A virtual power plant energy management method and system based on air conditioning load

Abstract

The present invention discloses a virtual power plant energy management method and system based on air conditioning load, which relates to the field of energy management. The management method includes: obtaining electricity consumption data within the management scope of the virtual power plant, integrating operation information data, and statistically dividing the operation information data according to operation rules and potential allocation models to generate a batch monitoring management set; constructing a virtual power plant electric energy trading framework based on the batch monitoring management set, analyzing the power supply and demand difference and the power balance relationship between virtual power plants, and outputting energy scheduling results within the management scope of the virtual power plant; using air conditioning load data as a demand response resource, optimizing the energy scheduling results to generate air conditioning control instructions, and recording and decomposing the air conditioning control instructions, and outputting the control status of the air conditioner. Through the optimized scheduling of the air conditioning load, the present invention can reduce the demand for air conditioning load during peak power demand periods, thereby effectively reducing system load pressure and avoiding excessive energy consumption.

Description

Technical Field

[0001] The present invention relates to the field of energy management, and in particular to a virtual power plant energy management method and system based on air conditioning load. Background Art

[0002] A virtual power plant (VPP) is a system that uses information technology to intelligently dispatch and manage dispersed, heterogeneous distributed energy resources (such as wind power, solar power, energy storage, electric vehicles, electric water heaters, air conditioners, etc.), thereby presenting large-scale electricity production and consumption capabilities in a "virtual" form in the power market. It achieves supply and demand balance and improves the stability and economy of the power system through demand response, energy storage, and predictive scheduling.

[0003] Energy management in a virtual power plant (VPP) is one of its core capabilities, determining whether the VPP can achieve efficient scheduling, load balancing, economical operation, and the ability to participate in market transactions. The so-called VPP energy management refers to the entire process of intelligent monitoring, optimized scheduling, and dynamic coordination of the distributed power sources, energy storage devices, and load resources aggregated by the VPP.

[0004] However, in the existing technology, virtual power plants often ignore the load changes of air conditioners within the management scope during the process of implementing energy scheduling management, resulting in large fluctuations in the usage patterns of air conditioners when a large number of air conditioners participate in energy scheduling, leading to inaccurate energy scheduling results, excessive energy consumption or insufficient energy for other electrical equipment. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes a virtual power plant energy management method and system based on air conditioning load to achieve the purpose of avoiding excessive energy consumption.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a virtual power plant energy management method based on air conditioning load, the management method comprising:

[0008] Obtain electricity consumption data within the virtual power plant management scope, integrate operational information data, and statistically divide the operational information data according to operational rules and potential allocation models to generate batch monitoring management sets;

[0009] Build a virtual power plant energy trading framework based on batch monitoring management sets, analyze the power supply and demand gap and the power balance relationship between virtual power plants, and output energy scheduling results within the virtual power plant management scope;

[0010] Air conditioning load data is used as a demand response resource, and the energy scheduling results are optimized to generate air conditioning control instructions. The air conditioning control instructions are recorded and decomposed, and the control status of the air conditioner is output.

[0011] Preferably, statistically dividing the operation information data according to the operation rules and the potential allocation model to generate a batch monitoring management set includes:

[0012] Analyze the joint probability of the Naive Bayes model based on the hyperparameters of the labels and values ​​involved in the operational information data. Based on the analysis results, optimize the Naive Bayes model to determine the correlation between labels and values.

[0013] Based on the correlation between labels and the correlation between values, the weights of labels and values ​​are analyzed respectively, and the main features of the operation information data are output according to the weight results. The classification category labels of the operation information data are determined in combination with the operation rules. The classification category labels include industry classification, regional classification and user classification;

[0014] Analyze the joint distribution correlation between industry classification, regional classification, user classification and subject characteristics, cluster the correlation, and train the naive Bayes model based on the clustering results to obtain the potential distribution model;

[0015] Based on the training results, classification models with industry classification, regional classification and user classification as themes are generated respectively. The classification models are used to collect statistics on operation information data according to industry, region and user to obtain a batch monitoring management set.

[0016] Preferably, a virtual power plant electricity energy trading framework is constructed based on the batch monitoring management set, the power supply and demand difference and the power balance relationship between virtual power plants are analyzed, and the energy scheduling results within the virtual power plant management scope are output, including:

[0017] Obtain the historical power generation of each energy source within the virtual power plant management scope, and combine it with environmental factor data to predict the energy regeneration capacity, and build a virtual power plant electricity energy trading framework based on the regeneration capacity and electricity price;

[0018] Based on batch monitoring management, the load fluctuation trend of virtual power plant management is predicted in batches, and the power supply and demand gap in the virtual power plant management process is analyzed according to the energy regeneration amount and load fluctuation trend;

[0019] Based on the virtual power plant electricity energy trading framework, the electricity supply and demand gap and the power balance relationship between virtual power plants are evaluated, and the difference is determined according to the power balance relationship to analyze the scheduling mechanism of virtual power plant electricity energy trading.

[0020] Preferably, the historical power generation of each energy source within the management scope of the virtual power plant is obtained, and the renewable energy amount of the energy is predicted in combination with the environmental factor data. The virtual power plant electric energy trading framework is constructed based on the renewable energy amount and the electricity price, including:

[0021] Extract the historical power generation of each energy source within the virtual power plant management scope based on the grid monitoring equipment, and obtain the environmental factor data within the virtual power plant management scope according to the demand during the forecast period;

[0022] Define forecast consecutive days and forecast consecutive weeks based on historical power generation, analyze the number of power generation change periods and power generation data within the forecast consecutive days, and calculate the daily power generation change factor and daily average change factor;

[0023] Analyze and predict the number of power generation change periods and power generation data within consecutive weeks, calculate the weekly power generation change factor and the weekly average change factor, and combine the calculation results with environmental factor data to predict the amount of energy regeneration;

[0024] Based on the prediction results, the virtual power plant electricity energy trading rules are designed, the virtual power plant electricity energy trading framework is constructed, and the energy cost within the virtual power plant management scope under the predicted regeneration conditions is analyzed.

[0025] Preferably, combining the calculation results with environmental factor data to predict the amount of energy regeneration includes:

[0026] Determine the simulated power generation during the forecast period based on the daily average change factor and the weekly average change factor, and determine the factors influencing the energy power generation results based on the environmental factor data;

[0027] Based on the influencing factors, the ratio of the daily average change factor to the weekly average change factor in the simulated power generation is considered respectively, and the sum of square errors of the simulated power generation is analyzed according to the ratio results;

[0028] The comprehensive weight factor is determined by minimizing the sum of squared errors, and the selection results of the forecast consecutive days and forecast consecutive weeks are determined based on the comprehensive weight factor, and the forecast regeneration amount of energy is obtained according to the selection results;

[0029] Obtain day-ahead electricity price data, and formulate quotation rules for virtual power plants based on the predicted energy regeneration volume. Select trading markets based on the quotation results to provide a basis for building a virtual power plant electricity energy trading framework.

[0030] Preferably, the calculation formula for the predicted regeneration amount of energy is:

[0031] ;

[0032] Where, E a,b represents the predicted energy regeneration amount in time period b during the a-th acquisition process, ω represents the comprehensive weight factor, R represents the total number of predicted consecutive days, G r,b represents the power generation in time period b during the rth forecast consecutive day, represents the average daily power generation of the rth forecast consecutive day, D represents the total number of forecast consecutive weeks, and D d,b represents the power generation in the b time period of the d-th consecutive week, represents the weekly average power generation of the dth forecast consecutive week, Indicates the average power generation within the management scope of the virtual power plant.

[0033] Preferably, the load fluctuation trend of the virtual power plant management is predicted in batches based on the batch monitoring management set, and the power supply and demand difference in the virtual power plant management process is analyzed according to the energy regeneration amount and the load fluctuation trend, including:

[0034] Based on the batch monitoring management set, the load simulation targets and load evaluation indicators corresponding to the industry classification, regional classification and user classification are determined respectively, and the random sampling distribution and parameters are set for random sampling;

[0035] Based on random sampling results and statistical simulation technology, a difference matrix is ​​constructed to generate random operation information samples, and a trend prediction model is constructed by simplifying the random operation information samples using attribute theory based on information theory;

[0036] Based on the trend prediction model, the load fluctuation trends corresponding to industry classification, regional classification and user classification are predicted respectively, and the electricity demand within the virtual grid management range is analyzed according to the load fluctuation trends;

[0037] The difference between electricity demand and energy regeneration forecast results is compared, and the energy profit and loss status in the virtual power plant management process is determined based on the comparison results. The electricity supply and demand difference is analyzed based on the energy profit and loss status.

[0038] Preferably, constructing a difference matrix based on random sampling results and statistical simulation technology to generate random operation information samples, and using attribute theory based on information theory to simplify the random operation information samples to construct a trend prediction model includes:

[0039] The reliability of random sampling results is judged based on load evaluation index simulation, and the random sampling results are optimized and adjusted according to the reliability results until the reliability meets the load simulation target to obtain random operation information samples;

[0040] Map random operation information samples to the attribute space to construct a discernibility matrix, traverse and search the data in the discernibility matrix until an empty set is encountered, and then stop searching. The reduced triangular matrix is ​​obtained based on the search results.

[0041] Analyze the original attribute assignments of random operation information samples in the reduced triangular matrix and verify them with the original attribute assignments corresponding to the discernibility matrix. Based on the verification results, determine the reduction accuracy of the reduced triangular matrix.

[0042] The random operation information samples corresponding to the simplified triangular matrix that meets the reduction requirements are used as the data set, and the data set is used as input, the load fluctuation is used as output and the Monte Carlo algorithm is combined to build a trend prediction model.

[0043] Preferably, the air conditioning load data is used as a demand response resource, the energy scheduling result is optimized to generate an air conditioning control instruction, and the air conditioning control instruction is recorded and decomposed, and the control status of the air conditioner is output including:

[0044] Obtain the power load data of various air conditioners within the virtual power plant management scope as demand response resources, determine the demand response amount of the air conditioner power load data within the forecast period, and adjust the energy scheduling results based on the demand response amount;

[0045] Determine the control strategy for each type of air conditioner based on the adjusted energy scheduling results, determine the air conditioner control instructions, and send the air conditioner control instructions to the device gateway layer through the message transmission protocol;

[0046] The device gateway layer receives and parses the air conditioning control instructions, obtains the power information required for each type of air conditioner to adjust the control status of the air conditioner, and evaluates the execution status of the air conditioners involved in the control.

[0047] In a second aspect, the present invention further provides a virtual power plant energy management system based on air conditioning load, the management system comprising:

[0048] The monitoring and management set generation module is used to obtain electricity consumption data within the management scope of the virtual power plant, integrate operational information data, and statistically divide the operational information data according to operational rules and potential allocation models to generate batch monitoring and management sets;

[0049] The energy dispatch result output module is used to build a virtual power plant energy trading framework based on the batch monitoring management set, analyze the power supply and demand gap and the power balance relationship between virtual power plants, and output the energy dispatch results within the virtual power plant management scope;

[0050] The air conditioning control status output module is used to use the air conditioning load data as a demand response resource, optimize the energy scheduling results to generate air conditioning control instructions, record and decompose the air conditioning control instructions, and output the control status of the air conditioning.

[0051] The beneficial effects of the present invention are:

[0052] 1. The present invention uses air-conditioning load data as a demand response resource. The virtual power plant can flexibly adjust the air-conditioning load according to the fluctuation of electricity demand, thereby balancing the power supply and demand of the virtual power plant, greatly improving the adaptability of the power system in the face of load fluctuations. At the same time, the power energy trading framework is used to analyze the supply and demand gap, which can predict the load changes of the power system in advance. Through the optimized scheduling of air-conditioning load, the demand for air-conditioning load can be reduced during peak electricity demand periods, thereby effectively reducing the system load pressure and avoiding excessive energy consumption.

[0053] 2. The present invention predicts short-term and medium-term renewable energy output by combining historical power generation with environmental factors such as weather, temperature, and season. It can respond to the strong volatility of renewable energy in advance, avoid the abandonment of excess power generation or passive emergency response due to insufficient power generation. At the same time, when the difference between electricity supply and demand is evaluated, the virtual power plant can adopt different scheduling mechanisms according to the balance relationship, realize the automation and intelligence of energy scheduling, and improve the economic benefits of the virtual power plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0055] Figure 1 is a flow chart of a virtual power plant energy management method based on air conditioning load according to an embodiment of the present invention;

[0056] Figure 2 This is a principle block diagram of a virtual power plant energy management system based on air conditioning load according to an embodiment of the present invention.

[0057] In the picture:

[0058] 1. Monitoring and management set generation module; 2. Energy scheduling result output module; 3. Air conditioning control status output module. DETAILED DESCRIPTION

[0059] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0060] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0061] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0062] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0063] See also Figure 1 The present invention provides a virtual power plant energy management method based on air conditioning load, the management method comprising:

[0064] Step S1: obtain electricity consumption data within the management scope of the virtual power plant, integrate the operation information data, and statistically divide the operation information data according to the operation rules and potential allocation model to generate a batch monitoring management set.

[0065] In one embodiment, in the process of statistically dividing the operation information data according to the operation rules and the potential allocation model and generating a batch monitoring management set, the joint probability of the naive Bayes model can be analyzed according to the hyperparameters of the labels and values ​​involved in the operation information data, and the naive Bayes model can be optimized based on the analysis results to judge the correlation between labels and the correlation between values; the weights of labels and values ​​are analyzed respectively based on the correlation between labels and the correlation between values, and the main features of the operation information data are output according to the weight results, and the classification category labels of the operation information data are determined in combination with the operation rules, and the classification category labels include industry classification, regional classification and user classification; the industry is analyzed The joint distribution between classification, regional classification and user classification and the main features is correlated, and the correlation is clustered. The naive Bayes model is trained based on the clustering results to obtain the potential allocation model; based on the training results, classification models with industry classification, regional classification and user classification as themes are generated respectively, and the classification models are used to count the operation information data according to industry, region and user to obtain the batch monitoring management set. At the same time, in the process of training the naive Bayes model, in order to prevent the model from overfitting, the cross-validation mechanism is introduced in the training process, and the model structure is simplified. At the same time, regularization strategies (such as L1 / L2) and early stopping method are used to control the number of training rounds. The naive Bayes model evaluation adopts comprehensive measurement of indicators such as accuracy, precision, recall rate and F1 score. Finally, the effectiveness and robustness of the naive Bayes model are verified based on the performance of accuracy and F1 score on the test set.

[0066] It should be explained that in the process of obtaining batch monitoring management sets, the joint probability of various labels (such as industry, region, user enterprise type, etc.) and values ​​(such as sales volume, electricity consumption, temperature, etc.) in the data is evaluated through hyperparameter analysis of labels and values. Hyperparameters are parameters that affect model performance, including weights, biases, distribution parameters, etc. By adjusting and optimizing hyperparameters, the relationship between data can be more accurately represented, and then the potential allocation model can be optimized, so as to better predict and identify the relationship between labels and values.

[0067] Optimize the potential allocation model through statistical and machine learning methods to determine the correlation between different labels and between labels and values. Specifically, use techniques such as Pearson correlation coefficient, mutual information, and covariance analysis to analyze the relationship between labels and values. Based on the correlation between labels and values, calculate the weight of each label and value. The purpose is to determine which factors are more important to the overall operational data through quantitative analysis.

[0068] According to the determined label weights and subject characteristics, combined with the existing operation rules, appropriate classification category labels are assigned to each operation information data. The classification labels include but are not limited to: industry classification, regional classification, and user classification, which can then be automatically classified and included in different management, making operation management more orderly and efficient. After obtaining the industry, region, and user classification labels, the joint distribution between the classification labels and subject characteristics is analyzed, and cluster analysis is performed based on the correlation of the joint distribution. The naive Bayes classification model is trained using the clustering results and the labels and numerical weights. Naive Bayes is a classification method based on probability theory. It learns how to predict the classification results of data based on characteristics such as industry classification, regional classification, and user classification, and generates a set of classification models. Based on the training results, the classification model is generated to perform statistics and analysis on the operation information data for the three dimensions of industry, region, and user.

[0069] Specifically, the industry classification model can analyze the trend of operational data based on industry characteristics; the regional classification model can analyze data such as electricity demand based on regional characteristics; the user classification model can analyze and predict individual needs based on user behavior, and generate batch monitoring management sets through the classification model.

[0070] Step S2: construct a virtual power plant electricity energy trading framework based on the batch monitoring management set, analyze the power supply and demand difference and the power balance relationship between virtual power plants, and output the energy scheduling results within the management scope of the virtual power plant.

[0071] In one embodiment, a virtual power plant energy trading framework is constructed based on the batch monitoring management set, the power supply and demand gap and the power balance relationship between virtual power plants are analyzed, and the energy scheduling results within the virtual power plant management scope are output, including:

[0072] Obtain the historical power generation of each energy source within the virtual power plant management scope, and combine it with environmental factor data to predict the energy regeneration capacity, and build a virtual power plant electricity energy trading framework based on the regeneration capacity and electricity price;

[0073] Based on batch monitoring management, the load fluctuation trend of virtual power plant management is predicted in batches, and the power supply and demand gap in the virtual power plant management process is analyzed according to the energy regeneration amount and load fluctuation trend;

[0074] Based on the virtual power plant electricity energy trading framework, the electricity supply and demand gap and the power balance relationship between virtual power plants are evaluated, and the difference is determined according to the power balance relationship to analyze the scheduling mechanism of virtual power plant electricity energy trading.

[0075] Specifically, in the process of obtaining the historical power generation of each energy source within the management scope of the virtual power plant, predicting the energy regeneration amount in combination with environmental factor data, and constructing a virtual power plant electric energy trading framework based on the regeneration amount and electricity price, the historical power generation of each energy source within the management scope of the virtual power plant can be extracted based on the power grid monitoring equipment, and the environmental factor data within the management scope of the virtual power plant can be obtained according to the demand of the forecast period; the forecast consecutive days and forecast consecutive weeks are defined according to the historical power generation, and the number of power generation change periods and power generation data within the forecast consecutive days are analyzed, and the daily power generation change factor and the daily average change factor are calculated; the number of power generation change periods and power generation data within the forecast consecutive weeks are analyzed, and the weekly power generation change factor and the weekly average change factor are calculated, and the calculation results are combined with the environmental factor data to predict the energy regeneration amount; the virtual power plant electric energy trading rules are designed based on the forecast results, the virtual power plant electric energy trading framework is constructed, and the energy cost within the management scope of the virtual power plant under the forecast regeneration amount condition is analyzed.

[0076] Specifically, combining the calculation results with the environmental factor data to predict the regeneration amount of energy includes: judging the simulated power generation in the prediction period according to the daily average change factor and the weekly average change factor, and determining the influencing factors affecting the energy power generation results based on the environmental factor data; considering the ratio of the daily average change factor and the weekly average change factor in the simulated power generation based on the influencing factors, and analyzing the error sum of squares of the simulated power generation according to the ratio results; determining the comprehensive weight factor by minimizing the error sum of squares, determining the selection results of the predicted consecutive days and the predicted consecutive weeks based on the comprehensive weight factor, and obtaining the predicted regeneration amount of energy according to the selection results; obtaining the day-ahead electricity price data, and formulating the quotation rules of the virtual power plant in combination with the predicted regeneration amount of energy, selecting the trading market according to the quotation results, so as to provide a basis for the construction of the virtual power plant electricity energy trading framework.

[0077] The calculation formula for the predicted regeneration amount of energy is:

[0078] ;

[0079] Where, E a,b represents the predicted energy regeneration amount in time period b during the a-th acquisition process, ω represents the comprehensive weight factor, R represents the total number of predicted consecutive days, G r,b represents the power generation in time period b during the rth forecast consecutive day, represents the average daily power generation of the rth forecast consecutive day, D represents the total number of forecast consecutive weeks, and D d,b represents the power generation in the b time period of the d-th consecutive week, represents the weekly average power generation of the dth forecast consecutive week, Indicates the average power generation within the management scope of the virtual power plant.

[0080] It should be explained that in the process of building a virtual power plant electricity energy trading framework, historical power generation data is extracted, which can be specifically extracted from power grid monitoring equipment, usually at the hourly or 15-minute level, to provide high-quality training samples for subsequent prediction models. At the same time, environmental factor data including light intensity, wind speed, temperature, humidity, and weather conditions are obtained. For the future forecast day or forecast week (such as the next week or the next three days), the collection frequency can be adaptively modified according to actual needs, for example, every 5 minutes or one hour. Environmental factors are key driving variables for the output of renewable energy (especially wind and solar energy). Representative continuous time periods are selected based on historical power generation data. It is assumed that the forecast continuous day is 3 to 5 days and the forecast continuous week is 1 to 2 weeks. A reference basis for future forecasts is established by selecting a reasonable historical time window.

[0081] The number of time slices with obvious changes in electricity consumption within a day or a week is analyzed, the ratio or difference of electricity consumption changes in each period is analyzed, and the average value of the electricity consumption change factor of a certain day / week is determined to reflect the fluctuation trend of energy output in the time dimension, providing support for simulation generation and regression modeling. Since sunlight has the greatest impact on the output of photovoltaic systems and wind speed has a significant impact on wind power systems, it is clear which environmental variables dominate the power generation of each energy source.

[0082] Through regression model or weighted function, the following are considered: daily average change factor, weekly average change factor, and environmental impact factor weights to simulate the power generation value at each future moment. At the same time, the square sum of the errors of the simulated power generation is calculated based on the simulated power generation to measure the difference between the simulated power generation and the true value, and the square sum of the errors is minimized to determine the comprehensive weight factor to obtain the optimal comprehensive weight distribution, making the prediction more accurate. According to the weight factor with the minimum error, it is inferred which "continuous day" and "continuous week" samples are more in line with the current prediction target, thereby optimizing the selection of training samples and improving the generalization ability of the model.

[0083] Then, based on the predicted energy regeneration amount and the day-ahead electricity price (which can be obtained from the market interface or historical data), the rule setting idea can be based on the principle that supply exceeds demand → bid a low price to seize the market, and demand exceeds supply → bid a slightly higher price to prioritize quantity. The purpose is to formulate a reasonable bidding strategy to maximize profits or grid acceptance. In the process of selecting a trading market, the day-ahead market or the real-time market can be selected to provide decision-making support for virtual power plants to participate in market transactions.

[0084] Specifically, in the process of predicting the load fluctuation trend of virtual power plant management in batches based on the batch monitoring management set, and analyzing the power supply and demand difference in the virtual power plant management process according to the energy regeneration and load fluctuation trend, the load simulation targets and load evaluation indicators corresponding to the industry classification, regional classification and user classification can be determined based on the batch monitoring management set, and the distribution and parameters of random sampling can be set for random sampling; a difference matrix is ​​constructed based on the random sampling results and statistical simulation technology to generate random operation information samples, and a trend prediction model is constructed by simplifying the random operation information samples using the attribute theory based on information theory; based on the trend prediction model, the load fluctuation trends corresponding to the industry classification, regional classification and user classification are predicted respectively, and the power demand within the virtual power grid management scope is analyzed according to the load fluctuation trend; the power demand is compared with the energy regeneration prediction results, and the energy profit and loss status in the virtual power plant management process is determined according to the comparison results, and the power supply and demand difference is analyzed based on the energy profit and loss status.

[0085] Among them, in the process of constructing a difference matrix based on random sampling results and statistical simulation technology to generate random operation information samples, and using the attribute theory based on information theory to simplify the random operation information samples to construct a trend prediction model, the reliability of the random sampling results can be judged based on the load evaluation index simulation, and the random sampling results can be optimized and adjusted according to the reliability results until the reliability meets the load simulation target to obtain the random operation information samples; the random operation information samples are mapped to the attribute space to construct a difference matrix, and the data in the difference matrix is ​​traversed and searched until an empty set is encountered, and the search is stopped, and a reduced triangular matrix is ​​obtained according to the search results; the original attribute assignments of the random operation information samples in the reduced triangular matrix are analyzed and verified with the original attribute assignments corresponding to the difference matrix, and the simplification accuracy of the reduced triangular matrix is ​​judged according to the verification results; the random operation information samples corresponding to the reduced triangular matrix that meets the simplification requirements are used as a data set, and the data set is used as input, and the load fluctuation is used as output to combine with the Monte Carlo algorithm to construct a trend prediction model.

[0086] It needs to be explained that the corresponding load simulation targets are determined according to industry classification, regional classification, and user classification, and corresponding load evaluation indicators are set, such as maximum load, minimum load, load fluctuation range, etc. The purpose is to set differentiated load simulation targets for different types of users (industries, regions, user types). The load evaluation indicators are used to quantify the characteristics of load fluctuations and are the basis for subsequent forecasting analysis and optimization.

[0087] Set the random sampling distribution of load data, select the probability distribution function (including but not limited to normal distribution, Poisson distribution, etc.) to simulate load fluctuations, and determine the parameters of random sampling, such as sample size, sampling frequency, sampling range, etc. The purpose is to use statistical simulation technology to construct an efficient sample data set for load fluctuation prediction, and obtain load data samples through random sampling to simulate load fluctuations during operation.

[0088] Based on the random sampling results, a difference matrix is ​​used to generate random operation information samples. The difference matrix is ​​used to measure the differences between different operation information samples. By comparing the differences between samples, the diversity of load fluctuations is evaluated. The difference matrix helps to understand the similarity between different load data samples and provides a basis for further trend prediction. At the same time, the random operation information samples reflect the sample set of simulated load fluctuation trends.

[0089] The attribute theory simplification method in information theory is used to simplify random operation information samples and remove redundant or unimportant sample features. Through the simplified samples, attribute simplification can improve the accuracy and efficiency of the model and avoid overfitting. The simplified samples will more effectively capture the trend of load fluctuations. It should be explained that in the process of simplifying random operation information samples based on the attribute theory of information theory, the cross-fusion of rough set theory and feature selection method based on information theory is mainly used to achieve the simplification of random operation information samples.

[0090] The reliability of random sampling results is judged based on load evaluation indicators to evaluate whether the samples meet the expected load simulation targets. Reliability judgment ensures the representativeness of the samples and the reliability of the predictions, reduces the impact of low-quality data on the model, maps random operation information samples to the attribute space, constructs a difference matrix, performs a traversal search, and stops searching when an empty set is encountered to obtain a reduced triangular matrix, which is an optimized data structure containing the minimum necessary information. The difference matrix can help understand the differences between each data sample, and the reduced triangular matrix further simplifies and extracts the most valuable prediction information. The reduced triangular matrix can be used as the core data set for subsequent analysis to reduce unnecessary calculations and data processing.

[0091] The samples in the simplified triangular matrix are verified to ensure that the simplified data set can effectively reflect the load fluctuation trend and that key features are not lost in the process of data simplification. The optimized random operation information samples are used as input and the load fluctuation is used as output. The Monte Carlo algorithm is combined with the Monte Carlo algorithm for simulation and prediction. The Monte Carlo algorithm simulates the load fluctuation trend under different scenarios through multiple random sampling and calculations. Based on the output of the trend prediction model (i.e. the predicted load fluctuation trend), combined with the energy regeneration prediction results, the power supply and demand gap is analyzed. By comparing the load demand and renewable energy supply, the operating status of the virtual power plant is evaluated to help the dispatching system make flexible adjustments.

[0092] Step S3: Use the air conditioning load data as a demand response resource, optimize the energy scheduling result to generate an air conditioning control instruction, record and decompose the air conditioning control instruction, and output the control status of the air conditioning.

[0093] In one embodiment, in the process of using air-conditioning load data as demand response resources, optimizing energy scheduling results to generate air-conditioning control instructions, recording and decomposing the air-conditioning control instructions, and outputting the control status of the air-conditioning, the power load data of various air-conditioners within the management scope of the virtual power plant can be obtained as demand response resources, and the demand response amount of the air-conditioning power load data in the prediction period can be judged, and the energy scheduling results can be adjusted based on the demand response amount; the control strategies of various air-conditioners are determined according to the adjusted energy scheduling results, the air-conditioning control instructions are determined, and the air-conditioning control instructions are sent to the device gateway layer through the message transmission protocol; the device gateway layer receives and parses the air-conditioning control instructions, obtains the power information required for execution of various air-conditioners, adjusts the control status of the air-conditioners, and evaluates the effectiveness of the execution status of the air-conditioners involved in the control.

[0094] It needs to be explained that the core purpose of using air-conditioning load as a demand response resource to optimize the complete closed-loop control process of the virtual power plant energy dispatch results is to achieve flexible response on the power demand side through the adjustability of air-conditioning, and to incorporate this response capability into the energy dispatch system, thereby achieving load peak shaving and valley filling, supply and demand balance and operation optimization. It first collects historical power load data of various air-conditioning (split air-conditioning, VRV multi-split units, water-cooled units, magnetic levitation chillers, intelligent variable frequency air-conditioning) within the management scope of the virtual power plant, and the real-time power curve of the air-conditioning, to provide basic data for subsequent prediction and regulation, identify the fluctuation characteristics of air-conditioning load, and use it as a dispatch resource. The system uses important inputs to determine changes in user comfort levels within the forecast period, and combines load data to estimate the adjustable load of the air conditioner, i.e., the demand response amount. This demand response amount is fed back to the virtual power plant dispatch system to dynamically adjust the energy supply and demand matching, thereby improving the flexibility and economy of power system dispatch. During peak hours, the air conditioner load is reduced to avoid system overload; during off-peak hours, the utilization efficiency is improved. Based on the dispatch optimization results, a personalized control strategy is generated for each type of air conditioner, including control type (temperature set point change, on / off control, fan speed adjustment, etc.), control duration, start time, recovery strategy, etc., and the demand response results are concretized into executable control instructions.

[0095] The control strategy is converted into standardized instructions (such as Modbus, BACnet, and MQTT protocol formats). Each instruction contains the target device, control parameters, execution time, etc. The purpose is to build a standard control instruction set to ensure that the control commands can be received and executed by the actual system. The control instructions are sent to the device gateway layer through the message transmission protocol (such as MQTT, Modbus, etc.), realizing the command transmission path from the virtual power plant control center to the device layer.

[0096] The gateway parses the received control instructions, translates them into control actions that can be recognized by the device layer, obtains the control content (such as adjusting the temperature set point to 25°C) and executes it, thereby ensuring the accurate implementation of the control strategy and compatibility with different types of air-conditioning equipment. At the same time, it records the execution status of the air-conditioning control in real time, including: whether it is successfully executed and the actual power change, providing data basis for subsequent control effect evaluation.

[0097] In the process of evaluating the effectiveness of the air conditioner execution status involved in the control, power regulation error and response time can be used as evaluation indicators to construct a control closed-loop mechanism. The power regulation error is used to measure the deviation between the actual air conditioner power output and the expected control value, and the response time is used to evaluate the speed and timeliness of the command response. Assuming that during the actual measurement process, the air conditioner status data is collected every 10 seconds, the results show that the control response time is an average of 14.6 seconds, the average power regulation error is ±0.23kW, and the control success rate is 97.8%. This shows that the air conditioner execution end can respond to the control signal stably and efficiently, thereby ensuring the effectiveness of the overall load regulation strategy and closed-loop performance.

[0098] It should be explained that when obtaining historical power generation and air-conditioning load data, it is necessary to pre-process the historical power generation and air-conditioning load data to detect missing data, noise, abnormal values ​​or jump points in the data to avoid prediction bias.

[0099] See also Figure 2 The present invention also provides a virtual power plant energy management system based on air conditioning load, the management system comprising:

[0100] Monitoring and management set generation module 1 is used to obtain electricity consumption data within the management scope of the virtual power plant, integrate operation information data, and statistically divide the operation information data according to operation rules and potential allocation models to generate batch monitoring and management sets;

[0101] Energy dispatch result output module 2 is used to build a virtual power plant energy trading framework based on the batch monitoring management set, analyze the power supply and demand gap and the power balance relationship between virtual power plants, and output the energy dispatch results within the virtual power plant management scope;

[0102] The air conditioning control state output module 3 is used to use the air conditioning load data as a demand response resource, optimize the energy scheduling results to generate air conditioning control instructions, record and decompose the air conditioning control instructions, and output the control state of the air conditioning.

[0103] In summary, with the help of the above technical solutions of the present invention, the present invention uses air-conditioning load data as a demand response resource, and the virtual power plant can flexibly adjust the air-conditioning load according to the fluctuation of electricity demand, thereby balancing the power supply and demand of the virtual power plant, greatly improving the adaptability of the power system during load fluctuations, and at the same time using the electric energy trading framework to analyze the supply and demand difference, it can predict the load changes of the power system in advance, and through the optimized scheduling of the air-conditioning load, it can reduce the demand for air-conditioning load during the peak period of electricity demand, thereby effectively reducing the system load pressure and avoiding excessive energy consumption. The present invention predicts the short-term and medium-term renewable energy output by combining historical power generation with environmental factors such as weather, temperature, and season, and can respond to the problem of strong volatility of renewable energy in advance, avoid the abandonment of excess power generation or passive emergency response due to insufficient power generation, and at the same time, when the power supply and demand difference is evaluated, the virtual power plant can adopt different scheduling mechanisms according to the balance relationship, realize the automation and intelligence of energy scheduling, and improve the economic benefits of the virtual power plant.

[0104] Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0105] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A virtual power plant energy management method based on air conditioning load, characterized in that: include: Obtain electricity consumption data within the virtual power plant's management scope, integrate operational information data, and analyze the joint probability of the Naive Bayes model based on the hyperparameters of the labels and values ​​involved in the operational information data. Based on the analysis results, optimize the Naive Bayes model to determine the correlation between labels and values. Based on the correlation between labels and the correlation between values, the weights of labels and values ​​are analyzed respectively, and the main features of the operation information data are output according to the weight results. The classification category labels of the operation information data are determined in combination with the operation rules. The classification category labels include industry classification, regional classification and user classification; Analyze the joint distribution correlation between industry classification, regional classification, user classification and subject characteristics, cluster the correlation, and train the naive Bayes model based on the clustering results to obtain the potential distribution model; Based on the training results, classification models are generated based on industry classification, regional classification, and user classification. The classification models are used to collect statistics on operational information data according to industry, region, and user to obtain a batch monitoring management set. Obtain the historical power generation of each energy source within the virtual power plant management scope, and combine it with environmental factor data to predict the energy regeneration capacity, and build a virtual power plant electricity energy trading framework based on the regeneration capacity and electricity price; Based on batch monitoring management, the load fluctuation trend of virtual power plant management is predicted in batches, and the power supply and demand gap in the virtual power plant management process is analyzed according to the energy regeneration amount and load fluctuation trend; Based on the virtual power plant energy trading framework, the power supply and demand gap and the power balance relationship between virtual power plants are evaluated. Based on the power balance relationship, the gap is determined and the scheduling mechanism of virtual power plant energy trading is analyzed. Air conditioning load data is used as a demand response resource, and the energy scheduling results are optimized to generate air conditioning control instructions. The air conditioning control instructions are recorded and decomposed, and the control status of the air conditioner is output.

2. The virtual power plant energy management method based on air conditioning load according to claim 1 is characterized in that: The process of obtaining the historical power generation of each energy source within the management scope of the virtual power plant, predicting the renewable energy amount of energy in combination with environmental factor data, and constructing a virtual power plant electric energy trading framework based on the renewable energy amount and electricity price includes: Extract the historical power generation of each energy source within the virtual power plant management scope based on the grid monitoring equipment, and obtain the environmental factor data within the virtual power plant management scope according to the demand during the forecast period; Define forecast consecutive days and forecast consecutive weeks based on historical power generation, analyze the number of power generation change periods and power generation data within the forecast consecutive days, and calculate the daily power generation change factor and daily average change factor; Analyze and predict the number of power generation change periods and power generation data within consecutive weeks, calculate the weekly power generation change factor and the weekly average change factor, and combine the calculation results with environmental factor data to predict the amount of energy regeneration; Based on the prediction results, the virtual power plant electricity energy trading rules are designed, the virtual power plant electricity energy trading framework is constructed, and the energy cost within the virtual power plant management scope under the predicted regeneration conditions is analyzed.

3. The method for energy management of a virtual power plant based on air conditioning load according to claim 2, characterized in that: The method of combining the calculation results with environmental factor data to predict the amount of energy regeneration includes: Determine the simulated power generation during the forecast period based on the daily average change factor and the weekly average change factor, and determine the factors influencing the energy power generation results based on the environmental factor data; Based on the influencing factors, the ratio of the daily average change factor to the weekly average change factor in the simulated power generation is considered respectively, and the sum of square errors of the simulated power generation is analyzed according to the ratio results; The comprehensive weight factor is determined by minimizing the sum of squared errors, and the selection results of the forecast consecutive days and forecast consecutive weeks are determined based on the comprehensive weight factor, and the forecast regeneration amount of energy is obtained according to the selection results; Obtain day-ahead electricity price data, and formulate quotation rules for virtual power plants based on the predicted energy regeneration volume. Select trading markets based on the quotation results to provide a basis for building a virtual power plant electricity energy trading framework.

4. The virtual power plant energy management method based on air conditioning load according to claim 3 is characterized in that: The calculation formula for the predicted regeneration amount of the energy is: Where, E a,b represents the predicted energy regeneration amount in time period b during the a-th acquisition process, ω represents the comprehensive weight factor, R represents the total number of predicted consecutive days, G r,b represents the power generation in time period b during the rth forecast consecutive day, represents the average daily power generation of the rth forecast consecutive day, D represents the total number of forecast consecutive weeks, and D d,b represents the power generation in the b time period of the d-th consecutive week, represents the weekly average power generation of the dth forecast consecutive week, Indicates the average power generation within the management scope of the virtual power plant.

5. The method for energy management of a virtual power plant based on air conditioning load according to claim 4, characterized in that: The batch monitoring management set-based batch prediction of load fluctuation trends in virtual power plant management, and the analysis of the power supply and demand gap in the virtual power plant management process based on the energy regeneration amount and load fluctuation trends include: Based on the batch monitoring management set, the load simulation targets and load evaluation indicators corresponding to the industry classification, regional classification and user classification are determined respectively, and the random sampling distribution and parameters are set for random sampling; Based on random sampling results and statistical simulation technology, a difference matrix is ​​constructed to generate random operation information samples, and a trend prediction model is constructed by simplifying the random operation information samples using attribute theory based on information theory; Based on the trend prediction model, the load fluctuation trends corresponding to industry classification, regional classification and user classification are predicted respectively, and the electricity demand within the virtual grid management range is analyzed according to the load fluctuation trends; The difference between electricity demand and energy regeneration forecast results is compared, and the energy profit and loss status in the virtual power plant management process is determined based on the comparison results. The electricity supply and demand difference is analyzed based on the energy profit and loss status.

6. The method for energy management of a virtual power plant based on air conditioning load according to claim 5, characterized in that: The method of constructing a difference matrix based on random sampling results and statistical simulation technology to generate random operation information samples, and constructing a trend prediction model based on the attribute theory of information theory to simplify the random operation information samples includes: The reliability of random sampling results is judged based on load evaluation index simulation, and the random sampling results are optimized and adjusted according to the reliability results until the reliability meets the load simulation target to obtain random operation information samples; Map random operation information samples to the attribute space to construct a discernibility matrix, traverse and search the data in the discernibility matrix until an empty set is encountered, and then stop searching. The reduced triangular matrix is ​​obtained based on the search results. Analyze the original attribute assignments of random operation information samples in the reduced triangular matrix and verify them with the original attribute assignments corresponding to the discernibility matrix. Based on the verification results, determine the reduction accuracy of the reduced triangular matrix. The random operation information samples corresponding to the simplified triangular matrix that meets the reduction requirements are used as the data set, and the data set is used as input, the load fluctuation is used as output and the Monte Carlo algorithm is combined to build a trend prediction model.

7. The method for energy management of a virtual power plant based on air conditioning load according to claim 1, characterized in that: The air conditioning load data is used as a demand response resource, the energy scheduling result is optimized to generate air conditioning control instructions, and the air conditioning control instructions are recorded and decomposed, and the control status of the air conditioner is output, including: Obtain the power load data of various air conditioners within the virtual power plant management scope as demand response resources, determine the demand response amount of the air conditioner power load data within the forecast period, and adjust the energy scheduling results based on the demand response amount; Determine the control strategy for each type of air conditioner based on the adjusted energy scheduling results, determine the air conditioner control instructions, and send the air conditioner control instructions to the device gateway layer through the message transmission protocol; The device gateway layer receives and parses the air conditioning control instructions, obtains the power information required for each type of air conditioner to adjust the control status of the air conditioner, and evaluates the execution status of the air conditioners involved in the control.

8. A virtual power plant energy management system based on air conditioning load, used to implement the virtual power plant energy management method based on air conditioning load according to any one of claims 1 to 7, characterized in that: The management system includes: The monitoring and management set generation module is used to obtain electricity consumption data within the management scope of the virtual power plant, integrate operation information data, and analyze the joint probability of the naive Bayes model based on the hyperparameters of the labels and values ​​involved in the operation information data, optimize the naive Bayes model based on the analysis results, and judge the correlation between labels and the correlation between values; analyze the label and value weights based on the correlation between labels and the correlation between values, output the main characteristics of the operation information data according to the weight results, and determine the classification category labels of the operation information data in combination with the operation rules, the classification category labels include industry classification, regional classification and user classification; analyze the joint distribution correlation between industry classification, regional classification and user classification and the main characteristics, and cluster the correlation, and train the naive Bayes model based on the clustering results to obtain a potential allocation model; generate classification models with industry classification, regional classification and user classification as themes based on the training results, and use the classification models to perform statistics on the operation information data according to industry, region and user to obtain a batch monitoring management set; The energy scheduling result output module is used to obtain the historical power generation of each energy source within the management scope of the virtual power plant, and predict the energy regeneration amount in combination with environmental factor data, and build a virtual power plant electricity energy trading framework based on the regeneration amount and electricity price; based on the batch monitoring management set, it predicts the load fluctuation trend of the virtual power plant management in batches, and analyzes the power supply and demand difference in the virtual power plant management process based on the energy regeneration amount and load fluctuation trend; based on the virtual power plant electricity energy trading framework, it evaluates the power supply and demand difference and the power balance relationship between virtual power plants, and determines the difference based on the power balance relationship to analyze the scheduling mechanism of virtual power plant electricity energy trading; The air conditioning control status output module is used to use the air conditioning load data as a demand response resource, optimize the energy scheduling results to generate air conditioning control instructions, record and decompose the air conditioning control instructions, and output the control status of the air conditioning.

Images