Micro-grid electrical load prediction method and system

Abstract

The invention discloses a microgrid electrical load prediction method and system, and the method comprises the steps: building a power grid load prediction model based on an energy consumption data classification record, and generating a real-time load prediction result; based on the real-time load prediction result, identifying a power use mode of the user, and generating a user behavior analysis result; optimizing the power grid load prediction model based on the user behavior analysis result, and generating an energy influence evaluation result; generating a node stability evaluation result by considering node positions and connectivity through an energy influence evaluation result; and based on a node stability evaluation result, adjusting abnormal load early warning parameters by considering node stability, predicting a power grid abnormal load in combination with the optimized power grid load prediction model, and generating power grid load prediction data. The monitoring capability and response speed of the power system are improved by analyzing the user power use data in the micro-grid in real time, and the accuracy of the prediction model is enhanced by combining classified processing of the user power data and analysis of the use mode.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to a method and system for predicting power load in a microgrid. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] The field of smart grid technology uses information technology, communication technology and grid automation technology to improve the reliability, efficiency and security of the power system, enable the grid to achieve two-way communication and energy flow, monitor the grid status in real time and respond quickly to changes in power demand and supply, including various applications of renewable energy integration and microgrid management, optimize energy utilization, reduce waste, and enhance the system's adaptability to various operating conditions and external interference. Through smart meters and energy management systems and other tools, it effectively manages user-end energy demand and enhances energy efficiency and economic benefits.

[0004] Among them, the microgrid electricity load forecasting method focuses on using data analysis technology in a microgrid environment to predict short-term and long-term electricity demand. It aims to optimize the allocation of power resources by accurately predicting power demand, improve energy utilization efficiency, and ensure the reliability of power supply. It helps microgrid operators effectively manage energy output, manage power generation and storage equipment by predicting load, responding to consumption peaks and troughs, and supporting the effective use of renewable energy, thereby reducing energy costs and improving the stability of grid operation, optimizing power supply and demand, and reducing carbon emissions.

[0005] Traditional microgrid electricity load forecasting methods have limitations in processing real-time data collection and load forecasting. They cannot fully consider the impact of various user behaviors and meteorological conditions, resulting in inaccurate load forecasting and insufficient response capabilities. It is difficult to respond to power peaks in a timely manner, affecting the reliability of the power grid and the effective use of energy. Insufficient consideration is given to the output changes and node stability of distributed resources, resulting in unstable grid operation when sudden load changes occur. In terms of stability assessment of key nodes in the power grid, it cannot effectively identify and respond to node failures, resulting in chain reactions, reducing the stability of power grid operation, and causing power outages and energy waste under high load and extreme weather conditions. Summary of the Invention

[0006] To overcome the deficiencies of the above-mentioned prior art, the present invention provides a method and system for predicting electricity load in a microgrid, which fully considers the influence of user electricity consumption behavior and external environmental factors when predicting load. On the basis of the constructed load prediction model, the stability and responsiveness of microgrid nodes are also considered, thereby enhancing the prediction of risk loads in the power grid.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions: In a first aspect, the present invention provides a method for predicting power load in a microgrid, comprising: Collect power usage data of users in the microgrid in real time, classify the power usage data according to user type, usage time, and meteorological conditions, and generate classified records of energy consumption data; Based on the classified records of the energy consumption data, a power grid load forecasting model is constructed by calculating the correlation and weight of multiple external factors to generate real-time load forecast results; Based on the real-time load forecast results, analyze the user's power usage data, identify the power usage patterns of multiple types of users and evaluate their impact on the microgrid load, and generate user behavior analysis results; Based on the user behavior analysis results, analyze the impact of output changes and charging and discharging behaviors of multiple distributed energy sources in the microgrid on the load, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate energy impact assessment results; By using the energy impact assessment results, considering the node locations and connectivity, identifying the key nodes in the microgrid and evaluating their stability and response capabilities, and generating a node stability assessment result; Based on the node stability evaluation result, the abnormal load warning parameters are adjusted taking into account the node stability, and the abnormal load of the power grid is predicted in combination with the optimized power grid load prediction model to generate power grid load prediction data.

[0008] A further technical solution is to generate classified records of energy consumption data, specifically including: Collect power usage data of microgrid users in real time, including power consumption, time records and meteorological information, and generate power data collection results; Based on the power data collection results, the data is classified according to user type, usage time, and meteorological conditions to generate classified power data; Based on classified power data, pattern recognition is performed on various power usage data, users' power usage behaviors and changing trends in various situations are analyzed, and classified records of energy consumption data are generated.

[0009] Further technical solutions to generate real-time load forecast results include: Based on the classified records of energy consumption data, the correlation between various external factors and load is analyzed, the influence degree of various external factors is evaluated, and the load impact analysis results are generated; Using the load impact analysis results, a logistic regression algorithm is used to construct a power grid load forecasting model and calculate the weights of multiple external factors on the load to generate a power grid load forecasting model; Based on the power grid load forecasting model, the power load change trend under multiple time periods and differentiated meteorological conditions is predicted to generate real-time load forecast results.

[0010] In a further technical solution, the logistic regression algorithm calculates the power load state according to the formula, which is expressed as:

[0011] in, Indicates the power load status, represents the base of natural logarithms, is the intercept term, represents the weight of temperature, represents the weight of humidity, represents the weight of the electricity market price, represents the weight of sunshine duration, represents the weight of a special event, represents the weight of the power grid maintenance status, Indicates temperature, Indicates humidity, represents the electricity market price, Indicates the duration of sunshine. Indicates special events, Indicates the power grid maintenance status.

[0012] Further technical solutions to generate user behavior analysis results include: Based on real-time load forecast results, analyze the power usage data of multiple user categories in the microgrid, identify the power usage patterns of multiple user categories in different time periods, and generate power usage pattern data; Analyzing and identifying power usage characteristics of multiple user categories based on the power usage pattern data to generate user power usage characteristics; Based on the user power usage characteristics, the impact of multiple user categories on the microgrid load in multiple time periods is evaluated, the load contribution ratios of the multiple user categories are calculated, and user behavior analysis results are generated.

[0013] Further technical solutions are used to obtain an optimized grid load forecasting model and generate energy impact assessment results, including: Based on the results of user behavior analysis, the output data of distributed energy resources in the microgrid is collected and analyzed to generate energy output data; Based on the energy output data, analyzing the impact of charging and discharging behavior of the power storage device on the grid load in multiple time periods to generate storage behavior analysis results; Based on the storage behavior analysis results, the load forecasting model is adjusted to match the impact of seasonal changes, weather conditions, and time changes, and an energy impact assessment result is generated.

[0014] Further technical solutions to generate grid load forecast data include: Based on the node stability assessment results, analyze the load data of multiple nodes in multiple time periods, identify the load carrying capacity of multiple nodes, and generate node load capacity data; Based on the node load capacity data, and according to the stability and load carrying capacity of the nodes, adjusting abnormal load warning threshold parameters of multiple nodes to generate load warning parameters; Based on the load warning parameters and in combination with meteorological conditions, the optimized power grid load forecasting model is used to predict risk load nodes at multiple time points and generate power grid load forecasting data.

[0015] In a second aspect, the present invention provides a microgrid power load forecasting system, comprising: The electricity consumption data classification module is configured to: collect electricity consumption data of users in the microgrid in real time, classify the electricity consumption data according to user type, usage time, and meteorological conditions, and generate energy consumption data classification records; An external factor analysis module is configured to: construct a power grid load forecasting model based on the classified records of the energy consumption data by calculating the correlation and weight of multiple external factors, and generate real-time load forecast results; A user group evaluation module is configured to: analyze user power usage data based on the real-time load forecast results, identify power usage patterns of multiple types of users, evaluate their impact on the microgrid load, and generate user behavior analysis results; a model parameter optimization module configured to: analyze the output changes and charging and discharging behaviors of multiple distributed energy resources in the microgrid on the load based on the user behavior analysis results, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate an energy impact assessment result; a node stability assessment module configured to: identify key nodes in the microgrid and assess their stability and response capabilities based on the energy impact assessment results, taking into account node locations and connectivity, and generate a node stability assessment result; The node influence assessment module is configured to: based on the node stability assessment result, adjust the abnormal load warning parameters considering the node stability, predict the abnormal load of the power grid in combination with the optimized power grid load prediction model, and generate power grid load prediction data.

[0016] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a microgrid power load forecasting method as described in the first aspect.

[0017] In a fourth aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of a microgrid power load forecasting method as described in the first aspect are implemented.

[0018] One or more of the above technical solutions have the following beneficial effects: The present invention improves the monitoring capability and response speed of the power system by performing real-time analysis on the power usage data of users in the microgrid. It also combines the classification and processing of users' power data with the analysis of usage patterns to enhance the accuracy of the prediction model, analyzes user behavior and the output changes of distributed energy resources, effectively predicts and manages power loads, and utilizes the assessment of the stability and response capability of key nodes in the microgrid to enhance the grid's ability to predict and identify risky loads, reduce economic losses caused by grid instability, and achieve accurate prediction of user power loads in the microgrid environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] 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.

[0020] Figure 1 This is a flow chart of a method for predicting power load in a microgrid according to an embodiment of the present invention; Figure 2 is a flow chart of generating energy consumption data classification records according to an embodiment of the present invention; Figure 3 is a flow chart of generating real-time load forecast results according to an embodiment of the present invention; Figure 4 is a flow chart of generating user behavior analysis results according to an embodiment of the present invention; Figure 5 is a flow chart of generating energy impact assessment results according to an embodiment of the present invention; Figure 6 is a flow chart of generating node stability evaluation results according to an embodiment of the present invention; Figure 7 This is a flow chart of generating power grid load forecast data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] 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 herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0022] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0023] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0024] Example 1 like Figure 1 As shown, this embodiment discloses a method for predicting power load of a microgrid, which includes the following steps: S1: Collect power usage data of users in the microgrid in real time, classify the power usage data according to user type, usage time, and meteorological conditions, and generate energy consumption data classification records; In this embodiment, real-time electricity usage data from users in the microgrid is collected and categorized based on user type, usage time, and weather conditions to generate categorized energy consumption data records. These records include power usage patterns for various user categories, time series analysis results, and records of weather conditions.

[0025] S101: Collect power usage data of microgrid users in real time, including power consumption, time records, and meteorological information, and generate power data collection results. Set the automatic collection frequency, match the real-time data stream, use data integration algorithms to pre-process the collected information, identify and remove abnormal data points, correlate and analyze time records and meteorological information, ensure the integrity and accuracy of the data, classify it, and index it according to power consumption, time records, and meteorological conditions. The formula is expressed as follows:

[0026] in, Indicates the storage capacity, Indicates power consumption, represents the weighted value of meteorological information, Indicates the base storage capacity, Indicates the influence coefficient of electricity consumption, The system optimizes the data storage structure and estimates future data growth, generating power data collection results, namely power usage data.

[0027] When training a model based on historical data, multivariate regression analysis or logistic regression is used. By fitting the target variable (such as load status), the coefficients (such as the regression coefficients of temperature, humidity, electricity price, etc.) are obtained by reverse calculation and the regression coefficients are used as the influencing coefficients.

[0028] S102: Classify electricity usage data based on user type, usage time, and weather conditions to generate classified electricity data. Deeply analyze electricity usage data, grouping it by user type and time series, with weather conditions as a secondary grouping factor. Classification algorithms (such as decision trees, K-means clustering, support vector machines, and random forests) based on statistical principles perform pattern analysis, categorize the data, and generate classified electricity data.

[0029] S103: Based on the classified electricity data, perform pattern recognition on various electricity usage data, analyze user electricity usage behaviors and trends under various circumstances, and generate classified records of energy consumption data. Use multivariate analysis methods (such as principal component analysis, factor analysis, multiple linear regression, and cluster analysis) to analyze electricity usage patterns of differentiated user categories under various meteorological conditions, identify key trends and behavioral patterns, and perform complex trend analysis based on the classified electricity data. Use pattern recognition algorithms (such as hidden Markov models, KNN, neural networks, and DBSCAN) to determine seasonality and random fluctuations in electricity consumption.

[0030] Key trends are patterns of regularity, cyclicality, or trends identified from historical electricity usage data across different user categories, over different time periods, and under different meteorological conditions. Examples include peak electricity consumption during the summer midday hours, higher weekend electricity consumption for residential users compared to weekdays, and increased load for commercial users before specific holidays. These trends are typically statistically significant, reflecting the overall response of user behavior to temporal or environmental variables. They are often identified through time series analysis and seasonal decomposition models.

[0031] Behavioral patterns refer to the classification and identification of patterns in the electricity usage behavior of user groups. Examples include users with high-frequency, short-term electricity use (such as temporary processing equipment), users with consistently low loads (such as small offices), and users whose loads vary significantly with temperature (such as air-conditioning-dominated venues). These patterns focus on analyzing behavioral patterns of individuals or groups and can be extracted through pattern recognition algorithms such as cluster analysis and association rule mining.

[0032] Energy consumption data classification records are generated based on the pattern recognition results. The energy consumption data classification records include behavior patterns of multiple user categories, time series analysis results, and meteorological conditions impact records.

[0033] S2: Based on the classified records of the energy consumption data, a power grid load forecasting model is constructed by calculating the correlation and weight of multiple external factors to generate real-time load forecast results; In this embodiment, based on categorized energy consumption data, the correlation and weighting of various external factors affecting power load are calculated to predict load trends, build a grid load forecasting model, and generate real-time load forecast results. The real-time load forecast results include load trends, seasonal and weather forecast data, and time-dependent load fluctuation charts.

[0034] S201: Based on the classified records of energy consumption data, analyze the correlation between various external factors and power load, assess the impact of various external factors, and generate load impact analysis results. Based on the classified records of energy consumption data, quantitatively analyze the relationship between each external factor and power load. Use multivariate analysis methods to statistically evaluate the impact of external factors on power load, including climate conditions and changes in user behavior. The impact of external factors is assessed and the load impact analysis results are generated.

[0035] External factors include meteorological conditions (such as temperature, humidity, and sunshine), user behavior (such as electricity usage habits and load response), market prices, holidays, and equipment status.

[0036] S202: Using the results of the load impact analysis, a logistic regression algorithm is used to construct a power grid load forecasting model and calculate the weights of various external factors on the load, thereby generating a power grid load forecasting model. Using the results of the load impact analysis, a logistic regression algorithm is used to construct a power grid load forecasting model and calculate the weights of various external factors on the power load. The model matches the actual impact of various external conditions, analyzes the impact of various external factors on the power load, assigns weights to the factors, constructs a logistic regression model, optimizes parameter settings to improve prediction accuracy, and generates a power grid load forecasting model.

[0037] Using external factors as independent variables and grid load status as the dependent variable, a logistic regression model is trained to produce a mathematical expression capable of outputting prediction probabilities, known as the grid load forecasting model. Based on the load impact analysis results, a preliminary weighting guide is provided for each external variable (i.e., external factor) for model initialization and feature selection. The weight of each external factor is used to measure its impact on load status prediction. All weights are automatically learned by the model during training.

[0038] Logistic regression algorithm calculates the power load status according to the formula, which is expressed as:

[0039] in, Indicates the power load state, used to indicate the possibility of the power grid reaching various load states; Represents the base of the natural logarithm, used for exponential transformation in logistic regression; is the intercept term, which is used to adjust the baseline probability that is independent of the input variables; The weight of temperature is used to measure the impact of ambient temperature changes on the probability of power load status, indicating how the probability of power load status changes when the temperature changes by one unit; Represents the weight of humidity, which is used to measure the impact of ambient humidity on power load, indicating the change in power load state probability when humidity increases by one unit; The weight of the electricity market price is used to express the impact of market price changes on electricity usage behavior and load state probability, reflecting the potential impact of price changes on electricity load; The weight of sunshine duration is used to measure the impact of sunshine time on power load, indicating how the increase or decrease of sunshine duration affects power demand and its load status; The weight of special events is used to represent the impact of holidays or other large-scale events on power load, and to illustrate the influence of these events on the probability change of load status; The weight representing the maintenance status of the power grid is used to measure the impact of maintenance activities on the power load capacity and to show how the maintenance status regulates the load pressure of the power system; Indicates temperature, used to indicate the impact of ambient temperature on power load; Indicates humidity, used to indicate the impact of ambient humidity on power load; Represents the electricity market price, used to indicate the impact of price changes on electricity usage behavior; Indicates the duration of sunshine, used to indicate the impact of solar power generation and load demand; Indicates special events, including holidays and major events, and is used to indicate changes in power load caused by special events; Indicates the maintenance status of the power grid and is used to indicate the impact of maintenance activities on power load capacity.

[0040] Obtain parameters by collecting relevant meteorological data (temperature and humidity data), power market data, sunshine data, and power grid maintenance schedule 、 、 、 、 、 According to the relationship between historical load data and these variables, the weight coefficients of multiple parameters are obtained through logistic regression analysis. 、 、 、 、 、 , substitute the parameters into the formula and calculate the probability , which is used to predict the possibility of the power system entering a high-load or low-load state in a certain period of time in the future.

[0041] Load states refer to a number of possible load levels that may occur in a microgrid under different operating conditions. These include: low load (system operating load is low, with the risk of idle resources or energy waste); medium load (load is within the normal fluctuation range of the design capacity, with stable system operation); high load (load approaches or exceeds the system design limit, with the risk of power congestion or overload); and abnormal load (such as atypical load fluctuations caused by emergencies, extreme weather, or equipment failures). These states can be classified based on thresholds set based on statistical analysis of historical operating data, using indicators such as load level proportion, voltage variation range, and equipment utilization rate to facilitate classification and risk management of model output results.

[0042] To determine the load state under the current input conditions, a logistic regression model is used. This model uses external factors such as temperature, humidity, electricity market prices, and sunshine duration as input variables, and outputs the probability of a specific load state occurring. For a binary classification model, a probability threshold (e.g., p ≥ 0.7 for a "high load state") can be set for discrimination. For a multi-classification model, probabilities are output for each state, with the final prediction being the state with the highest probability. This probability is not only used for classification but also serves as a quantitative basis for the "load risk level" in subsequent early warning mechanisms, enhancing the system's ability to perceive and respond to dynamic load states.

[0043] Through the above technical solution, the grid load forecasting model finally generated introduces a variety of external variables and their mapping relationships evaluated by S201 on the basis of the original input features, and optimizes the various parameters in the model through the training process, so that the model has stronger prediction and generalization capabilities.

[0044] S203: Based on the power grid load forecasting model, power load trends are predicted for multiple time periods and under differentiated meteorological conditions, generating real-time load forecast results. These real-time load forecast results include load trends, seasonal and weather forecast data, and time-dependent load fluctuation charts. The power grid load forecasting model uses real-time inputs such as current and predicted meteorological conditions and user behavior data to perform time series analysis, predicting short-term and long-term load changes. Model parameters are regularly updated to match real-time data trends, generating real-time load forecast results.

[0045] During the actual forecasting process, meteorological forecast data for future periods is required. This is because power load is highly sensitive to climatic conditions. For example, high temperatures can significantly increase air conditioning power consumption, affecting the overall load state. Therefore, these predictive variables must be introduced in advance to improve the model's foresight and accuracy. Furthermore, the user behavior data used in the grid load forecasting model comes from the "user behavior analysis results" generated in S3, rather than the load impact analysis results in S201. User behavior data reflects the electricity consumption characteristics of different user groups in different time periods and their contribution to the system load, and is an important component of the model's input variables.

[0046] S3: Based on the real-time load forecast results, analyze the user's power usage data, identify the power usage patterns of multiple types of users and evaluate their impact on the microgrid load, and generate user behavior analysis results; In this embodiment, real-time load forecasting results are used to analyze user electricity usage data, identify various user electricity usage patterns, and assess their impact on the microgrid load, generating user behavior analysis results. These user behavior analysis results include user group electricity demand characteristics, peak and off-peak electricity usage period identification records, and electricity usage pattern comparison results.

[0047] Although the "electricity usage pattern" and the "electricity usage pattern of various user categories in the energy consumption data classification record" in this step are related, their meanings are not exactly the same. The "electricity usage pattern of various user categories in the energy consumption data classification record" refers to the original electricity usage behavior characteristics initially obtained after classifying and pattern identifying the electricity usage data of different users at different times and meteorological conditions in step S1. It focuses on the preliminary division of data and static feature extraction. The "electricity usage pattern" is based on the aforementioned classification data, combined with the real-time load forecast results, to further dynamically identify and summarize the electricity usage behavior of each user category in differentiated time periods, and generate a more representative and predictive behavior pattern. Therefore, the electricity usage pattern is a deep behavioral abstraction and evolution result based on the classification record, and there is a progressive and processing relationship between the two.

[0048] S301: Based on real-time load forecast results, analyze the power usage data of multiple user categories within the microgrid, identify the power usage patterns (power consumption behaviors) of these user categories during different time periods, and generate power usage pattern data. Collect power usage records for each user category during different time periods. Use data clustering to group users based on their power usage habits and time preferences. Perform time series analysis on each group's power usage data to identify typical power usage patterns.

[0049] User categories are categorized based on their electricity usage characteristics, usage, and load attributes within the microgrid, and are used for classification analysis and behavioral modeling of energy consumption data. These categories include, but are not limited to, residential users (such as homes and apartments, which experience peak morning and evening electricity consumption), commercial users (such as shops and office buildings, which experience concentrated daytime electricity consumption), industrial users (such as factories and manufacturing workshops, which experience heavy and sustained loads), public service users (such as schools, hospitals, and government agencies, which experience distinct weekday load patterns), and special users (such as data centers and charging stations, which have specific uses and operating modes). This classification approach helps identify differences in electricity usage among different users at different times and under different environmental conditions, providing refined input for subsequent load forecasting models.

[0050] S302: Based on the power consumption pattern data, consider user categories, analyze and identify the power consumption characteristics of multiple user categories, and generate user power usage characteristics. Combining user behavior and statistical methods, analyze the power consumption data of each group, extract the key features that affect the microgrid load, and use factor analysis technology to simplify the complex data set into a few representative features. The formula is expressed as:

[0051] in, Indicates the user's power usage characteristics, represents a constant term, Indicates the The weight of the feature, Indicates user groups No. Electricity usage characteristics, Represents the total number of features analyzed.

[0052] Using raw electricity consumption data as input and electricity pattern data as a reference category or division basis, the system extracts key internal variables within each user group. Based on the electricity pattern data, factor analysis and other methods are used to further extract quantifiable and discriminative statistical characteristic indicators from the multi-dimensional data, such as average load, peak-to-valley ratio, and frequency of change. This provides parameter variables that can be directly input for subsequent feature weighting and modeling.

[0053] S303: Based on user power usage characteristics, evaluate the impact of multiple user groups on the microgrid load over multiple time periods, calculate the load contribution ratios of multiple user groups, and generate user behavior analysis results. Use multiple regression analysis to estimate the contribution of each user group to the system load over different time periods.

[0054] While the above technical solution uses electricity usage data as the primary data source, this step is not simply repetitive data processing. Instead, it further analyzes user electricity usage behavior based on the real-time load forecast results generated in S2. Specifically, S2 provides load forecast trends at different time points and environmental conditions. S3 uses these trends to reverse-map and identify the specific electricity response patterns of each user under different load conditions, thereby generating power usage patterns and user electricity usage characteristics. Therefore, S3 not only utilizes the original data but also integrates the forecast results, achieving a transition from "forecasting total load" to "understanding the contribution of user behavior," forming a complete prediction-explanation-feedback closed loop. Thus, S1 provides static classification data, S2 performs dynamic load forecasting based on the forecast model, and S3, using the forecast results as a reference, further analyzes the specific impact of user behavior in different load ranges, thereby more accurately identifying the behavioral patterns and load contributions of each user category. While S3 still uses some of the original data from S1, its processing objectives, analysis depth, and data dimensionality have been enhanced. This represents data reuse rather than process duplication, and the three processes form a clear logical and progressive relationship.

[0055] S4: Based on the user behavior analysis results, analyze the impact of output changes and charging and discharging behaviors of multiple distributed energy sources in the microgrid on the load, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate energy impact assessment results; In this embodiment, based on the results of user behavior analysis, the impact of the output changes and charging and discharging behaviors of multiple distributed energy resources in the microgrid on the grid load is analyzed, the load forecasting model is optimized, and the energy impact assessment results are generated. The energy impact assessment results include the output stability of distributed energy resources, storage device charging and discharging cycle data, and seasonal output forecasting models. The user behavior analysis results provide the power consumption patterns and load contribution ratios of each user category in different time periods, providing a basis for determining the temporal structure and fluctuation trend of the overall power demand of the microgrid. The energy output and storage behavior analysis is based on the temporal changes in user power demand and is used to further infer the dispatch response of each distributed energy source in the microgrid and the charging and discharging rhythm of the energy storage system.

[0056] S401: Based on the results of user behavior analysis, collect and analyze the output data of distributed energy resources in the microgrid, including solar energy and wind energy, and record performance data under various weather and seasonal conditions to generate energy output data. The user behavior analysis results determine the demand curve for energy output and energy storage scheduling. Collect various energy outputs, including the power output of photovoltaic panels and wind turbines, and use data acquisition systems for real-time monitoring under different meteorological conditions. By applying data analysis tools, evaluate the efficiency and output of each energy resource under different conditions. The formula is expressed as:

[0057] in, Represents energy output data, Indicates the The output coefficient of the energy under target conditions, Indicates time No. The measured output of the energy source, is the total number of energy types.

[0058] The results of user behavior analysis provide electricity consumption behavior trends and load contribution forecasts for different user groups in different time periods, enabling the system to predict which periods will have load peaks or troughs. Therefore, when analyzing energy output data, the system can use demand-side trends as a reference to evaluate the adaptability of supply-side resources. For example, when the user load forecast shows that there will be a large amount of electricity demand in a certain time period, the system needs to focus on analyzing whether the renewable energy output during this time period can meet the corresponding load. Therefore, although the energy output data itself is generated through equipment collection and is not directly derived from the user behavior analysis results, the two have the same application goals, and the evaluation process of the former is based on the latter as a reference. The two constitute a data closed loop on both the supply and demand sides.

[0059] In an actual microgrid environment, the factors that affect energy output are complex and changeable. There are dynamic interference factors such as weather fluctuations, equipment aging, shading, power limitations, and system scheduling strategies, which often cause deviations between the calculated results of ideal efficiency and output and the actual output. Therefore, the present invention introduces the "analysis and evaluation" step of efficiency and output. Its purpose is to dynamically identify the actual output capacity of the current equipment under a specific environment through statistical modeling of historical operating data and comparative analysis of actual performance, to ensure that the results of the supply and demand matching analysis are closer to the actual operating status, thereby improving the accuracy of the prediction. For example, when predicting the peak load of users, it is necessary to determine whether the solar energy or wind energy in the corresponding period can still maintain its rated efficiency, and this judgment needs to rely on the evaluation of the actual performance of the equipment under similar conditions.

[0060] S402: Based on the energy output data, analyze the impact of the charging and discharging behavior of the power storage device on the grid load over multiple time periods and generate storage behavior analysis results. Record the charging and discharging cycles of each storage unit, and use statistical analysis methods to evaluate the contribution and impact to the overall grid load. Use multivariate regression analysis to quantify the relationship between storage behavior and grid load. The formula is expressed as:

[0061] in, Indicates the results of storage behavior analysis. represents the baseline influence, Indicates the The charge and discharge behavior of each storage device affects the weight. Indicates the storage devices at time The charge and discharge capacity, The total number of storage devices.

[0062] There is a close and significant relationship between microgrid load forecasting and distributed energy generation output. Microgrid systems inherently possess strong local self-sufficiency, and their supply and demand balance relies heavily on the output capacity of distributed energy resources (such as photovoltaics and wind power). Therefore, when forecasting future power load, it is also necessary to comprehensively consider supply fluctuations on the generation side. This impact is particularly pronounced in scenarios where renewable energy accounts for a high proportion and energy storage capacity is limited. This invention introduces an evaluation mechanism for energy output and storage behavior. The goal is to analyze the generation side's ability to meet the expected load based on previous user behavior and load trend forecasts. In other words, power generation output is not only a key resource for responding to forecasted loads but also, in turn, influences the system's load regulation capabilities (e.g., peak shaving and valley shifting, and load pressure relief). Therefore, the two factors form a closed-loop model of forecast-driven and supply-feedback.

[0063] S403: Based on the results of storage behavior analysis, adjust the load forecasting model to match the influence of seasonal changes, weather conditions, and time changes, and generate energy impact assessment results. Based on the results of storage behavior analysis, dynamically adjust the parameter level of the load forecasting model to reflect the comprehensive influence of external conditions such as seasonal changes, weather factors, and time periodic fluctuations during the operation of the microgrid. Specifically, first, based on the charging and discharging behavior of various energy storage devices in different time periods, identify their ability to regulate the overall grid load, and form an impact mapping relationship for different characteristic parameters, including temperature changes, battery state of charge, and discharge frequency. Then, using these influencing factors as input, perform weighted corrections on the key parameters in the original load forecasting model (such as meteorological related factors and time weights) to ensure that the prediction curve is consistent with the actual operation. The environment remains consistent, and the adjustment process does not change the model structure. Instead, by introducing correction parameters representing seasonal changes, weather conditions and time factors into the prediction function, its pulling or suppressing effect on the load trend is dynamically calculated. For example, if the analysis shows that the proportion of energy storage discharge increases during winter due to short daylight hours and concentrated electricity consumption at night, the model will automatically increase the load forecast base value for this period to reflect the greater power supply pressure. Ultimately, the adjusted model output reflects the predicted value of electricity demand under current or future conditions, which serves as the core basis for generating energy impact assessment results and provides accurate input for subsequent node stability assessment and abnormal warning.

[0064] S5: Based on the energy impact assessment results, considering the node location and connectivity, identifying the key nodes in the microgrid and evaluating their stability and response capabilities, and generating a node stability assessment result; In this embodiment, the energy impact assessment results are used to identify key nodes in the microgrid, consider node location and connectivity, and evaluate their stability and responsiveness to generate node stability assessment results. These node stability assessment results include the load management capabilities of key nodes, node influence scores, and network connection stability analysis results.

[0065] S501: Based on the energy impact assessment results, analyze and identify the location and connection data of multiple nodes in the microgrid, evaluate the connection strength, record the energy output information of multiple nodes, and generate node distribution location information. Collect and analyze the geographic location, connection type, and corresponding energy output of each node. Use network analysis tools to evaluate the connection strength between nodes. Utilize geographic information systems and network topology analysis technology to ensure the accuracy and practicality of the data. The formula is expressed as:

[0066] in, Representation node location and connectivity data, Indicates connection To Node The strength coefficient, Representation node To Node The connection data, Indicates connection to a node The total number of .

[0067] The purpose of this step is to fully understand the structural relationships and electrical connection characteristics between nodes within the microgrid, providing a foundation for subsequent node stability assessment and risk load forecasting. By quantifying the strength of connections between nodes through network analysis tools, it is possible to identify which nodes are core to the network and which nodes have weaker connections or present bottleneck risks. Combining geographic information systems (GIS) and network topology analysis techniques allows for an integrated modeling of the spatial location of nodes and their electrical connections, ensuring data integrity and practical operability at both the spatial layout and electrical structure levels. This analysis is useful for determining the impact of a node on the overall system under specific load fluctuations or fault conditions. It also assists in determining which critical nodes require stricter abnormal load warning thresholds or increased redundancy, thereby improving the robustness and stability of the microgrid under dynamic operating conditions. Therefore, this step is an essential structural foundation for building a risk-responsive predictive system.

[0068] S502: Based on node distribution information, analyze node connectivity and location information, evaluate the mutual influence between multiple nodes, and generate influence assessment data. Use mathematical modeling to quantify the interaction between nodes. By constructing a multi-level influence model, the importance of nodes in the microgrid and their impact on system behavior are revealed.

[0069] The core function of the multi-level influence model is to quantify the influence of a node in the microgrid structure, and to support node stability assessment and load warning parameter setting by weighted summarization of the spatial and structural relationships between it and other nodes.

[0070] S503: Based on the impact assessment data, evaluate the stability and responsiveness of multiple nodes under various load conditions and generate node stability assessment results. Analyze the performance of nodes under different operating conditions, including their adaptability in high-load and complex environments. Use dynamic simulation testing combined with real-time data monitoring to evaluate node performance and stability. The formula is:

[0071] in, Representation node The stability evaluation results of represents the baseline stability, Representation node Under load conditions The response capability influence coefficient under Representation node Under load conditions The response data below, Indicates the total number of nodes that affect node performance.

[0072] Dynamic simulation tests are used in combination with real-time data monitoring to evaluate the performance and stability of nodes. Specifically, a simulation model is first constructed that includes the electrical connection relationship and historical load data of each node, and typical operating scenarios are set, such as sudden load increases, day-night load switching, and load fluctuations caused by extreme weather. The voltage, current, and power response changes of each node are dynamically observed through simulation tools to determine its electrical stability performance under stress conditions. At the same time, the collected real-time operating data is used to compare and verify the simulation results, and nodes with delayed response, large voltage fluctuations, or abnormally frequent alarms are identified. Combined with their influence scores, their stability level and operating risks in the microgrid are comprehensively evaluated to obtain node stability assessment results, providing a basis for subsequent load warnings and network optimization.

[0073] S6: Based on the node stability evaluation result, abnormal load warning parameters are adjusted in consideration of node stability, abnormal load of the power grid is predicted in combination with the optimized power grid load prediction model, and power grid load prediction data is generated.

[0074] In this embodiment, the abnormal load warning parameters are adjusted based on the node stability assessment results, combined with the load forecasting model to predict abnormal grid loads and generate grid load forecast data. The grid load forecast data includes parameters based on the load anomaly interval, warning threshold settings, and load data forecast results.

[0075] S601: Based on the node stability evaluation results, analyze the load data of multiple nodes in multiple time periods, identify the load carrying capacity of multiple nodes, and generate node load capacity data. Through integrated analysis tools, the historical load data of each node is analyzed to identify the performance of the node under high load and low load conditions. Specifically, the historical load data of each node in the past period of time, including indicators such as power consumption, voltage level, and power factor, are first retrieved and sorted according to the time dimension to construct a node-level load time series data set. Then, based on the set load threshold, the data is divided into high-load and low-load segments. The average load, maximum / minimum value, change rate and stability index of the node under different load conditions are calculated using the aggregation statistical analysis method. Under high load conditions, the node is analyzed for abnormal operating behaviors such as frequent overload, instantaneous voltage drop, and current shock. Under low load conditions, it is evaluated whether the node runs in a low load state for a long time, and there is low energy efficiency or resource waste. In addition, the system uses volatility analysis, standard deviation calculation, peak recognition and other technical means to further judge the stability and response capability of the node load in different load segments, and identify the characteristics of nodes that are susceptible to disturbances, response delays or extreme operations. The analysis results are used to construct a load capacity profile for each node, providing quantitative support for the subsequent abnormal load warning parameter setting. The statistical model is used to calculate the maximum and minimum load bearing thresholds of the node and evaluate performance stability. The formula is expressed as:

[0076] in, Representation node Load capacity data, Indicates time period The load factor, Representation node In the period The load data, Indicates the total number of time periods analyzed.

[0077] S602: Based on the node load capacity data, and in accordance with the stability and load carrying capacity of the nodes, the abnormal load warning threshold parameters of the multiple nodes are adjusted to generate load warning parameters.

[0078] By comparing the historical load performance and stability assessment results of the node, the early warning threshold of each node is quantitatively set to warn of load anomalies in advance. The formula is expressed as:

[0079] in, Representation node Load warning parameters, Indicates the basic warning threshold, The adjustment factor for load carrying capacity, Representation node Stability evaluation data of Indicates the total number of evaluation parameters considered.

[0080] This step adjusts the abnormal load warning threshold parameters for each node based on the node's stability assessment results and load carrying capacity data. This step closely relies on and feeds back to the power load forecast. Specifically, the power load forecast model primarily predicts load trends for the entire system over future time periods. However, to achieve more operational predictive responses, the system-level load forecast must be broken down to the node level, and appropriate warning thresholds must be set based on each node's physical carrying capacity and operational stability. This step is crucial for transforming prediction results into actionable strategies. Based on high-risk periods identified in the forecast results, the system appropriately lowers the warning thresholds for highly sensitive nodes to enable early response. For nodes with strong stability and high redundancy, more relaxed response parameters can be set to avoid false alarms. Ultimately, these adjusted warning threshold parameters are fed back into the load forecast model and used in S603 to identify "risky load nodes," thus completing the closed loop of prediction, analysis, and warning. Therefore, this step is the intermediary link connecting the "load forecasting model" and the "risk management mechanism", and its output directly affects the distribution accuracy and practical applicability of the forecast results.

[0081] S603: Based on the load warning parameters and meteorological conditions, the optimized grid load forecasting model is used to predict the risk load nodes at multiple time points and generate grid load forecast data. The load forecasting model is used to integrate meteorological condition data and node warning parameters to perform load risk assessment. The formula is expressed as

[0082] in, Indicates time Power grid load forecast data, represents the forecast benchmark, Indicates the influence coefficient of load warning parameters, represents the influence coefficient of meteorological conditions, Indicates that the node is at time Warning parameters, Indicates time weather conditions.

[0083] The predicted risky load nodes and the overall grid load forecast form a subordinate relationship between local refinement and result implementation. Grid load forecasting primarily provides system-level load trend changes at multiple future time points, such as overall load peaks and fluctuation ranges. However, in actual microgrid operation, changes in the total system load are the result of the combined distributed load behavior of each node. Therefore, node-level risk control is difficult to achieve with only a global forecast. Building on this foundation, S603 refines the total system load forecast into a combination of "critical time points + high-risk nodes" by integrating node warning parameters, meteorological conditions, and forecast model outputs. This identifies which nodes are most likely to trigger load anomalies or become bottlenecks within a specific time period, namely "risky load nodes." This node-level risk assessment not only enhances the practicality of the forecast but also provides precise guidance for subsequent operational and maintenance strategies such as load scheduling, load shedding, or distributed energy deployment. Therefore, the prediction of risky load nodes is an extended application of grid load forecasting and a means of risk identification. It embodies a closed-loop model from macro-trend analysis to micro-risk management and control, and is a key manifestation of the refined management capabilities of the load forecasting system of the present invention.

[0084] Example 2 This embodiment discloses a microgrid power load forecasting system, including: The electricity consumption data classification module is configured to: collect electricity consumption data of users in the microgrid in real time, classify the electricity consumption data according to user type, usage time, and meteorological conditions, and generate energy consumption data classification records; An external factor analysis module is configured to: construct a power grid load forecasting model based on the classified records of the energy consumption data by calculating the correlation and weight of multiple external factors, and generate real-time load forecast results; A user group evaluation module is configured to: analyze user power usage data based on the real-time load forecast results, identify power usage patterns of multiple types of users, evaluate their impact on the microgrid load, and generate user behavior analysis results; a model parameter optimization module configured to: analyze the output changes and charging and discharging behaviors of multiple distributed energy resources in the microgrid on the load based on the user behavior analysis results, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate an energy impact assessment result; a node stability assessment module configured to: identify key nodes in the microgrid and assess their stability and response capabilities based on the energy impact assessment results, taking into account node locations and connectivity, and generate a node stability assessment result; The node influence assessment module is configured to: based on the node stability assessment result, adjust the abnormal load warning parameters considering the node stability, predict the abnormal load of the power grid in combination with the optimized power grid load prediction model, and generate power grid load prediction data.

[0085] Example 3 The purpose of this embodiment is to provide a computing device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method of embodiment 1 when executing the program.

[0086] Example 4 The purpose of this embodiment is to provide a computer-readable storage medium, a computer-readable storage medium having a computer program stored thereon, which performs the steps of the method of embodiment 1 when executed by a processor.

[0087] The steps involved in the apparatuses of Examples 3 and 4 above correspond to those of Method Example 1. For detailed implementation, please refer to the relevant description of Example 1. The term "computer-readable storage medium" should be understood to mean a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and causing the processor to perform any of the methods of the present invention.

[0088] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computer device. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0089] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0090] 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 method for predicting power load of a microgrid, characterized in that: include: Collect power usage data of users in the microgrid in real time, classify the power usage data according to user type, usage time, and meteorological conditions, and generate classified records of energy consumption data; Based on the classified records of the energy consumption data, a power grid load forecasting model is constructed by calculating the correlation and weight of multiple external factors to generate real-time load forecast results; Based on the real-time load forecast results, analyze the user's power usage data, identify the power usage patterns of multiple types of users and evaluate their impact on the microgrid load, and generate user behavior analysis results; Based on the user behavior analysis results, analyze the impact of output changes and charging and discharging behaviors of multiple distributed energy sources in the microgrid on the load, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate energy impact assessment results; By using the energy impact assessment results, considering the node locations and connectivity, identifying the key nodes in the microgrid and evaluating their stability and response capabilities, and generating a node stability assessment result; Based on the node stability evaluation result, the abnormal load warning parameters are adjusted taking into account the node stability, and the abnormal load of the power grid is predicted in combination with the optimized power grid load prediction model to generate power grid load prediction data.

2. A microgrid power load prediction method according to claim 1, characterized in that: Generate classified records of energy consumption data, including: Collect power usage data of microgrid users in real time, including power consumption, time records and meteorological information, and generate power data collection results; Based on the power data collection results, the data is classified according to user type, usage time, and meteorological conditions to generate classified power data; Based on classified power data, pattern recognition is performed on various power usage data, users' power usage behaviors and changing trends in various situations are analyzed, and classified records of energy consumption data are generated.

3. A microgrid power load prediction method according to claim 1, characterized in that: Generate real-time load forecast results, including: Based on the classified records of energy consumption data, the correlation between various external factors and load is analyzed, the influence degree of various external factors is evaluated, and the load impact analysis results are generated; Using the load impact analysis results, a logistic regression algorithm is used to construct a power grid load forecasting model and calculate the weights of multiple external factors on the load to generate a power grid load forecasting model; Based on the power grid load forecasting model, the power load change trend under multiple time periods and differentiated meteorological conditions is predicted to generate real-time load forecast results.

4. A microgrid power load prediction method according to claim 3, characterized in that: The logistic regression algorithm calculates the power load state according to the formula, which is expressed as: in, Indicates the power load status, represents the base of natural logarithms, is the intercept term, represents the weight of temperature, represents the weight of humidity, represents the weight of the electricity market price, represents the weight of sunshine duration, represents the weight of a special event, represents the weight of the power grid maintenance status, Indicates temperature, Indicates humidity, represents the electricity market price, Indicates the duration of sunshine. Indicates special events, Indicates the power grid maintenance status.

5. A microgrid power load prediction method according to claim 1, characterized in that: Generate user behavior analysis results, including: Based on real-time load forecast results, analyze the power usage data of multiple user categories in the microgrid, identify the power usage patterns of multiple user categories in different time periods, and generate power usage pattern data; Analyzing and identifying power usage characteristics of multiple user categories based on the power usage pattern data to generate user power usage characteristics; Based on the user power usage characteristics, the impact of multiple user categories on the microgrid load in multiple time periods is evaluated, the load contribution ratios of the multiple user categories are calculated, and user behavior analysis results are generated.

6. A microgrid power load prediction method according to claim 1, characterized in that: Obtain an optimized grid load forecasting model and generate energy impact assessment results, including: Based on the results of user behavior analysis, the output data of distributed energy resources in the microgrid is collected and analyzed to generate energy output data; Based on the energy output data, analyzing the impact of charging and discharging behavior of the power storage device on the grid load in multiple time periods to generate storage behavior analysis results; Based on the storage behavior analysis results, the load forecasting model is adjusted to match the impact of seasonal changes, weather conditions, and time changes, and an energy impact assessment result is generated.

7. A microgrid power load prediction method according to claim 1, characterized in that: Generate grid load forecast data, including: Based on the node stability assessment results, analyze the load data of multiple nodes in multiple time periods, identify the load carrying capacity of multiple nodes, and generate node load capacity data; Based on the node load capacity data, and according to the stability and load carrying capacity of the nodes, adjusting abnormal load warning threshold parameters of multiple nodes to generate load warning parameters; Based on the load warning parameters and in combination with meteorological conditions, the optimized power grid load forecasting model is used to predict risk load nodes at multiple time points and generate power grid load forecasting data.

8. A microgrid power load forecasting system, characterized in that: include: The electricity consumption data classification module is configured to: collect electricity consumption data of users in the microgrid in real time, classify the electricity consumption data according to user type, usage time, and meteorological conditions, and generate energy consumption data classification records; An external factor analysis module is configured to: construct a power grid load forecasting model based on the classified records of the energy consumption data by calculating the correlation and weight of multiple external factors, and generate real-time load forecast results; A user group evaluation module is configured to: analyze user power usage data based on the real-time load forecast results, identify power usage patterns of multiple types of users, evaluate their impact on the microgrid load, and generate user behavior analysis results; a model parameter optimization module configured to: analyze the output changes and charging and discharging behaviors of multiple distributed energy resources in the microgrid on the load based on the user behavior analysis results, optimize the grid load forecasting model, obtain the optimized grid load forecasting model, and generate an energy impact assessment result; a node stability assessment module configured to: identify key nodes in the microgrid and assess their stability and response capabilities based on the energy impact assessment results, taking into account node locations and connectivity, and generate a node stability assessment result; The node influence assessment module is configured to: based on the node stability assessment result, adjust the abnormal load warning parameters considering the node stability, predict the abnormal load of the power grid in combination with the optimized power grid load prediction model, and generate power grid load prediction data.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of a microgrid power load forecasting method as described in any one of claims 1 to 7 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the microgrid power load prediction method according to any one of claims 1 to 7 are implemented.

Images