual behaviors of the power systems, and cannot accurately simulate and predict dynamic events in the power systems, such as faults, load changes, etc.

Disclosure of Invention

Based on the above, the present invention provides a simulation method based on electric performance data to solve at least one of the above technical problems.

In order to achieve the above purpose, the power performance data based simulation method comprises the following steps:

Step S1: acquiring power generation equipment group data; collecting historical power generation performance data of the power generation equipment group data to generate historical power generation performance data; carrying out multi-mode data fusion according to the historical power generation performance data to generate multi-mode power performance data; manifold learning evolution processing is carried out on the multi-mode power performance data to obtain power performance manifold evolution data;

Step S2: detecting abnormal power generation modes according to the power performance manifold evolution data to generate abnormal power generation mode data; carrying out cooperative vector estimation on the abnormal power generation mode data to generate cooperative vector estimation data; carrying out causal relation test according to the estimated data of the coordination vector to obtain variable relation data of the electric power performance;

step S3: performing multi-level power generation network simulation according to the power performance manifold evolution data to generate multi-level power network simulation data; carrying out electric power performance model construction through multi-level electric power network simulation data based on a preset graph neural network model to obtain a power generation performance simulation model;

Step S4: carrying out power production index processing according to the power performance variable relation data, and making power generation constraint conditions to obtain power generation constraint condition data; performing multi-objective power generation optimization processing on the power generation constraint condition data by using an artificial bee colony algorithm, and performing optimization strategy integration on a power generation performance simulation model to generate an optimized power generation performance simulation model;

Step S5: real-time power data acquisition is carried out according to the power generation equipment group data, and real-time power data are generated; analyzing dynamic performance influence factors of the real-time power data to generate dynamic performance influence data; performing dynamic event response simulation on the dynamic performance influence data by using the optimized power generation performance simulation model to generate response event simulation data; carrying out power network reconstruction processing according to the response event simulation data to generate power network reconstruction data;

step S6: and carrying out power supply side energy analysis on the power network reconstruction data, carrying out supply side elasticity evaluation, and generating supply side elasticity evaluation data.

The invention acquires the data of the power generation equipment group to collect various operation parameters and performance indexes related to the power generation equipment so as to comprehensively know the operation state of the equipment. The historical power generation performance data acquisition is performed on the data, so that the performance file of the equipment in a historical period can be built. Through multi-mode data fusion, data from different sources are comprehensively processed, so that the comprehensiveness and accuracy of the data are improved. Through manifold learning evolution processing, high-dimensional multi-mode electric performance data are mapped into a low-dimensional manifold space, so that the internal structure and rules of the data can be found. The abnormal power generation mode detection is carried out through the power performance manifold evolution data, so that the abnormal condition in the power generation system can be timely identified, and the potential problem is prevented or timely treated. And carrying out cooperative vector estimation on abnormal power generation mode data, and helping to understand long-term relations and common trends among different performance variables. And the causal relation between the performance variables is determined through causal relation detection, so that the influence mechanism of each factor on the system operation is cleared. And carrying out multi-level power generation network simulation according to the power performance manifold evolution data, so that the power generation network structure and the running condition of different levels can be simulated, and more comprehensive data support is provided for the simulation and analysis of the system. By constructing the electric power performance model based on a preset graph neural network model, the model can be built based on real data, and the running state and performance of the system can be predicted through the model. And the power production index is processed according to the power performance variable relation data, so that the production capacity and efficiency of the power system can be accurately estimated. Secondly, the generation constraint condition is formulated, and the limitation and the requirement of the generation system under various conditions can be clearly defined. The artificial bee colony algorithm is utilized to perform multi-objective power generation optimization processing on power generation constraint condition data, so that an optimal power generation scheme can be found under the condition of considering a plurality of objectives and constraints, and the overall efficiency and performance of the system are improved. And finally, carrying out optimization strategy integration on the power generation performance simulation model, and directly applying an optimization result to the model to improve the accuracy and applicability of the model. The real-time power data acquisition is carried out according to the power generation equipment group data, the actual data of the system operation can be timely obtained, the dynamic performance influence factor analysis is carried out on the real-time power data, the key influence factors in the system operation process can be found, and references are provided for timely adjustment and coping. And the dynamic event response simulation is carried out on the dynamic performance influence data by utilizing the optimized power generation performance simulation model, so that the response and performance of the system under different conditions can be simulated. And the power network reconstruction processing is carried out according to the response event simulation data, so that the structure and configuration of the power system can be optimized, and the reliability and stability of the system are improved. The power supply side energy analysis is carried out on the power network reconstruction data, so that the power supply capacity and the resource allocation situation of the power system can be comprehensively evaluated, a basis is provided for management and planning of the power supply side, the elasticity evaluation of the power supply side is carried out, and the flexibility and the adaptability of the power supply side in coping with various changes and challenges can be evaluated. Therefore, the simulation method based on the power performance data performs multi-mode data fusion and manifold learning evolution processing through the historical power generation performance data, and accurately digs out the built-in structure and rule of the data in a large amount of historical power performance data. And considering an abnormal power generation mode in the power generation system, carrying out cooperative vector estimation on abnormal power generation mode data, and helping to understand long-term relations and common trends among different performance variables, so that the actual behavior of the power system is captured. By constructing the power generation performance optimization simulation model, the response and performance of the power system under different conditions can be accurately simulated, and the optimization of each node of the power system and the reconstruction of the power network are realized. And carrying out supply side elasticity evaluation on the reconstructed power network to comprehensively evaluate the adjusted power supply capacity and resource allocation condition of the power system.

Preferably, step S1 comprises the steps of:

step S11: acquiring power generation equipment group data; collecting historical power generation performance data of the power generation equipment group data to generate historical power generation performance data;

Step S12: performing Z-score standardization on the historical power generation performance data to generate standard historical power performance data;

Step S13: carrying out multi-mode data fusion according to the standard historical power performance data to generate multi-mode power performance data;

step S14: performing time window division on the multi-mode power performance data through preset time window data to generate multi-mode window division data;

Step S15: performing intra-window similarity calculation according to the multi-mode window division data to obtain window similarity data;

step S16: carrying out dynamic adjacency graph construction on the multi-mode power performance data by utilizing the window similarity data to generate dynamic power adjacency graph data;

Step S17: carrying out Laplace feature mapping according to the dynamic power adjacency graph data to generate dynamic power feature mapping data;

step S18: and performing manifold evolution processing on the dynamic power characteristic mapping data so as to obtain power performance manifold evolution data.

According to the invention, by acquiring the data of the power generation equipment group, the running condition, various indexes and related operation records of the equipment group in a period of time can be obtained. Through Z-score standardization, historical power generation performance data can be converted into standardized data with the mean value of 0 and the standard deviation of 1, so that the influence caused by different scales is eliminated, and the comparison between different equipment and different indexes is more objective and accurate. The data of different modes are fused, so that factors in multiple aspects can be comprehensively considered, the comprehensive evaluation capability of the electric power performance is improved, and the operation condition of equipment can be more comprehensively known. The data can be divided according to time windows to extract the power performance characteristics in different time periods, which is helpful for capturing the dynamic change rule of the operation of the equipment. By calculating the data similarity in the window, the change rule of the equipment performance under similar working conditions can be found. By constructing the dynamic adjacency graph, the data in different time windows can be connected to reflect the evolution relation of the equipment performance. The high-dimensional dynamic power data can be mapped into a low-dimensional space through Laplace feature mapping, so that the local structure information of the data is reserved, and the subsequent manifold evolution processing and feature extraction are facilitated. Through manifold evolution processing, manifold structures and evolution rules of data in a low-dimensional space can be found, and the change trend of equipment performance along with time is revealed.

Preferably, step S2 comprises the steps of:

step S21: performing Fourier transform processing according to the electric power performance manifold evolution data to generate electric power performance spectrum data;

step S22: detecting an abnormal power generation mode according to the electric power performance frequency spectrum data to generate abnormal power generation mode data;

step S23: processing the abnormal power generation mode data in a multi-element time sequence manner to generate multi-element time sequence power abnormal data;

step S24: carrying out synergistic relation analysis according to the multi-element time sequence power abnormal data to obtain power performance variable synergistic data;

step S25: carrying out cooperative vector estimation on the electric performance variable cooperative data to generate cooperative vector estimation data;

step S26: and carrying out causal relation inspection on the multi-element time sequence power abnormal data by utilizing the cooperative vector estimation data to obtain power performance variable relation data.

The Fourier transform can convert the power performance data from the time domain to the frequency domain, so that the contribution of different frequency components in the power performance can be found, and the periodic variation and the frequency characteristic can be identified. The abnormal power generation mode detection can help identify abnormal behaviors or unusual modes in the power performance, so that potential faults or abnormal conditions are revealed, equipment faults can be prevented, and reliability and stability of a power system are improved. Through multi-element time sequence processing, the relation and evolution trend among different variables in the abnormal power generation mode data can be comprehensively analyzed, and more comprehensive and finer power abnormal data can be generated. The analysis of the synergistic relationship can reveal the long-term stable relationship among different variables in the electric performance, help understand the common evolution trend among the variables, and the estimation of the synergistic vector can provide the coefficient and the direction of the synergistic relationship among the variables, so that the understanding of the long-term equilibrium relationship among the electric performance variables is further enhanced. The coefficient and the direction of the coordination relation among the variables can be obtained through the coordination vector estimation, and the long-term equilibrium relation among the variables is further revealed.

Preferably, step S22 comprises the steps of:

Step S221: performing multi-scale decomposition on the electric performance spectrum data to obtain multi-scale decomposition data;

Step S222: performing multi-scale entropy calculation according to the multi-scale decomposition data to generate multi-scale entropy data; performing multi-scale variance calculation according to the multi-scale decomposition data to generate multi-scale variance data;

step S223: performing multi-scale feature fusion on the multi-scale decomposition data through the multi-scale variance data and the multi-scale entropy data to generate multi-scale feature data;

Step S224: detecting abnormal frequency components of the multi-scale characteristic data by using a preset abnormal frequency detection threshold value to generate abnormal frequency component data;

step S225: extracting trend components from the abnormal frequency component data to generate performance trend component data;

step S226: residual term calculation is carried out on the multi-scale characteristic data through the performance trend component data, and electric performance residual term data are obtained;

Step S227: and carrying out periodic pattern analysis according to the power performance residual error item data, and carrying out abnormal power generation pattern matching so as to obtain abnormal power generation pattern data.

According to the multi-scale decomposition method, the power performance spectrum data can be decomposed according to different scales, so that the change conditions in different frequency ranges are revealed, and the complexity and the variability of the power performance data can be captured under different scales by calculating the multi-scale entropy and the multi-scale variance. The multi-scale feature fusion can combine the feature information under a plurality of different scales, and improves the representation capability and analysis precision of the data. Abnormal frequency component detection can discover the existence of abnormal frequency components at different scales, helping to identify abnormal changes in the power performance data. The trend component extraction can extract trend information in the abnormal frequency components, and helps to understand long-term change trend in the power performance data. After the trend information can be removed by residual term calculation, residual term data with stronger variability is obtained, and the method is beneficial to highlighting abnormal changes and periodic modes in the data. The periodic rule and the abnormal pattern in the power performance data can be found through the periodic pattern analysis and the abnormal pattern matching, the potential abnormal power generation condition can be identified, an important basis is provided for equipment fault diagnosis and prediction, and the stability and the reliability of a power system are improved.

Preferably, step S3 comprises the steps of:

step S31: performing multi-level power generation network simulation according to the power performance manifold evolution data to generate multi-level power network simulation data;

step S32: generating a generating set topological graph according to the multi-level power network simulation data to generate generating set topological graph data;

Step S33: establishing a mapping relation between a power generation equipment group and power generation performance through power generation group topological graph data based on a preset graph neural network model, so as to establish an initial power generation performance simulation model;

Step S34: carrying out random data division on the multi-mode power performance data to respectively obtain training set data and verification set data;

Step S35: and carrying out model training on the initial power generation performance simulation model by using training set data, and carrying out model verification by using verification set data, thereby obtaining the power generation performance simulation model.

The multi-level power generation network simulation can simulate the operation conditions of all components in the power system in different levels, and comprehensively reflects the complexity and the dynamic property of the power system from whole to local and from macroscopic to microscopic. The topological graph of the generating set can clearly show the connection relation and the topological structure among all components in the generating system, and visual network graphic representation is provided for establishing a generating performance simulation model. Through modeling of the graph neural network model, the relation between the power generation equipment groups can be fully mined, and the complexity and nonlinear characteristics of the power system can be reflected more accurately. Through random data partitioning, the data set can be divided into a training set and a verification set for training and verification of the model. By the method, generalization capability and reliability of the model can be guaranteed, the situation of over-fitting or under-fitting of the model is avoided, and prediction accuracy of the model is improved. Through model training and verification, model parameters can be continuously optimized, the fitting capacity and generalization capacity of the model are improved, and the actual running condition of the power system is reflected by the model more accurately.

Preferably, step S4 comprises the steps of:

step S41: carrying out power production extraction according to the power performance variable relation data to respectively obtain generating capacity data, generating type data and generating quality data;

Step S42: calculating the load rate according to the generated energy data, and generating set load capacity data;

Step S43: analyzing a power generation mode according to the power generation type data, and performing power generation cost processing to generate power generation cost data;

Step S44: performing quality index calculation according to the power generation quality data to generate power generation quality index data;

Step S45: constraint condition setting is carried out on the power generation quality index data, the power generation cost data and the power generation group load capacity data by utilizing a preset power generation rule of power generation equipment, so that power generation constraint condition data are obtained;

step S45: performing multi-objective power generation optimization processing on the power generation constraint condition data by using an artificial bee colony algorithm to generate a multi-objective power generation optimization strategy;

step S46: and carrying out optimization strategy integration on the power generation performance simulation model through a multi-target power generation optimization strategy to generate an optimized power generation performance simulation model.

The invention extracts key power production data including information such as generated energy, generation type, generation quality and the like by analyzing the relation among the power performance variables. These data are key indicators for assessing the efficiency and quality of power production. The load data of the generating set is an important index for measuring the utilization rate and the running state of the generating equipment, and the running condition of the generating equipment can be accurately reflected through load rate calculation. By analyzing the power generation types, the cost structure and the cost difference of different power generation modes can be determined, so that the power generation mode selection is optimized, the power generation cost is reduced, and the economic benefit is improved. The power generation quality index is an important basis for evaluating the operation quality and the power quality of power generation equipment, and the quality problem in the power generation process can be timely found and solved through calculation and analysis of the quality index. By setting constraint conditions, various indexes in the power generation process can be limited in a reasonable range. Through the optimization processing of the artificial bee colony algorithm, the optimal balance point among a plurality of targets can be found, the overall efficiency and performance of the power generation system are improved, and the optimization of multiple targets is realized. The optimized power generation performance simulation model can more accurately reflect the running state and performance of the power system.

Preferably, step S46 comprises the steps of:

Step S461: generating initial group data through the power generation constraint condition data to generate optimization target initial group data;

step S462: carrying out sharing communication processing on the initial group data of the optimization target through preset sharing triggering condition data to generate group sharing communication data;

step S463: performing multi-group cooperative processing according to the group sharing communication data to generate a multi-group cooperative strategy;

Step S464: performing multi-objective initial solution calculation on the multi-population collaborative strategy based on an artificial bee colony algorithm to obtain objective initial solution data;

step S465: and carrying out maximum adaptability mapping processing on the target initial solution data, thereby obtaining a multi-target power generation optimization strategy.

According to the power generation constraint condition data, initial optimization target group data with diversity is generated under the multidimensional constraint condition, and a starting point and a search space are provided for a subsequent optimization algorithm. The shared communication processing can utilize the existing shared triggering condition to promote information exchange and collaboration among group members, accelerate the convergence speed and the optimization effect of the optimization algorithm, and improve the efficiency and the robustness of the algorithm. The multi-group cooperative processing can fully utilize information sharing and cooperation among group members, effectively adjust optimization strategies in groups, improve global searching capability and convergence performance of an optimization algorithm, and enable an optimization result to be more robust and reliable. The artificial bee colony algorithm can obtain a group of initial solution data through searching and adjusting on the basis of multi-colony cooperation according to the optimization requirements of a plurality of targets, and provides a starting point and reference data for multi-target power generation optimization. The maximum fitness mapping processing can screen out the optimization strategy with the most advantages and potential according to the fitness value of the target initial solution data, and provides important guidance and direction for further iteration and optimization of the optimization algorithm.

Preferably, step S5 comprises the steps of:

step S51: real-time power data acquisition is carried out according to the power generation equipment group data, and real-time power data are generated;

step S52: analyzing dynamic performance influence factors according to the real-time power data to generate dynamic performance influence data;

step S53: carrying out characteristic engineering processing on the dynamic performance influence data to generate dynamic performance influence characteristic data;

Step S54: transmitting the dynamic performance influence characteristic data to an optimized power generation performance simulation model to perform dynamic event response simulation, and generating response event simulation data;

Step S55: performing Lagrange sensitivity analysis according to the response event simulation data to obtain power generation performance fault sensitivity data;

step S56: carrying out key equipment identification according to the power generation performance fault sensitive data to generate key power generation equipment data;

Step S57: carrying out fragile node identification according to the power generation performance fault sensitive data to generate power generation structure fragile node data;

Step S58: and carrying out power network reconstruction processing based on the key power generation equipment data and the power generation structure fragile node data to generate power network reconstruction data.

The invention monitors the running state of the power generation equipment group in real time and acquires data, and can acquire the running condition and performance index of the power system in time. The dynamic performance influence factor analysis can identify and analyze dynamic factors influencing the performance of the power system, such as load fluctuation, temperature change and the like, and the characteristic engineering processing can extract effective characteristic information, such as frequency, amplitude, trend and the like, from the original data, so that the change rule of the dynamic performance influence factor can be accurately described. By simulating the dynamic event response process, the performance and response capability of the power system under different conditions can be evaluated. The Lagrange sensitivity analysis can evaluate the influence degree of system parameters on performance indexes, and identify the parameters which are most sensitive to the influence of the system performance. The critical device identification enables determination of the critical device that has the greatest impact on the operational stability and performance of the power system. The fragile node identification can identify the node with larger influence on the structural stability of the power system, thereby being beneficial to optimizing the system structure and improving the operation strategy and improving the robustness and the reliability of the system. The power network reconstruction can optimize the structural layout and equipment configuration of the power system, and improve the anti-interference capability and the operation efficiency of the system.

Preferably, step S6 comprises the steps of:

Step S61: carrying out power supply side energy analysis according to the power network reconstruction data to respectively obtain energy resource type data and energy facility data;

Step S62: performing energy resource characteristic analysis on the energy resource type data to generate energy resource characteristic data;

step S63: performing production elasticity index calculation according to the energy facility data to generate energy elasticity index data;

Step S64: acquiring electric power market demand data; and carrying out supply side elasticity assessment on the power market demand data by using a power supply side assessment algorithm based on the energy elasticity index data and the energy characteristic data, and generating supply side elasticity assessment data.

According to the invention, the power supply side energy analysis is carried out according to the power network reconstruction data, and the power supply side energy of the power system can be comprehensively analyzed, including the type and distribution condition of the energy and related energy facility information. The energy resource characteristic analysis can be used for deeply knowing the characteristics of various energy resources, such as yield, efficiency, reproducibility and the like, and is helpful for identifying and utilizing the advantages and limitations of different energy resources. The calculation of the production elasticity index can evaluate the adaptability and the flexibility of the energy facility to different energy demands, and is beneficial to optimizing energy configuration and improving power supply efficiency. The power supply side elasticity evaluation method has the advantages that the power supply side elasticity evaluation algorithm is utilized to evaluate the power market demand data based on the energy elasticity index data and the energy characteristic data, the demand characteristics of the power market and the elasticity condition of energy supply can be fully considered, the flexibility and the adaptability of the power supply side in the aspect of meeting the market demand are evaluated, and important support is provided for the supply and demand balance and the running stability of the power market.

Drawings

Fig. 1 is a schematic flow chart of steps of a simulation method based on electric power performance data according to the present invention.

FIG. 2 is a flowchart illustrating the detailed implementation of step S2 in FIG. 1;

fig. 3 is a detailed implementation step flow diagram of step S4 in fig. 1.

The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.

Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.

It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.

In order to achieve the above objective, referring to fig. 1 to 3, the present invention provides a simulation method based on electric power performance data, comprising the following steps:

In the embodiment of the present invention, as described with reference to fig. 1, a step flow diagram of a simulation method based on electric power performance data according to the present invention is provided, and in the embodiment, the simulation method based on electric power performance data includes the following steps:

In the embodiment of the invention, the data of the power generation equipment group is obtained, wherein the data comprise parameters such as current, voltage, frequency and the like of each generator and the load condition of the whole system. Historical power generation performance data are collected, then Z-score standardization processing is carried out, and the data are standardized into the form of zero mean and unit standard deviation. And then, fusing the performance data of different types to form the multi-mode electric power performance data. Dividing the multi-mode data by using a preset time window, and calculating the similarity of the data in the window. And determining an adjacency window according to the similarity, and constructing a dynamic power adjacency graph. And decomposing the characteristic value by using the Laplace matrix to obtain a characteristic vector and a characteristic value, thereby obtaining dynamic electric power characteristic mapping data. And finally, processing the feature mapping data through a manifold learning algorithm to obtain electric performance manifold evolution data.

In the embodiment of the invention, the Fourier transform is utilized to process the electric performance manifold evolution data, and the time domain signal is converted into the frequency domain representation. Frequency and amplitude information is calculated by a fast fourier transform algorithm to generate electrical performance spectrum data. An abnormal peak is analyzed in the spectral data, possibly representing a system frequency imbalance or failure. An appropriate anomaly detection algorithm is selected for analysis, such as a statistical or machine learning based approach. And carrying out anomaly detection on the frequency spectrum data, and identifying an anomaly mode different from a normal mode. Modeling the interaction relation of abnormal power generation mode data by using a multivariate time series processing method, such as a vector autoregressive model. And predicting and generating abnormal data according to the model parameter estimation and residual analysis. And carrying out synergistic relation analysis on variables such as abnormal frequency components, amplitude changes and the like, and determining the causal relation among the variables by using a Grangel causal test. And obtaining electric performance variable relation data according to the test result, and describing causal relation among the variables.

In the embodiment of the invention, the simulation parameters of the multi-level power generation network are set according to the power performance manifold evolution data, including time period and precision, and an applicable simulation method is selected. And carrying out multi-level power generation network simulation on the data by using the selected simulation method, and updating the state and processing the event. And selecting a topological graph generating algorithm, such as a minimum spanning tree algorithm, according to the requirements to generate generating set topological graph data. And constructing a mapping relation between the power generation equipment group and the power generation performance by using the graph neural network model, wherein the mapping relation comprises model initialization, forward propagation and backward propagation. And (3) training the model through multiple iterations, dividing the data into a training set and a verification set by using random sampling, and updating model parameters by using the training set data. A loss function was calculated in each epoch and model fitting was assessed. And evaluating the generalization capability and the prediction accuracy of the model through the verification set, and further constructing a power generation performance simulation model.

In the embodiment of the invention, causal relation verification is carried out according to the electric performance variable relation data, remarkable causal relation is confirmed, and related data are extracted as a part of generating capacity, type and quality. And calculating the load rate of the generating set and analyzing the generating mode, and calculating the generating cost by taking the cost factors into consideration. And (5) formulating a power generation rule according to the quality index data to ensure the stability and economy of the system. And converting the rule into constraint condition data to prepare for algorithm processing. And carrying out multi-objective power generation optimization processing on the constraint conditions by using an artificial bee colony algorithm, and integrating an optimization strategy into a power generation performance simulation model. And finally, evaluating the performance of the model through simulation and adjusting and optimizing to generate an optimized power generation performance simulation model comprehensively considering the multi-objective optimization strategy.

In the embodiment of the invention, the power data of the power generation equipment group, including current, voltage, power and the like, are collected in real time according to the preset configuration. And monitoring various dynamic performance influence factors such as the influence of load change on the power characteristics by using real-time data, and analyzing by using a statistical or machine learning algorithm. And carrying out feature processing on the dynamic influence data, including selection, dimension reduction, standardization and the like, so as to facilitate subsequent modeling. And transmitting the characteristic data to an optimized power generation performance simulation model, and performing dynamic event response simulation to generate simulation data. For example, the data after feature engineering processing is input into a neural network for simulation, such as simulating the response of load sudden increases to power generation equipment. And analyzing the influence degree of each factor on the power generation performance by using a Lagrange multiplier method, and calculating a sensitivity index to obtain power generation performance fault sensitivity data. Critical devices are identified based on the sensitivity index, e.g., cooling systems where temperature increases have the greatest impact on performance are identified as critical devices. Frequent faults are found to affect overall performance, identifying it as a weak node. And the power network connection is redesigned, so that the connection stability of key equipment is ensured, and the influence of fragile nodes on a system is reduced.

In the embodiment of the invention, according to the power network reconstruction data, the energy type and the energy facility information of each area are determined, including solar power stations, wind power stations, hydroelectric power stations and the like, the capacity, the power generation mode and the geographic position of the solar power stations, the wind power stations, the hydroelectric power stations and the like are recorded, and the energy facility data are generated. And (3) aiming at each energy type, carrying out energy characteristic analysis such as sunlight time and illumination intensity of a solar energy resource, wind speed and wind direction of a wind energy resource and the like. According to the energy facility data, calculating production elasticity indexes including power supply capacity, adjustability, response speed and the like, and generating energy elasticity index data. And the elasticity of the supply side is evaluated by utilizing a power supply side evaluation algorithm in combination with market demands, and the response speed and the resource allocation flexibility of the facility are considered. And scoring the facility according to the evaluation result, so as to generate the supply side elasticity evaluation data.

Preferably, step S1 comprises the steps of:

In the embodiment of the invention, the real-time operation data is acquired through the monitoring interface of the data acquisition system connected to the power generation equipment group. For example, parameters such as current, voltage, frequency, etc. of each generator, and the load condition of the whole system can be obtained. Then, historical power generation performance data collection is performed on these data. Z-score standardization is carried out on the historical power generation performance data, and the mean value and standard deviation of each performance index are calculated first. For example, for the current data of the generator, we can calculate the mean value to be 100A and the standard deviation to be 10A. Then, a Z-score normalization process is performed on each data point, i.e., each data point is subtracted by the mean value and divided by the standard deviation to obtain a normalized value. For example, if the current value of a certain data point is 110A, the value normalized by Z-score is (110-100)/10=1. The multi-mode data fusion is carried out according to standard historical power performance data, and firstly, various data types such as current, voltage, frequency and the like which need to be fused are determined. These different types of data are then integrated and combined to form a comprehensive multi-modal dataset. For example, the current and voltage data may be combined into one data set. And carrying out time window division on the multi-mode power performance data through preset time window data, and firstly determining the size and the sliding step length of the time window. For example, each time window contains one hour of data, and the windows overlap each other for half an hour. Then, the data is divided from the history data according to the set time window size and the sliding step size. For example, starting from the first timestamp, taking one hour of data in succession as the data for the first time window, then sliding the window back half an hour, taking the data for the adjacent window, and so on until the entire historical dataset is covered. Common similarity measurement methods such as euclidean distance or cosine similarity are used. Then, for each time window, the similarity between it and the other windows is calculated. For example, there are 10 time windows, and then 10×10=100 similarity values need to be calculated. Finally, the similarity values are organized into a matrix or other form of data structure. And determining a threshold value or other conditions according to the window similarity data, and regarding the window with higher similarity as an adjacent window. For example, a similarity threshold of 0.8 may be set, and then all windows having a similarity greater than 0.8 are considered contiguous windows. Then, a dynamic power adjacency graph is constructed from the adjacencies. For example, a weighted graph may be constructed with windows as nodes and similarities as weights for edges. And calculating the Laplacian matrix according to the constructed dynamic power adjacency graph. For example, a standard laplacian matrix or a normalized laplacian matrix may be used. And then, carrying out eigenvalue decomposition on the Laplace matrix to obtain eigenvectors and eigenvalues. Finally, the first several feature vectors are selected as new feature space, and the original data are mapped into the new feature space to obtain dynamic power feature mapping data. And performing manifold evolution processing on the obtained dynamic power characteristic mapping data by using a manifold learning algorithm. For example, manifold learning algorithms such as Local Linear Embedding (LLE), isometric mapping (Isomap), etc. may be used. And then mapping the original data into manifold space to obtain the electric performance manifold evolution data.

Preferably, step S2 comprises the steps of:

As an example of the present invention, referring to fig. 2, a detailed implementation step flow diagram of step S2 in fig. 1 is shown, where step S2 includes:

In the embodiment of the invention, fourier transform processing is carried out on the prepared power performance manifold evolution data. Fourier transform is the process of converting a time domain signal into the frequency domain, which can decompose the signal into components of different frequencies. For time series data, the computation is typically performed using a Fast Fourier Transform (FFT) algorithm. For example, the current data is fourier transformed using an FFT algorithm to obtain frequency and amplitude information. Then, from the result of the fourier transform, power performance spectrum data may be generated, which contains current intensity information at different frequencies. Finally, the spectral data is analyzed to find that an abnormal peak exists at a particular frequency, which may represent a frequency imbalance or fault in the system.

In the embodiment of the invention, a proper abnormality detection algorithm is selected for analysis according to the characteristics of the frequency spectrum data. Common anomaly detection algorithms include statistical-based methods, machine learning-based methods, and the like. For example, anomaly detection may be performed using a threshold-based method or a clustering algorithm. And performing anomaly detection on the electric performance spectrum data. This may involve analysis and comparison of information such as frequency and amplitude to identify abnormal patterns that differ from normal patterns. For example, it is possible to detect whether there is a frequency component out of the normal range or an abnormal frequency distribution pattern.

In the embodiment of the invention, a proper multi-element time sequence processing method is selected according to the characteristics and the analysis purpose of the abnormal power generation mode data. Common processing methods include vector autoregressive models (VAR), vector Error Correction Models (VECM), and the like. For example, a vector autoregressive model may be selected taking into account interactions between the anomaly frequency components. And modeling the abnormal power generation mode data in a multi-element time sequence. This may involve steps of estimation of model parameters, residual analysis, etc. For example, for a vector autoregressive model, it is necessary to estimate the coefficient matrix of the model and check the residual sequence to ensure the rationality and stability of the model. And according to the established multivariate time sequence model, carrying out prediction and generation of abnormal data. This can be achieved by a predictive function of the model, predicting future anomalies from historical data. For example, the VAR model may be used to predict the anomaly frequency components to obtain anomalies over a period of time in the future.

In the embodiment of the invention, for example, a CVAR model is used for carrying out collaborative relation analysis on the multi-element time sequence power abnormal data. Modeling the relationship between a plurality of variables such as abnormal frequency components and amplitude variations, and estimating a coefficient matrix of the model. Then, existence and stability of the synergistic relationship are verified by means of unit root test and the like. If the test result shows that the coordination relation exists, coordination data among the power performance variables can be obtained, wherein the coordination data comprises information such as coefficient matrixes, residual sequences and the like.

In the embodiment of the invention, a proper cooperative vector estimation method is selected according to the characteristics and the analysis purpose of the cooperative data. Common methods include least squares, maximum likelihood, and the like. For example, the coefficient of the co-vector is estimated, and the least squares method may be selected for estimation. And carrying out the coordination vector estimation on the electric performance variable coordination data. This may involve processing and analysis of the coefficient matrix and residual sequence to obtain estimates of the co-vector. For example, for the least squares method, the coefficients of the co-vector may be estimated by minimizing the sum of squares of the residuals. And verifying the accuracy and stability of the estimation result of the cooperative vector. The verification can be performed by methods such as checking residual sequences and confidence intervals of parameter estimation. If the estimation results meet expectations and meet statistical assumptions, it is shown that the cooperative vector estimation is reliable.

In an embodiment of the invention, causal verification of multivariate time series power anomaly data is performed, for example, using a Grangel causal verification. The relationship between the variables is examined and the grange causal test statistic is calculated. The level of significance of the test results is then determined by a significance test. If the test results indicate that a certain variable has significant causal relationship with another variable, then electrical performance variable relationship data may be obtained describing the causal relationship between the variables. For example, if the test results show that a voltage anomaly has a significant causal effect on a current anomaly, causal relationship data between the voltage anomaly and the current anomaly may be obtained.

Preferably, step S22 comprises the steps of:

In the embodiment of the invention, the electric performance spectrum data is subjected to wavelet decomposition to obtain spectrum components on different scales. This may be achieved by wavelet transform algorithms such as Discrete Wavelet Transform (DWT). And carrying out wavelet decomposition on the electric performance spectrum data to obtain spectrum components on different scales. This may be achieved by wavelet transform algorithms such as Discrete Wavelet Transform (DWT). For example, daubechies wavelet decomposition is performed on the power performance spectrum data to obtain spectrum components on different scales, such as spectrum data of a low-frequency scale and a high-frequency scale. And calculating entropy of the frequency spectrum data on each scale to obtain multi-scale entropy data. Shannon entropy or the like may be used as a measure of entropy. And calculating entropy and variance of the spectrum data of each scale to obtain multi-scale entropy data and multi-scale variance data. The multi-scale variance data and the multi-scale entropy data are fused, and the multi-scale characteristic data can be generated by adopting a simple weighted average or other specific fusion methods. And detecting abnormal frequency components of the multi-scale characteristic data through a preset abnormal frequency detection threshold value, and identifying abnormal frequency components exceeding the set threshold value. And extracting a trend component of a certain frequency in the abnormal frequency component data by using a Loess smoothing method to obtain performance trend component data. And subtracting trend component data on the corresponding scale from the multi-scale characteristic data to obtain residual item data. The residual term reflects fluctuations in the abnormal frequency component data without considering long-term trends. Wavelet transformation is carried out on the power performance residual term data, and periodic fluctuation exists on a certain scale. For example, the periodic pattern analysis of the power performance residual term data is selected using an autocorrelation function analysis. And carrying out autocorrelation function calculation on the residual error item data to obtain an autocorrelation coefficient. Then, a period in which significant periodicity exists is identified from the change in the autocorrelation coefficient. For example, if the autocorrelation coefficient exceeds the significance threshold at a certain point in time, it can be considered that a periodic pattern exists at that point in time. The abnormal power generation mode is defined as oscillation with the duration exceeding 3 months and within a specific range, and the periodic analysis result is matched. And comparing the periodic analysis result with the abnormal mode to find out the part meeting the condition. And then, obtaining abnormal power generation mode data according to the matching result, and describing abnormal conditions in the residual error item data.

Preferably, step S3 comprises the steps of:

In the embodiment of the invention, parameters of multi-level power generation network simulation are set according to the characteristics of the power performance manifold evolution data and the simulation purpose, including simulation time period, simulation precision and the like. And selecting a proper multi-level simulation method according to the simulation requirement. Common methods include discrete event simulation, continuous system simulation, and the like. And carrying out multi-level power generation network simulation on the power performance manifold evolution data. This may involve state updates, event handling during simulation. And selecting a proper topological graph generation algorithm according to the characteristics of the simulation data and the requirements of topological graph generation. Common algorithms include depth-first search, breadth-first search, minimum spanning tree algorithm, and the like. For example, the multi-level power network simulation data is analyzed, the position and the connection relation of each power generation group are extracted, and the minimum spanning tree algorithm is selected as the topology graph generation algorithm. And then, according to a minimum spanning tree algorithm, adding nodes and connecting edges to the position information of the generating set to obtain generating set topological graph data. And constructing a mapping relation between the power generation equipment group and the power generation performance by using the graph neural network model through the power generation group topological graph data and the corresponding power generation performance data. This process includes initialization of the model, forward propagation, backward propagation, etc. Training the constructed model and training data to adjust model parameters so as to enable the model parameters to be better fit with the power generation performance data. This typically involves a number of iterative training processes, each iteration using a batch of data for parameter updating. The multi-modal power performance data is divided into a training set and a validation set using a random sampling method. The division may be generally made in a proportion, such as 70% of the data for training and 30% for verification. And training the model by using the training set data, and updating model parameters through a back propagation algorithm so that the model can gradually fit the training set data. Training of multiple epochs, each containing a complete traversal of the entire training set, is typically performed. During each epoch training process, the value of the loss function (e.g., mean square error) is calculated for evaluating the fit of the model. By calculating the difference between the model predictions and the validation set's true tags, for example, the mean square error or decision coefficients are calculated. By evaluating the performance of the model on the verification set, the generalization ability and the prediction accuracy of the model can be judged.

Preferably, step S4 comprises the steps of:

As an example of the present invention, referring to fig. 3, a detailed implementation step flow diagram of step S4 in fig. 1 is shown, where step S4 includes:

In the embodiment of the invention, according to the electric performance variable relation data, the causal relation among the variables is checked by adopting methods such as the Grangel causal test, and the like, the relevant statistics are calculated and the saliency test is carried out, so as to confirm the saliency level of the causal relation. Subsequently, variable data having a significant causal influence is extracted as a part of the power generation amount, the power generation type, and the power generation quality data, based on the inspection result. For example, if the test result shows that a certain variable has a significant influence on the amount of power generation, the type of power generation, or the quality of power generation, the data of the variable is extracted accordingly, thereby obtaining corresponding power generation data.

In the embodiment of the invention, the load rate of the power generation group is calculated for each time period. Load factor is generally defined as the ratio of actual power generation to maximum power generation capacity, typically expressed in percent.

In the embodiment of the invention, the power generation mode is analyzed according to the power generation type data, and the application conditions of various power generation modes under different time periods or conditions are analyzed, for example, coal-fired power generation is mainly adopted in the morning and evening, and solar energy and wind power generation are mainly adopted in the daytime. Subsequently, for each power generation mode, the power generation cost is calculated in consideration of various cost factors including fuel cost, equipment maintenance cost, operation cost, and the like. For example, coal-fired power generation costs $0.05/kilowatt-hour, solar power generation costs $0.08/kilowatt-hour, and wind power generation costs $0.06/kilowatt-hour. And finally, arranging the calculated power generation cost data to generate the power generation cost data.

In the embodiment of the invention, the power generation quality data comprise data of indexes such as voltage stability, frequency stability, harmonic content and the like. These indices are then calculated according to the selected power generation quality index using corresponding calculation formulas or algorithms, for example, calculating values of voltage fluctuation, frequency deviation and harmonic content. For example, an average value of the voltage fluctuation in each period of time in one week may be calculated as the voltage stability index, and likewise, an average value of the frequency deviation and the harmonic content may be calculated as the frequency stability and the harmonic content index.

In the embodiment of the invention, the preset power generation rule of the power generation equipment is formulated according to the technical characteristics, the running conditions and the related regulation standards of the power generation equipment, so that the stability, the economy and the reliability of the power generation system are ensured. For example, considering the characteristics of different power generation equipment, rules such as the operation of the coal-fired power generation unit in a high-load period, the preferential utilization of renewable energy sources by the solar power generation panel and the like are formulated. Then, according to the rules, constraint condition setting is carried out on the power generation quality index data, the power generation cost data and the power generation group load capacity data so as to ensure that the operation of the power generation system meets the expected target. For example, voltage stability and frequency stability of the coal-fired power generation unit are set to meet requirements in a certain range, power generation cost is controlled in a budget range, and load capacity of the power generation unit is adjusted to meet load requirements. And finally, converting the formulated constraint condition into a data form to form power generation constraint condition data.

Step S46: performing multi-objective power generation optimization processing on the power generation constraint condition data by using an artificial bee colony algorithm to generate a multi-objective power generation optimization strategy;

in the embodiment of the invention, the power generation constraint condition data is prepared and is arranged into a format suitable for algorithm processing. Then, the artificial bee colony algorithm is utilized to perform multi-objective power generation optimization processing. In the process, the process of searching the optimal solution by the bees is simulated by initializing the bees colony, setting the searching range, the objective function and the like. The bees search according to the information of the current position and the adjacent information, and select the position with better adaptability to update until the stopping condition is reached. Finally, a multi-objective power generation optimization strategy is generated based on the optimization results, representing the best operational strategy that the power generation system can take in consideration of multiple objectives.

Step S47: and carrying out optimization strategy integration on the power generation performance simulation model through a multi-target power generation optimization strategy to generate an optimized power generation performance simulation model.

In the embodiment of the invention, the applicability and effect of various optimization strategies are comprehensively considered and integrated into the power generation performance simulation model, and the method may involve adjusting model parameters, setting of an optimization algorithm or running modes of the model. Then, the performance of the integrated model is evaluated according to the simulation or emulation of the integrated model, and the model is adjusted and optimized. Finally, an optimized power generation performance simulation model is generated, and the model can effectively consider the influence of a multi-objective power generation optimization strategy and provide decision support for the optimized operation of the power generation system. For example, if the optimization strategies include adjusting the generator set load rate and optimizing the manner of power generation, integrating these strategies into the model may involve modifying the load rate settings, updating the generator set operating mode, and adjusting parameters of the model based on the results of the optimization algorithm, etc. Finally, the performance of the integrated model is evaluated by simulation or emulation.

Preferably, step S46 comprises the steps of:

In the embodiment of the invention, constraint condition data is converted into a format suitable for algorithm processing, for example, the power generation load rate, the power generation cost, the quality index and the like are converted into numerical data. Generating optimization target initial group data according to the power generation constraint condition data, namely randomly generating a certain number of initial solutions according to the constraint condition. Ensuring that the generated initial population data meets constraints of the power generation system, e.g., power generation constraints include maximum load rate, minimum power generation cost, and minimum power generation quality requirements, then the initial population generation process may randomly generate a set of initial solutions that meet these constraints, e.g., different power generation load rates, power generation costs, and quality index combinations. And carrying out sharing communication processing on the initial group data of the optimization target according to preset sharing triggering condition data, namely determining which groups need to be subjected to information exchange and sharing. For example, if the sharing trigger condition is based on the distance between groups, then it may be determined that information is exchanged between groups that are closer in distance. For example, based on geographical position information of power generation facilities, it is determined that power stations closer to each other communicate with each other, and optimization information of each other is shared. And exchanging and integrating optimization information among the multiple groups according to the cooperative strategy so as to promote cooperative optimization among the groups. And solving the multi-objective optimization problem by using an artificial bee colony algorithm according to the multi-colony cooperation strategy to obtain objective initial solution data. The multi-objective problem to be optimized comprises two objectives of maximizing the generated energy and minimizing the generating cost, and the artificial bee colony algorithm is utilized for optimization. The initial solution may be a combination of a set of power generation and cost that are randomly generated. Then, according to the search mechanism and the information exchange rule of the artificial bee colony algorithm, the initial solution is gradually updated until a converged optimization result is obtained. The maximum fitness mapping processing is to weight the fitness value of each target to obtain a comprehensive fitness value. For example, two targets of maximizing the power generation amount and minimizing the power generation cost need to be considered at the same time, the power generation amount may be weighted, and then the weighted power generation amount and the fitness value of the power generation cost are added to obtain the comprehensive fitness value. And then, determining a multi-target power generation optimization strategy according to the comprehensive fitness value.

Preferably, step S5 comprises the steps of:

In the embodiment of the invention, the power data of the power generation equipment group, including current, voltage, power and the like, are collected in real time according to the preset configuration. Each dynamic performance impact factor is monitored and analyzed using real-time power data, for example, by statistical methods or machine learning algorithms to analyze the power signature changes of the power plant under different loads. And carrying out feature processing on the dynamic performance influence data, including feature selection, dimension reduction, standardization and the like, so as to facilitate subsequent modeling and analysis. And transmitting the dynamic performance influence characteristic data to an optimized power generation performance simulation model. In the simulation model, dynamic event response simulation is carried out according to the received characteristic data, different electric power performance events are simulated, and corresponding simulation data are generated. For example, the dynamic performance influence characteristic data processed by the characteristic engineering is transmitted to a pre-built neural network model, and the model carries out dynamic event response simulation according to the characteristic data, such as simulating response behaviors of the power generation equipment when the load suddenly increases. And analyzing the influence degree of each input factor on the power generation performance by using a Lagrange multiplier method according to the response event simulation data. And calculating sensitivity indexes of each factor to obtain power generation performance fault sensitivity data. And identifying key equipment with larger influence on the power generation performance according to the power generation performance fault sensitive data by using a threshold value judging method based on the sensitivity index. For example, based on the result of the Lagrangian sensitivity analysis, a cooling system that has the greatest influence on the power generation performance at a temperature increase is identified as a key power generation device. Based on sensitive data analysis, it is found that a certain generator will cause a decrease in the performance of the whole power generation structure when frequently failed, and thus the generator is identified as a fragile node. And (3) redesigning the power network connection mode according to the identified key power generation equipment and the fragile node, so that the connection between the key equipment is ensured to be more stable, and the influence of the fragile node on the whole power system is reduced.

Preferably, step S6 comprises the steps of:

In the embodiment of the invention, the energy resource types of different areas or nodes are identified according to the power network reconstruction data, and the energy facility information of each node or area is identified and recorded according to the power network reconstruction data. For example, it is determined that there are energy facilities such as solar power stations, wind power stations, and hydroelectric power stations in a certain region based on the power network reconstruction data. Specific information of the energy facilities is recorded, including capacity, power generation mode, geographic position and the like, and energy facility data is generated. For each energy resource type, energy resource characteristic analysis is performed, including characteristics in terms of resource reproducibility, seasonality, predictability, and the like. For example, the solar energy resource is subjected to energy resource characteristic analysis, and characteristics such as sunlight time, illumination intensity and seasonal variation are evaluated; wind energy resources are analyzed, and characteristics such as wind speed change, wind direction change, seasonality and the like of the wind energy resources are evaluated. And recording the analysis result to generate the energy characteristic data of the solar energy and the wind energy resources. And calculating production elasticity indexes of each energy facility according to the energy facility data. The production elasticity index may include indexes in terms of power supply capacity, adjustability, response speed, and the like. And calculating the production elasticity index of each energy facility by using a proper mathematical calculation formula to generate energy elasticity index data. And (3) evaluating the elasticity of the supply side by using a power supply side evaluation algorithm in combination with the energy elasticity index and the market scene energy demand. This may relate to considerations such as response speed on the power supply side, flexibility in energy resource allocation, etc., for example, a certain energy facility may be able to increase output rapidly when there is a high market demand, and decrease output in time when there is a decrease in the market demand, indicating that the facility has a higher supply side elasticity. The assessment algorithm may score the supply side facilities based on these metrics.

Preferably, the power supply side evaluation algorithm formula in step S64 is as follows:

；

In the method, in the process of the invention, Represented as supply-side elasticity assessment data,Represented by the time interval(s) indicated,Denoted as the supply side evaluation start time,Represented as the supply side evaluation termination time,Represented as atWhen the energy source at the supply side generates power,Represented as an estimated time variable,Represented as a mathematical partial derivative symbol,Represented as atThe market energy demand value at the time of the time,Represented as a reference energy demand value,Represented as a power market weight value,Represented as energy characteristic data,Represented as energy facility supply weight values,Angular frequency expressed as a change in market demand,Expressed as energy elasticity index data.

The invention utilizes a power supply side evaluation algorithm which fully considers the time intervalSupply side evaluation start timeSupply side evaluation termination timeIn the presence ofPower generation at the time of supply sideEvaluating time variableIn the presence ofMarket energy demand value at timeReference energy demand valueElectric market weight valueEnergy characteristic dataSupply weight value of energy facilityAngular frequency of market demand changeElastic index data of energyAnd interactions between functions to form a functional relationship:

That is to say, The supply side elasticity ability scoring is achieved by calculating the average value of the supply side elasticity over a given time frame. Sub-itemsRepresenting the power output of the energy source at the supply sideWith respect to timeThe derivative of (a) i.e. the rate of change of power over time directly reflects the dynamic change in the supply side energy yield. If it isWhich means that the supply side can rapidly increase or decrease the energy yield to accommodate the rapid change in market demand. Conversely, if this rate of change is small, the supply side will respond slower to changes in market demand. Thus, the first and second substrates are bonded together,The larger the value of (c), the better the elasticity of the supply side. Sub-itemsRepresenting the energy demand value of the marketAnd the reference energy demand valueNatural logarithm of the ratio of (2). The function of this expression is to quantify the relative change in market demand, when the market demand is at the valueEqual to the reference demand valueWhen the logarithmic value is zero, no change in market demand is indicated. If it isGreater thanThe logarithmic value is positive, indicating an increase in market demand; if it isLess thanThe logarithmic value is negative, indicating a reduced market demand. This logarithmic change is multiplied by the rate of change of the power output of the source of energy on the supply side. The purpose of this is to relate the energy production capacity of the supply side to the change in market demand, and thereby evaluate the adaptability of the supply side at different demand levels. Sub-itemsThe characteristics and importance of the energy supply side can be evaluated,Representing specific properties of the energy source such as reliability, availability or efficiency, etc. WhileRepresenting weighted energy supply capacity, whereinIs a weight coefficient used to reflect the importance of different energy facilities in the overall supply. By combiningAnd (3) withThe multiplication results in a value that combines the energy characteristics and the delivery weight. Angular frequency of this logarithmic function versus market demand variationAnd energy elasticity index dataTogether, by a sine functionTo simulate the periodic variation of market demand, the purpose of this is to relate the characteristics and importance of the energy supply side to the dynamics of the market demand, and to evaluate the adaptability of the supply side under different market conditions. Finally, the formula goes through a limiting process, i.e. withApproaching zero to calculate an average elasticity evaluation value in a time interval. The process involves integrating all the above factors to obtain an evaluation score that integrates energy output power variation, market demand variation, energy characteristics and supply weight. By this evaluation score, supply side facilities can be scored to evaluate their adaptability to different market demand changes. The higher this score, the better the elasticity of the supply side, i.e. the more effective against changes in market demand.

The application has the beneficial effects that the purpose of acquiring the data of the power generation equipment group is to collect the operation parameters and performance indexes related to the power generation equipment so as to comprehensively know the operation state of the equipment. Historical power generation performance data collection helps to build up a performance profile of the device over a historical period of time. Through multi-mode data fusion, data from different sources are comprehensively processed, and the comprehensiveness and accuracy of the data are improved. Manifold learning evolution processing maps high-dimensional multi-modal power performance data into a low-dimensional manifold space, which is helpful for finding the inherent structure and rules of the data. The electric power performance manifold evolution data is used for detecting abnormal power generation modes, identifying abnormal conditions in a power generation system in time and preventing or treating potential problems. The synergistic vector estimation understands the long-term relationship and the common trend between different performance variables. And the causal relation test determines the causal relation among the performance variables, and clarifies the influence mechanism of each factor on the system operation. The power performance manifold evolution data supports multi-level power generation network simulation, and simulates power generation network structures and operation conditions of different levels. And constructing an electric power performance model based on a preset graph neural network model, and predicting the running state and performance of the system through the model. The electrical performance variable relationship data processing evaluates the productivity and efficiency of the electrical power system. The power generation constraint conditions set the limit and the requirement of the definite power generation system under various conditions. The artificial bee colony algorithm performs multi-objective power generation optimization, searches for an optimal power generation scheme, and improves system efficiency and performance. And the power generation performance simulation model is integrated by optimizing strategies, so that the accuracy and applicability of the model are improved. Real-time electric power data acquisition acquires actual data of system operation, analyzes dynamic performance influence factors, and provides references for timely adjustment and coping. And optimizing the power generation performance simulation model to simulate dynamic event response, and simulating the response and performance of the system under different conditions. And the response event simulation data are used for reconstructing the power network, optimizing the structure and configuration of the power system, and improving the reliability and stability. The power supply side energy analysis comprehensively evaluates the power supply capacity and resource allocation condition of the power system, provides basis for management and planning of the power supply side, and evaluates the flexibility and adaptability of the power supply side in coping with various changes and challenges.

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The power performance data based simulation method is characterized by comprising the following steps of:

Step S3: performing multi-level power generation network simulation according to the power performance manifold evolution data to generate multi-level power network simulation data; carrying out electric power performance model construction through multi-level electric power network simulation data based on a preset graph neural network model to obtain a power generation performance simulation model; the step S3 specifically includes:

step S35: model training is carried out on the initial power generation performance simulation model by using training set data, and model verification is carried out by using verification set data, so that a power generation performance simulation model is obtained;

step S5: real-time power data acquisition is carried out according to the power generation equipment group data, and real-time power data are generated; analyzing dynamic performance influence factors of the real-time power data to generate dynamic performance influence data; performing dynamic event response simulation on the dynamic performance influence data by using the optimized power generation performance simulation model to generate response event simulation data; carrying out power network reconstruction processing according to the response event simulation data to generate power network reconstruction data; the step S5 specifically includes:

step S58: carrying out power network reconstruction processing based on the key power generation equipment data and the power generation structure fragile node data to generate power network reconstruction data;

2. The power performance data based simulation method of claim 1 wherein step S1 comprises the steps of:

3. The power performance data based simulation method of claim 2 wherein step S2 comprises the steps of:

4. A power performance data based simulation method according to claim 3, wherein step S22 comprises the steps of:

5. The power performance data based simulation method of claim 1 wherein step S4 comprises the steps of:

6. The power performance data based simulation method of claim 5 wherein step S46 comprises the steps of:

7. The power performance data based simulation method of claim 1 wherein step S6 comprises the steps of:

8. The power performance data-based simulation method according to claim 7, wherein the power supply side evaluation algorithm formula in step S64 is as follows:

；