Flexible power supply distribution method and system based on load prediction

Abstract

The invention relates to a flexible power supply distribution method and system based on load prediction, and the method comprises the following steps: carrying out the multi-dimensional feature extraction of historical load data of a power supply system, and obtaining a load distribution feature matrix; analyzing the load distribution characteristic matrix to obtain a load change trend parameter; on the basis of the load change trend parameter, performing supply-demand balance calculation on the power supply system to obtain a dynamic power supply distribution scheme; performing stability evaluation and adjustment on the dynamic power supply distribution scheme to obtain a safe and stable power supply distribution scheme; and based on the safe and stable power supply distribution scheme, carrying out real-time correction calculation on the dynamic power supply distribution scheme to obtain a flexible power supply control parameter group, and controlling the power supply system based on the flexible power supply control parameter group, thereby solving the problem that a traditional power supply distribution method mainly depends on a static model and historical data to carry out planning. And the method is difficult to adapt to rapidly changing load conditions and fluctuation characteristics of an intermittent power supply.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply systems, and in particular to a flexible power supply distribution method and system based on load prediction. Background Art

[0002] In modern power supply systems, with the widespread access to distributed energy and the growing demand for electricity, how to achieve efficient and stable power supply has become a major challenge. Traditional power distribution methods mainly rely on static models and historical data for planning, which is difficult to adapt to rapidly changing load conditions and the fluctuating characteristics of intermittent power sources. This not only limits the flexibility and response speed of the power system, but may also lead to an imbalance between supply and demand, affecting the stable operation of the power grid.

[0003] In addition, existing technologies usually use a single time series analysis or a simple regression model when dealing with load forecasting. These methods cannot fully capture the multi-dimensional characteristics of the load and its inherent changing trends. Especially in the face of complex and changeable user behavior patterns and extreme weather conditions, the prediction accuracy of traditional methods has dropped significantly, making it difficult for the power supply system to make timely and effective adjustments. Therefore, it is particularly important to develop a method that can accurately predict load changes and dynamically adjust the power supply strategy accordingly.

[0004] Furthermore, with the rise of the concept of smart grid, higher stability and safety requirements are put forward for the power supply system. However, in actual operation, there is a lack of a complete mechanism to evaluate and correct the stability and safety of the power distribution scheme in real time, which increases the operation risk of the power system. Especially during peak load periods, unreasonable power distribution may lead to problems such as voltage drop and frequency deviation, and even cause large-scale paralysis of the power system in severe cases. To this end, a flexible power distribution method based on load prediction is studied to improve the adaptive ability and anti-interference ability of the power system and ensure its safe and stable operation. Summary of the invention

[0005] The main purpose of the present invention is to provide a flexible power supply distribution method and system based on load forecasting, which solves the technical problem that the traditional power supply distribution method mainly relies on static models and historical data for planning and is difficult to adapt to rapidly changing load conditions and the fluctuating characteristics of intermittent power supplies.

[0006] To achieve the above object, the present invention provides a flexible power supply distribution method based on load prediction, which is applied to a power supply system and includes the following steps: Performing multi-dimensional feature extraction on the historical load data of the power supply system to obtain a load distribution feature matrix; Performing multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters; Predicting whether there is a supply-demand imbalance in the power supply system based on the load change trend parameter, and if so, performing supply-demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation plan; By using a preset power flow analysis method, the dynamic power supply distribution scheme is evaluated for stability and adjusted to obtain a safe and stable power supply distribution scheme; Based on the safe and stable power supply allocation scheme, the dynamic power supply allocation scheme is corrected and calculated in real time to obtain a flexible power supply control parameter group, and the power supply system is controlled based on the flexible power supply control parameter group.

[0007] Furthermore, the historical load data of the power supply system is subjected to multi-dimensional feature extraction to obtain a load distribution feature matrix, including: Sampling the historical load data of the power supply system in different time periods to obtain a load time series data set, and performing spectrum decomposition processing on the load time series data set to obtain a load frequency domain feature sequence; Based on the load frequency domain feature sequence, correlation analysis is performed on the power load in the power supply system to obtain a load correlation feature matrix, and singular value decomposition is performed on the load correlation feature matrix to obtain a load principal component feature set, wherein the load principal component feature set includes a load change gradient, a power factor distribution, and a load density coefficient; The instantaneous features of the load principal component feature set are extracted by Hilbert transform to obtain the load instantaneous feature vector, and the load instantaneous feature vector is combined in multiple dimensions to obtain a load distribution feature matrix, wherein the load distribution feature matrix includes power fluctuation features, load distribution density features and power consumption cycle features.

[0008] Furthermore, the load distribution characteristic matrix is ​​subjected to multi-scale analysis by wavelet transform decomposition to obtain load change trend parameters, including: Performing multi-resolution wavelet decomposition on the load distribution characteristic matrix through wavelet transform decomposition to obtain a multi-scale characteristic coefficient matrix, and performing singular spectrum analysis on the multi-scale characteristic coefficient matrix to obtain a load fluctuation characteristic vector; Based on the load fluctuation feature vector, a time-frequency joint analysis is performed on the power supply system to obtain a time-frequency feature tensor, and an asymmetric orthogonal decomposition is performed on the time-frequency feature tensor to obtain a load dynamic feature set; The load dynamic feature set is subjected to feature fusion processing by wavelet packet entropy value calculation to obtain a multi-dimensional trend feature matrix, and the multi-dimensional trend feature matrix is ​​subjected to adaptive threshold segmentation to obtain load change trend parameters.

[0009] Furthermore, based on the load change trend parameter, the power supply system is subjected to supply and demand balance calculation to obtain a dynamic power supply allocation scheme, including: Performing dynamic time series decomposition on the load change trend parameter to obtain a load power fluctuation sequence, and performing multi-dimensional phase compensation calculation on the load power fluctuation sequence to obtain a load power compensation feature group; Based on the load power compensation feature group, power distribution calculation is performed on the power supply system to obtain a power distribution feature matrix, and based on the power distribution feature matrix, topology optimization analysis is performed on the power supply system to obtain a power supply network topology feature set, wherein the power supply network topology feature set includes node connection weights, branch impedance distribution, and voltage gradient coefficient; Calculating the supply-demand matching degree of the power supply network topology feature set by a multi-objective constraint optimization method to obtain a supply-demand balance feature vector, and dynamically allocating weights to the supply-demand balance feature vector to obtain a power dispatch parameter group; Based on the power dispatch parameter group, the power supply system is subjected to regional load distribution calculation to obtain a regional power supply feature sequence, and the regional power supply feature sequence is subjected to adaptive boundary constraint processing to obtain a power supply boundary feature set, wherein the power supply boundary feature set includes a regional power limit, a voltage fluctuation range, and a load transfer threshold; A dynamic power supply scheme is generated for the power supply boundary feature set through a hierarchical progressive optimization method to obtain a dynamic power supply allocation scheme, wherein the dynamic power supply allocation scheme includes a power allocation strategy, a voltage regulation scheme and load balancing parameters.

[0010] Furthermore, the power distribution calculation of the power supply system is performed based on the load power compensation feature group to obtain a power distribution feature matrix, including: Performing multi-dimensional space mapping decomposition on the load power compensation feature group to obtain a power compensation vector set, and performing phase difference calculation on the power compensation vector set to obtain a voltage-power coupling matrix; Based on the voltage-power coupling matrix, the power supply system is tracked for node power flow to obtain a power distribution map, and a topological sensitivity analysis is performed on the power distribution map to obtain a power supply network transmission characteristic vector; The power supply network transmission characteristic vector is subjected to power partition clustering by a double-layer recursive partitioning method to obtain a power sub-network set, and the interconnection coupling degree of the power sub-network set is calculated to obtain an inter-regional power exchange matrix; Based on the inter-regional power exchange matrix, a hierarchical coordination optimization calculation is performed on the power supply system to obtain a multi-level power supply coordination parameter group, and a robustness evaluation is performed on the multi-level power supply coordination parameter group to obtain a power distribution elasticity coefficient set; The power distribution elasticity coefficient set is dynamically integrated through an adaptive weight fusion mechanism to obtain a power distribution feature matrix, wherein the power distribution feature matrix includes node power distribution ratios, voltage control strategy parameters and load balancing adjustment data.

[0011] Furthermore, the power flow analysis method is a Newton-Raphson power flow iteration method. The dynamic power supply distribution scheme is evaluated and adjusted for stability by a preset power flow analysis method to obtain a safe and stable power supply distribution scheme, including: The nonlinear power flow calculation of the dynamic power supply distribution scheme is performed by the Newton-Raphson power flow iteration method to obtain a system node voltage distribution feature set, and a power flow sensitivity analysis is performed on the system node voltage distribution feature set to obtain a power grid power transmission characteristic matrix; When the minimum singular value of the power transmission characteristic matrix of the power grid is less than a preset threshold, a continuous power flow tracking analysis is performed on the power supply system based on the power transmission characteristic matrix of the power grid to obtain a system voltage stability margin curve, and critical points are identified on the system voltage stability margin curve to obtain a characteristic map of the vulnerable area of ​​the power grid; The power control scheme is optimized by using the AC power flow optimization analysis method to optimize the characteristic map of the vulnerable area of ​​the power grid, so as to obtain a safe and stable power supply distribution scheme, wherein the safe and stable power supply distribution scheme includes an optimal power distribution strategy, a dynamic reactive power compensation configuration and a voltage regulation control coefficient.

[0012] Furthermore, based on the safe and stable power supply allocation scheme, the dynamic power supply allocation scheme is corrected and calculated in real time to obtain a flexible power supply control parameter group, including: Performing high-order wavelet singular spectrum analysis on the safe and stable power supply distribution scheme to obtain a power supply scheme feature vector sequence, and performing phase compensation calculation on the power supply scheme feature vector sequence to obtain a real-time power supply control reference parameter group, wherein the real-time power supply control reference parameter group includes a power adjustment step coefficient, a node voltage control gain, and a load response time constant; Based on the real-time power supply control reference parameter group, the dynamic power supply allocation scheme is subjected to orthogonal decomposition processing to obtain a power supply scheme difference feature matrix, and multi-scale entropy value calculation is performed on the power supply scheme difference feature matrix to obtain a power supply correction quantization parameter set; The time-space domain features of the power supply correction quantization parameter set are fused through nonlinear adaptive filtering to obtain a power supply dynamic response feature sequence, and the power supply dynamic response feature sequence is subjected to fuzzy reasoning optimization processing to obtain a flexible power supply control parameter group, wherein the flexible power supply control parameter group includes real-time power allocation instructions, dynamic voltage regulation strategies and load response control signals.

[0013] The present invention also provides a flexible power supply distribution system based on load prediction, which is applied to a power supply system and includes: An extraction module, used for performing multi-dimensional feature extraction on the historical load data of the power supply system to obtain a load distribution feature matrix; An analysis module, used for performing multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters; A calculation module, used to predict whether there is a supply-demand imbalance in the power supply system based on the load change trend parameter, and if so, perform supply-demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation plan; An adjustment module, used to evaluate and adjust the stability of the dynamic power supply distribution scheme through a preset power flow analysis method to obtain a safe and stable power supply distribution scheme; A correction module is used to perform real-time correction calculation on the dynamic power supply allocation scheme based on the safe and stable power supply allocation scheme, obtain a flexible power supply control parameter group, and control the power supply system based on the flexible power supply control parameter group.

[0014] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0016] The present invention provides a flexible power supply distribution method based on load prediction, comprising the following steps: extracting multi-dimensional features from historical load data of the power supply system to obtain a load distribution feature matrix; performing multi-scale analysis on the load distribution feature matrix through wavelet transform decomposition to obtain load change trend parameters; predicting whether there is a supply and demand imbalance in the power supply system based on the load change trend parameters, and if so, performing supply and demand balance calculation on the power supply system based on the load change trend parameters to obtain a dynamic power supply distribution scheme; evaluating and adjusting the stability of the dynamic power supply distribution scheme through a preset power flow analysis method to obtain a safe and stable power supply distribution scheme; performing real-time correction calculation on the dynamic power supply distribution scheme based on the safe and stable power supply distribution scheme to obtain a flexible power supply control parameter group, and controlling the power supply system based on the flexible power supply control parameter group, thereby solving the technical problem that the traditional power supply distribution method mainly relies on static models and historical data for planning and is difficult to adapt to rapidly changing load conditions and the fluctuation characteristics of intermittent power supplies, and realizing the stability evaluation and adjustment of the dynamic power supply distribution scheme using a preset power flow analysis method, thereby ensuring the safety and stability of the final power supply distribution scheme. This method can effectively detect various factors that may affect the stability of the power grid and eliminate hidden dangers through appropriate adjustments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the steps of a flexible power supply distribution method based on load prediction in one embodiment of the present invention; Figure 2 is a structural block diagram of a flexible power supply distribution system based on load prediction in one embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] like Figure 1 As shown, Figure 1 It is a schematic diagram of the steps of a flexible power supply distribution method based on load prediction in one embodiment of the present invention; In one embodiment of the present invention, a flexible power supply distribution method based on load prediction is provided, which is applied to a power supply system and includes the following steps: Step S1, performing multi-dimensional feature extraction on historical load data of the power supply system to obtain a load distribution feature matrix.

[0021] Specifically, when discussing how to extract multi-dimensional features from the historical load data of the power supply system to obtain the load distribution feature matrix, we first need to understand that the core of this process is to identify key features from complex data sets that help understand and predict load behavior. This step is the basis of the flexible power supply allocation method based on load prediction, because it provides the necessary input for the subsequent wavelet transform decomposition, load change trend analysis, and supply and demand balance calculation. Specifically, the process involves extracting information from the massive historical data of the power supply system, which may include power consumption, weather conditions, user types, etc. at different time scales, aiming to build a matrix that comprehensively reflects the load characteristics. The first step to achieve this goal is to collect various historical data related to the power supply system. These data are not limited to power consumption, but also include various external factors that may cause load fluctuations, such as seasonal changes, holiday effects, temperature fluctuations, etc. For example, during the high temperature period in summer, the use of air conditioners increases significantly, which will directly affect the power demand; while in winter, the use of heating equipment becomes the main load growth point. By integrating these different data sources, a preliminary data set can be formed, which contains a variety of factors that affect the load of the power supply system. Next, in order to ensure the quality and applicability of the data, the raw data must be preprocessed. This includes data cleaning (removing erroneous or incomplete records), normalization (making data of different scales comparable), and feature selection (determining which variables are most critical for load forecasting). For example, when faced with a large amount of meteorological data, only parameters such as temperature and humidity that are directly related to power load are selected as features, while those with less correlation are ignored. In this way, we can reduce redundant information and improve the efficiency and accuracy of model training. Once the data preprocessing stage is completed, we enter the core link of multi-dimensional feature extraction. Here, "multi-dimensional" refers to analyzing load data from multiple angles to capture its inherent laws and potential patterns. For example, in addition to considering changes in time series, different regions can be divided according to geographical location to explore the differences in electricity usage habits of users in each region; or classified according to user types (such as industrial users, commercial users, and residential users) to understand the differences in electricity usage patterns between different types of users. In this way, the rich information in the load data can be revealed from different levels, so as to construct a more accurate load distribution feature matrix. The process of constructing the load distribution feature matrix is ​​actually to combine the information of the above various dimensions to form a structured representation. In this matrix, each row may represent a specific time period or user group, and each column corresponds to a characteristic attribute. For example, for an industrial area, a row of the matrix may be the average electricity consumption per hour on a certain day, and the corresponding column includes factors such as the highest temperature of the day, whether it is a weekday or a weekend, and whether it is a special holiday.This structured representation not only facilitates the subsequent wavelet transform decomposition, but also helps researchers to more intuitively understand the interaction between various factors and their impact on the load. For example, suppose we want to optimize the power supply system of a city, which is divided into several areas, and each area has a variety of user types, including residential areas, commercial centers, and industrial parks. By collecting and organizing data such as the average daily power consumption, weather conditions, and public holiday arrangements in each area over the past year, we can construct a detailed load distribution feature matrix. Then, using wavelet transform decomposition and other techniques to conduct in-depth analysis of the matrix, we can not only identify the peak hours of electricity consumption on a daily, weekly, or even annual basis, but also discover the instantaneous pressure on the local power grid caused by certain special events (such as large exhibitions and sports events). These insights provide a scientific basis for further formulating targeted power supply strategies, such as dispatching backup power sources in advance during expected high load periods, or adjusting electricity price policies to encourage peak power consumption, and ultimately achieve efficient and stable operation of the power supply system. In short, constructing an accurate load distribution feature matrix through meticulous historical data analysis is an important step in achieving smart grid management.

[0022] Step S2, performing multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters.

[0023] Specifically, in the implementation of the flexible power supply distribution method based on load forecasting, the process of multi-scale analysis of the load distribution feature matrix by wavelet transform decomposition to obtain the load change trend parameters is crucial. This process aims to analyze the potential patterns and dynamic characteristics in the load data from different time scales, thereby providing key inputs for subsequent steps. Specifically, wavelet transform is a powerful tool that can perform signal analysis in both time and frequency domains. It is particularly suitable for processing non-stationary signals, that is, signals whose statistical characteristics change over time, which is a significant feature of power load data. First of all, it is necessary to understand that the load distribution feature matrix contains historical data of the power supply system in multiple dimensions, which reflects the impact of different factors on the load. However, it is not easy to extract useful load change trends directly from these raw data because they usually contain a lot of noise and irregular fluctuations. Therefore, the wavelet transform decomposition method can effectively decompose these complex data into different frequency components, each of which represents the load change pattern at different time scales. For example, in a typical daily power consumption cycle, there may be a clear difference between the peak period during the day and the valley period at night; while in a longer time range, such as seasonal changes or holiday effects, a slower change trend will result. Wavelet transform can identify and separate the characteristics of changes on these different time scales. In actual operation, choosing the right wavelet basis function is a crucial step. Different wavelet basis functions have different mathematical properties and applicable scenarios. For example, Morlet wavelet is suitable for situations with good time-frequency localization, while Haar wavelet is often used for fast calculation due to its simple structure. According to the specific load data characteristics and analysis purposes, selecting the most suitable wavelet basis function is crucial to accurately capture the load change trend. For example, when analyzing the daytime peak load caused by the use of air conditioners in a city in summer, a wavelet basis function that can accurately reflect the rapid changes in a short period of time may be required; on the contrary, if the focus is on the annual electricity consumption growth trend, it is more appropriate to use a wavelet basis function that can effectively capture the long-term change pattern. Once the wavelet basis function is determined, the next step is to perform the wavelet transform decomposition process. In this process, the original load distribution feature matrix is ​​converted into a series of wavelet coefficients, which correspond to the load information at different scales. In this way, not only short-term fluctuations (such as daily or weekly changes) can be clearly seen, but also long-term trends (such as monthly or annual changes) can be observed. For example, when analyzing the electricity consumption records of an industrial area within a year, wavelet transform can help us find that in addition to the obvious weekly cycle (the difference between weekdays and weekends), there are also slower changing trends caused by factors such as production plan adjustments and changes in market supply and demand. In addition, by further processing the wavelet coefficients, we can extract load change trend parameters.This usually involves statistical analysis of wavelet coefficients at different scales, such as mean, variance, etc., to quantify the intensity and direction of changes at each scale. These parameters not only reflect the specific pattern of load changes over time, but also provide necessary inputs for subsequent steps. For example, when predicting load demand in the future, these trend parameters can be used to more accurately estimate the possible peak or valley periods, so as to make corresponding scheduling preparations in advance. To better understand the application scenario of this process, consider an example: in a power supply area with multiple user types, including residential areas, commercial centers, and industrial parks. The power supply system in this area hopes to improve efficiency and stability through optimized scheduling. By performing wavelet transform decomposition on the power consumption data of the past few years, the power consumption patterns of different types of users and their changes over time can be revealed. For example, it is found that the peak power consumption in residential areas is mainly concentrated in the evening to late night, while the commercial center reaches its peak during the daytime on weekdays; as for the industrial park, its power consumption pattern is greatly affected by the production cycle and presents a more complex periodic fluctuation. Based on these in-depth analysis results, the power supply system can formulate more flexible and effective power supply strategies, such as increasing power generation capacity in advance or optimizing the grid configuration during expected high load periods to ensure the stable operation of the entire system. In short, obtaining load change trend parameters through wavelet transform decomposition provides a solid foundation for the efficient management and optimization of smart grids.

[0024] Step S3, predicting whether there is a supply-demand imbalance in the power supply system based on the load change trend parameters, and if so, performing supply-demand balance calculation on the power supply system based on the load change trend parameters to obtain a dynamic power supply allocation plan.

[0025] Specifically, based on the load change trend parameters, it is predicted whether there is a supply and demand imbalance in the power supply system. This process is one of the key links in the entire flexible power supply distribution method. Its purpose is to predict the imbalance between power supply and demand that may occur in the future by analyzing historical data and current operating status. Once it is confirmed that there is a risk of supply and demand imbalance, the load change trend parameters are further used to calculate the supply and demand balance of the power supply system to formulate a dynamically adjusted power supply distribution plan to ensure the stable operation of the power system. First of all, to achieve this step, it is necessary to rely on the load change trend parameters obtained by wavelet transform decomposition. These parameters describe in detail the load fluctuation patterns at different time scales, including short-term daily changes and long-term trend changes. In the prediction stage, these trend parameters are input into one or more prediction models, which can be based on machine learning methods such as support vector machines, neural networks, etc., or traditional statistical methods such as regression analysis. The choice of model depends on the specific application scenario and the quality of available data. For example, when facing a power supply area with multiple types of users, considering the power consumption characteristics and change patterns of residential areas, commercial centers and industrial parks, we can build a multivariate prediction model that not only takes into account the load changes in time series, but also incorporates the influence of external factors such as weather conditions and holiday arrangements. When the prediction model receives the load change trend parameter as input, it will perform operations according to the pre-set algorithm logic to evaluate whether there will be a supply and demand imbalance in the power supply system in the future. Here, "supply and demand imbalance" refers to the state where the power supply capacity cannot meet the actual demand, which may be due to the risk of power outages caused by insufficient supply or the waste of resources caused by oversupply. Therefore, accurately identifying this potential mismatch is crucial to maintaining the stability of the power grid. For example, during the high temperature period in summer, with the surge in the use of air conditioners, the power demand in some areas may suddenly rise, which may lead to overload or even collapse of the local power grid if it is not fully prepared in advance. Once the prediction results show that the power supply system is at risk of supply and demand imbalance, it is necessary to calculate the supply and demand balance of the power supply system based on the load change trend parameter. This process involves complex mathematical optimization problems, which aims to find an optimal power distribution strategy that minimizes energy loss and improves the overall efficiency of the system while meeting the needs of all users. Specifically, the supply and demand balance calculation usually takes into account multiple constraints, such as the maximum output power of the power plant, the transmission capacity limit of the transmission line, and the charging and discharging rate of the energy storage device. At the same time, some objective functions are introduced, such as minimizing costs, maximizing reliability, or optimizing environmental impact, to guide the decision-making process.For example, suppose that in an urban power supply system with multiple types of users, we have obtained detailed load change trend parameters through wavelet transform decomposition, and predicted that there will be a significant demand peak in the upcoming weekend, mainly because a large music festival held locally attracted a large number of tourists. In this case, in order to cope with the possible imbalance between supply and demand, we need to develop a dynamic power supply allocation plan. This plan should not only take into account the existing power generation capacity and transmission infrastructure, but also flexibly mobilize distributed energy resources such as solar panels and wind turbines, and even energy storage batteries at electric vehicle charging stations. In addition, the electricity price mechanism can be adjusted in real time to encourage residents and enterprises to use electricity at off-peak times, thereby alleviating the pressure during peak periods. Finally, after a series of fine calculations and simulation tests, the final dynamic power supply allocation plan will be implemented in the actual power supply management system. This solution is highly flexible and adaptable, and can be continuously adjusted and optimized according to real-time monitoring data to ensure that the power supply system can maintain safe and stable operation both under normal operating conditions and when emergencies occur. In short, by predicting the imbalance between supply and demand based on load change trend parameters and performing supply and demand balance calculations based on them, a scientific and effective management method is provided for smart grids, which helps to improve the adaptive ability and anti-interference ability of the power system.

[0026] Step S4, using a preset power flow analysis method, the dynamic power supply distribution scheme is evaluated for stability and adjusted to obtain a safe and stable power supply distribution scheme.

[0027] Specifically, in the implementation of the flexible power supply distribution method based on load forecasting, the key step to ensure that a safe and stable power supply distribution scheme is finally obtained is to evaluate and adjust the stability of the dynamic power supply distribution scheme through the preset power flow analysis method. This process not only involves complex mathematical models and algorithms, but also requires a deep understanding of the working mechanism inside the power system and the influence of external environmental factors. Specifically, power flow analysis is a technology used to study the voltage, current, power distribution and their relationship in the power network. Through the precise calculation of these parameters, the stability and safety of the power grid under different operating conditions can be effectively evaluated. First, after obtaining the dynamic power supply distribution scheme, it needs to be input into the preset power flow analysis model. This model is usually built based on a series of assumptions, including but not limited to the maximum output power of the generator set, the transmission capacity limit of the transmission line, the voltage regulation range of the substation, etc. These assumptions reflect the physical characteristics and operating limitations of the actual power grid, and are therefore crucial for accurately simulating the behavior of the power grid. For example, in an urban power supply system that includes residential areas, commercial centers, and industrial parks, considering the power consumption characteristics and demand change laws of each area, we can set corresponding boundary conditions based on historical data. For example, industrial areas may have high electricity consumption due to the needs of production activities and have strict requirements for voltage stability; while residential areas are more concerned about the power supply guarantee during peak hours. Next, the dynamic power supply distribution scheme is calculated and evaluated in detail using the power flow analysis method. In this process, the core task is to perform power balance analysis on each node (i.e., substation or load point) in the power grid to determine the voltage level, power flow and potential overload risk of the entire system under the current power supply strategy. This step requires comprehensive consideration of multiple factors, such as coordination between various power generation sources, thermal limits of transmission lines, and rated capacity of transformers. For example, during the high temperature period in summer, with the surge in air conditioning use, some key transmission corridors may face overload risks. At this time, power flow analysis is needed to identify these problem areas and propose corresponding improvement measures. In order to improve the accuracy of the analysis results, modern power flow analysis often uses advanced numerical algorithms, such as the Newton-Raphson method or the fast decoupling method. These algorithms can efficiently solve large-scale nonlinear equations to quickly obtain the state variables of each node in the power grid. In addition, simulation software tools such as PSS / E or MATPOWER can be combined to conduct more intuitive and comprehensive simulation experiments. These tools can not only provide detailed calculation results, but also help engineers visualize the performance of the power grid under different operating conditions so as to make more scientific and reasonable decisions. Once the preliminary power flow analysis is completed, the next step is to adjust and optimize the problems found. If the analysis results show that there is a risk of voltage drop or line overload in certain areas, it is necessary to review and adjust the original power distribution plan.For example, in the case of the above-mentioned urban power supply system, if it is found that the power demand of an industrial park exceeds expectations, causing the main transmission line connecting the area to approach its maximum carrying capacity, this problem can be alleviated by increasing the investment in backup power, adjusting the load distribution in other areas, or upgrading existing transmission facilities. At the same time, it is also possible to consider introducing distributed energy resources (DERs), such as solar panels and wind turbines, as a supplementary power supply source to further enhance the flexibility and reliability of the power grid. Finally, after multiple iterations and optimizations, when all potential safety hazards are properly handled, a safe power supply distribution plan that meets user needs and ensures the stability of the power grid can be obtained. This plan should not only take into account short-term operational efficiency, but also take into account long-term development plans, such as the integration of renewable energy and the application of smart grid technology. For example, by reasonably arranging the location and scale of energy storage equipment, excess electricity can be stored during the low-power consumption period and released during the peak period, so as to smooth the overall load curve, reduce dependence on traditional power generation methods, and promote the widespread application of clean energy. In summary, the stability assessment and adjustment of the dynamic power supply distribution plan through the preset power flow analysis method is an important means to ensure the safe and stable operation of the power system. It not only helps us to find and solve the problems in the power grid in time, but also provides strong support for formulating scientific and reasonable power supply strategies. Just as when facing a complex power supply network with multiple types of users, only through accurate power flow analysis can we truly achieve a dynamic balance between supply and demand and provide a solid energy guarantee for the development of social economy.

[0028] Step S5: Based on the safe and stable power supply allocation scheme, perform real-time correction calculation on the dynamic power supply allocation scheme to obtain a flexible power supply control parameter group, and control the power supply system based on the flexible power supply control parameter group.

[0029] Specifically, based on the safe and stable power supply distribution scheme, the dynamic power supply distribution scheme is corrected and calculated in real time, and finally a flexible power supply control parameter group is obtained. This process aims to ensure that the power system can achieve efficient, reliable and flexible power supply under various conditions by continuously monitoring and adjusting the operating status of the power system. This step is one of the core links of the entire flexible power supply distribution method. It not only relies on the data and analysis results obtained in the previous steps, but also combines the actual operating status of the current power grid and the changes in external environmental factors to achieve optimal power dispatching and management. First, in order to perform real-time correction calculations, it is necessary to establish a powerful data acquisition and monitoring system (SCADA), which can collect operating data from various nodes (including power stations, substations, transmission lines, etc.) in real time, such as voltage level, current intensity, power factor, etc. These data provide the necessary input for subsequent analysis, so that we can keep track of the state changes of the power grid at any time. For example, in an urban power supply network including residential areas, commercial centers and industrial parks, due to the significant differences in the power consumption patterns of different types of users, it is necessary to closely monitor the load fluctuations in each area. Especially for industrial users, sudden changes in their production activities may cause supply and demand imbalances in local power grids; while residential users may increase the use of air conditioning or heating equipment due to changes in weather conditions (such as extreme heat or cold), thus affecting the overall load distribution. Once the latest grid operation data is obtained, the next step is to combine it with the previously determined safe and stable power supply distribution plan to perform real-time correction calculations. The goal of this step is to minimize the deviation between the actual operating state and the ideal state without violating any physical constraints. Specifically, this means dynamically adjusting the original planned power generation, transmission, and energy storage equipment operation strategies based on real-time data. For example, if it is found that the load rate of a major transmission line is close to its maximum capacity during a certain period of time, then immediate measures need to be taken, such as reducing the output power of the relevant generators or transferring part of the load to other paths to avoid potential risks. In addition, distributed energy resources (DERs), such as solar panels and wind turbines, can be used as supplementary power sources to further enhance the flexibility of the grid. When performing real-time correction calculations, advanced algorithms and technologies are usually used, such as model predictive control (MPC) and adaptive control. These methods can respond quickly in complex and changing environments and provide optimal solutions. Take MPC as an example. It predicts the operating status of the power grid in the future, and then formulates a series of control action sequences based on the prediction results, and then selects the best set to apply to the current moment. This method is particularly suitable for dealing with systems with obvious dynamic characteristics, such as power networks, because it not only takes into account immediate impacts, but also includes considerations of future trends.For example, in the face of the upcoming holiday peak period, the possible high load situation can be predicted in advance, and the start-up time and location of the backup power supply can be arranged accordingly, so as to respond to the sudden demand in time. Based on the above calculation results, a set of flexible power supply control parameter groups is finally formed. This set of parameters covers multiple aspects, such as power generation dispatching instructions, energy storage device charging and discharging plans, demand-side response signals, etc., which together constitute a specific guide for guiding the operation of the power system. For example, in a power supply area with multiple types of users, considering the differences in electricity consumption habits of various types of users in different time periods, specific demand-side response signals can be sent to encourage residential users to use electrical equipment during non-peak hours, while requiring industrial enterprises to flexibly adjust production shifts according to the actual situation of the power grid. In addition, a reasonable electricity price mechanism can be set to guide users to consciously participate in peak-shifting electricity consumption through price levers, thereby effectively alleviating the pressure on the power grid. Finally, precise control of the power supply system based on the flexible power supply control parameter group is the key to realizing the smart grid. Through automated control systems such as remote terminal units (RTUs) and programmable logic controllers (PLCs), instructions can be directly issued to each control point and feedback information can be received to ensure that all operations are carried out smoothly according to the predetermined plan. For example, during the peak period of electricity consumption in summer, when the power grid is detected to be facing the risk of overload, the pre-configured emergency plan can be quickly activated, including enabling backup power generation facilities, adjusting the working mode of energy storage equipment, and issuing emergency power restriction notices, so as to ensure the stable operation of the entire system. In short, through this continuous real-time correction and fine-tuning, not only the reliability of the power system is improved, but also a better service experience is provided to users.

[0030] In a specific embodiment, the multi-dimensional feature extraction is performed on the historical load data of the power supply system to obtain a load distribution feature matrix, including: Sampling the historical load data of the power supply system in different time periods to obtain a load time series data set, and performing spectrum decomposition processing on the load time series data set to obtain a load frequency domain feature sequence; Based on the load frequency domain feature sequence, correlation analysis is performed on the power load in the power supply system to obtain a load correlation feature matrix, and singular value decomposition is performed on the load correlation feature matrix to obtain a load principal component feature set, wherein the load principal component feature set includes a load change gradient, a power factor distribution, and a load density coefficient; The instantaneous features of the load principal component feature set are extracted by Hilbert transform to obtain the load instantaneous feature vector, and the load instantaneous feature vector is combined in multiple dimensions to obtain a load distribution feature matrix, wherein the load distribution feature matrix includes power fluctuation features, load distribution density features and power consumption cycle features.

[0031] Specifically, when discussing how to extract multi-dimensional features from the historical load data of the power supply system to obtain the load distribution feature matrix, we must first clarify that this process is based on an in-depth analysis of the historical load data of the power supply system, aiming to identify key features from complex data sets that help understand and predict load behavior. This process first involves sampling the historical load data of the power supply system in different time periods to obtain a load time series data set, and converting the load information in the time domain into a feature sequence in the frequency domain through spectral decomposition processing. This step not only lays the foundation for subsequent correlation analysis, but also reveals the periodic and non-periodic components in the load data. Specifically, after obtaining a large amount of historical load data of the power supply system, the first step is to sample these data in different time periods, which means extracting load data according to specific time intervals (such as hourly, daily, or weekly) to form a load time series data set. This time-division method can help us better capture the load change pattern in different time periods, such as the difference between weekdays and weekends, and between day and night. Next, in order to more fully understand the intrinsic characteristics of these load data, they need to be subjected to spectral decomposition processing. Spectral decomposition is a technique for transforming signals from the time domain to the frequency domain, and the commonly used one is the fast Fourier transform (FFT). Through this method, the information of different frequency components can be extracted from the original load time series data to form a load frequency domain feature sequence. For example, when analyzing the annual electricity consumption records of a city, it can be found that certain fixed frequency components correspond to daily, weekly and even annual electricity consumption patterns. Based on the load frequency domain feature sequence obtained above, the next step is to perform correlation analysis on the power loads in the power supply system to identify possible correlations between different loads. This step is crucial for constructing the load correlation feature matrix because it can help us understand which load factors have significant correlations and how these correlations change over time and with external conditions. For example, in an urban power supply network that includes residential areas, commercial centers and industrial parks, it can be found through analysis that the peak electricity consumption in industrial areas usually occurs during working hours on weekdays, while residential areas reach peaks from dusk to late at night. The identification of this correlation not only helps to optimize the power dispatching strategy, but also provides the necessary input for the subsequent singular value decomposition. Singular value decomposition is a powerful linear algebra technique that can decompose complex matrices into the form of several simple matrix products. Here, it is used to process the load correlation feature matrix. By performing singular value decomposition on the matrix, we can extract the main component feature set of the load, which includes key parameters such as load change gradient, power factor distribution and load density coefficient. These parameters reflect the speed of load change over time, an important indicator of power quality and the average load level per unit area.For example, by analyzing the electricity consumption data of a certain area in the past few years, its load change gradient can be calculated to understand the growth trend of electricity demand; at the same time, the power factor distribution in the area can be evaluated to determine whether there is a problem of reactive power excess, and then propose improvement measures. Subsequently, in order to further explore the instantaneous characteristics in the load data, the Hilbert transform is used to process the load principal component feature set. The Hilbert transform is an effective tool for analyzing the instantaneous properties of signals. It can extract the instantaneous amplitude and phase information of the signal without losing the original signal information. In this process, by applying the Hilbert transform to the load principal component feature set, the load instantaneous feature vectors can be obtained, which contain the specific state of the load at any given moment. For example, in the process of monitoring a city power grid, the instantaneous changes of the load in a certain area can be monitored in real time, and sudden high load events can be discovered and handled in time. Finally, by combining the load instantaneous feature vectors in multiple dimensions, the load distribution feature matrix is ​​finally obtained. This matrix integrates various features extracted in all previous steps, including but not limited to power fluctuation features, load distribution density features, and power consumption cycle features. For example, when considering an urban power supply system composed of multiple types of users, by constructing such a detailed load distribution feature matrix, we can not only clearly see the power consumption patterns of different types of users in each area and their changes over time, but also provide a scientific basis for formulating targeted power supply strategies. For example, measures such as pre-dispatching backup power sources during expected high load periods, or adjusting electricity price policies to encourage peak power consumption are all effective response strategies based on a deep understanding of these characteristics. In short, constructing an accurate load distribution feature matrix through meticulous historical data analysis is an important step in realizing smart grid management and the key to improving the efficiency and reliability of the entire power system.

[0032] In a specific embodiment, the multi-scale analysis of the load distribution characteristic matrix is ​​performed through wavelet transform decomposition to obtain load change trend parameters, including: Performing multi-resolution wavelet decomposition on the load distribution characteristic matrix through wavelet transform decomposition to obtain a multi-scale characteristic coefficient matrix, and performing singular spectrum analysis on the multi-scale characteristic coefficient matrix to obtain a load fluctuation characteristic vector; Based on the load fluctuation feature vector, a time-frequency joint analysis is performed on the power supply system to obtain a time-frequency feature tensor, and an asymmetric orthogonal decomposition is performed on the time-frequency feature tensor to obtain a load dynamic feature set; The load dynamic feature set is subjected to feature fusion processing by wavelet packet entropy value calculation to obtain a multi-dimensional trend feature matrix, and the multi-dimensional trend feature matrix is ​​subjected to adaptive threshold segmentation to obtain load change trend parameters.

[0033] Specifically, when discussing how to perform multi-scale analysis on the load distribution feature matrix through wavelet transform decomposition to obtain load change trend parameters, we first need to understand that this process aims to analyze the potential patterns and dynamic characteristics in the load data from different time scales. This step is based on the previously constructed load distribution feature matrix, which contains multi-dimensional information such as power fluctuations, load distribution density and power consumption cycle in the power supply system. Specifically, this process uses the powerful ability of wavelet transform to decompose complex load data into multiple frequency components and extract key features that help understand and predict load behavior. First, the load distribution feature matrix is ​​decomposed by multi-resolution wavelet through wavelet transform decomposition, which is the first step of the whole process. As a tool that can perform signal analysis in both time and frequency domains, wavelet transform is particularly suitable for processing non-stationary signals, that is, signals whose statistical characteristics change over time. At this stage, the original load distribution feature matrix is ​​converted into a feature coefficient matrix containing multi-scale information. For example, in a typical urban power supply network, there may be daily, weekly or even annual peak and trough periods of power consumption; through wavelet transform decomposition, we can identify and separate these change characteristics on different time scales. Each scale corresponds to a different frequency component, representing short-term fluctuations (such as intraday changes) and long-term trends (such as seasonal changes) in the load data. In addition, in order to further extract the key information in these multi-scale characteristic coefficient matrices, singular spectrum analysis is performed on them to obtain the load fluctuation eigenvectors. Singular spectrum analysis is an effective dimensionality reduction technique that can help us extract the most representative features from complex multi-scale characteristic coefficient matrices. These eigenvectors reflect the main patterns of load changes over time. Next, based on the load fluctuation eigenvectors obtained above, a time-frequency joint analysis is performed on the power supply system to obtain the time-frequency feature tensor. This process combines the advantages of time series analysis and spectrum analysis, allowing us to not only observe the state of the load at different time points, but also understand its performance at different frequencies. For example, when facing an urban power supply network containing residential areas, commercial centers, and industrial parks, the power consumption patterns of different types of users and their relationships can be clearly revealed through time-frequency joint analysis. For example, it is found that the peak electricity consumption in industrial areas usually occurs during the day, while in residential areas, it reaches its peak in the evening and late at night; at the same time, it is also possible to identify instantaneous high load phenomena caused by special events (such as large exhibitions or sports events) in certain time periods. In order to better understand and utilize these time-frequency features, the obtained time-frequency feature tensor is subjected to asymmetric orthogonal decomposition to extract the load dynamic feature set. Asymmetric orthogonal decomposition is a technique for processing high-dimensional data. It can effectively reduce data redundancy and highlight the most important dynamic characteristics.The result is a more concise and highly representative load dynamic feature set, which provides a solid foundation for subsequent feature fusion processing. Subsequently, the load dynamic feature set is subjected to feature fusion processing by wavelet packet entropy calculation to obtain a multi-dimensional trend feature matrix. Wavelet packet entropy is an indicator of signal complexity, which can quantify the degree of uncertainty of signals in different frequency bands. In this process, by applying wavelet packet entropy calculation to the load dynamic feature set, not only can the subtle changes in the load data be captured, but also the information from different frequency bands can be effectively fused to form a multi-dimensional trend feature matrix. This matrix integrates the various features extracted in all previous steps, including but not limited to short-term fluctuations, medium-term trends, and long-term change laws of the load. For example, in the process of monitoring a city power grid, by constructing such a detailed multi-dimensional trend feature matrix, the changes in the load in each area can be monitored in real time, and corresponding adjustment measures can be made in time. Finally, in order to ensure that the extracted features are practical and useful for the management of the power system, it is necessary to perform adaptive threshold segmentation on the multi-dimensional trend feature matrix to obtain the final load change trend parameters. Adaptive threshold segmentation is a method that automatically determines the best segmentation point based on the characteristics of the data itself. It can dynamically adjust the threshold according to the current operating status to adapt to the changing load conditions. For example, during the high temperature period in summer, with the surge in the use of air conditioners, the power demand in a certain area may suddenly rise; at this time, through the adaptive threshold segmentation method, this abnormal change can be accurately identified from the multi-dimensional trend feature matrix, and corresponding scheduling strategies can be formulated accordingly, such as increasing the power generation capacity in advance or optimizing the grid configuration to ensure the stable operation of the entire system. In short, through the above series of complex but orderly steps, from wavelet transform decomposition to adaptive threshold segmentation, we can accurately extract the load change trend parameters from the load distribution feature matrix. These parameters not only reflect the specific pattern of load change over time, but also provide a scientific basis for the efficient management and optimization of smart grids. For example, when dealing with a complex power supply network with multiple user types, only through in-depth analysis and accurate prediction can we truly achieve a dynamic balance between supply and demand and provide a solid energy guarantee for the development of social economy.

[0034] In a specific embodiment, the supply-demand balance calculation is performed on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation scheme, including: Performing dynamic time series decomposition on the load change trend parameter to obtain a load power fluctuation sequence, and performing multi-dimensional phase compensation calculation on the load power fluctuation sequence to obtain a load power compensation feature group; Based on the load power compensation feature group, power distribution calculation is performed on the power supply system to obtain a power distribution feature matrix, and based on the power distribution feature matrix, topology optimization analysis is performed on the power supply system to obtain a power supply network topology feature set, wherein the power supply network topology feature set includes node connection weights, branch impedance distribution, and voltage gradient coefficient; Calculating the supply-demand matching degree of the power supply network topology feature set by a multi-objective constraint optimization method to obtain a supply-demand balance feature vector, and dynamically allocating weights to the supply-demand balance feature vector to obtain a power dispatch parameter group; Based on the power dispatch parameter group, the power supply system is subjected to regional load distribution calculation to obtain a regional power supply feature sequence, and the regional power supply feature sequence is subjected to adaptive boundary constraint processing to obtain a power supply boundary feature set, wherein the power supply boundary feature set includes a regional power limit, a voltage fluctuation range, and a load transfer threshold; A dynamic power supply scheme is generated for the power supply boundary feature set through a hierarchical progressive optimization method to obtain a dynamic power supply allocation scheme, wherein the dynamic power supply allocation scheme includes a power allocation strategy, a voltage regulation scheme and load balancing parameters.

[0035] Specifically, the process of calculating the supply and demand balance of the power supply system based on the load change trend parameters to obtain a dynamic power supply distribution plan is a highly complex and sophisticated step, which aims to ensure the efficient and stable operation of the power system through precise analysis and optimization. First, the load change trend parameters need to be dynamically decomposed into time series data that is easier to handle, namely, the load power fluctuation sequence. This process uses advanced mathematical models and technologies to accurately capture the changing laws of loads at different time scales. For example, in an urban power supply network that includes residential areas, commercial centers, and industrial parks, dynamic time series decomposition can identify daily, weekly, and even annual peak and trough periods of electricity consumption, as well as instantaneous high load phenomena caused by special events (such as holidays or large-scale events). Next, the load power fluctuation sequence is multi-dimensionally compensated to eliminate errors caused by signal delay or distortion, thereby obtaining a load power compensation feature group. Phase compensation is a technology for correcting the signal phase, which can help us understand and predict the load change pattern more accurately. In this process, considering the differences in electricity consumption between different types of users, a more balanced power distribution can be achieved by adjusting the phase relationship between regions. For example, during the peak period of electricity consumption on working days in industrial areas, the phase compensation technology can be used to reduce the impact on the peak period of electricity consumption in residential areas in the evening, thereby optimizing the load distribution of the entire power grid. Based on the load power compensation feature group, the power distribution of the power supply system is further calculated to obtain a power distribution feature matrix. This matrix not only contains the power demand information of each node, but also reflects the flow of electricity in the network. In order to improve the efficiency and reliability of power distribution, it is necessary to perform topological structure optimization analysis on the power supply system based on the power distribution feature matrix to obtain a power supply network topology feature set. Here, the power supply network topology feature set includes key parameters such as node connection weights, branch impedance distribution, and voltage gradient coefficient, which together determine the transmission efficiency and stability of electricity in the network. For example, when facing an urban power supply system composed of multiple user types, by optimizing the topological structure of the power supply network, the line loss can be effectively reduced, the quality of power transmission can be improved, and the overall reliability of the system can be enhanced. Subsequently, the supply and demand matching degree of the power supply network topology feature set is calculated using a multi-objective constraint optimization method to obtain a supply and demand balance feature vector. This method can find the optimal supply and demand balance point under the premise of meeting multiple constraints. For example, during the high temperatures of summer, with the surge in air conditioning usage, the electricity demand in a certain area may suddenly rise; at this time, through the multi-objective constraint optimization method, the most suitable supply and demand balance strategy can be found from the topological feature set, which not only guarantees the user's electricity demand, but also avoids the risk of overload of the power grid. Then, the dynamic weight allocation is performed on the supply and demand balance feature vector to obtain the power dispatch parameter group.This process dynamically adjusts the weights of various parameters according to the current actual operating status to adapt to the changing load conditions. For example, when dealing with sudden high-load events, the pressure can be relieved by increasing the input weight of the backup power supply or adjusting the load distribution weights of other regions. Based on the power dispatch parameter group, the power supply system is subjected to regional load distribution calculation to obtain a regional power supply feature sequence. This process divides the entire power supply network into several regions and formulates specific power supply strategies for each region. In order to ensure the effectiveness and feasibility of these strategies, it is also necessary to perform adaptive boundary constraint processing on the regional power supply feature sequence to obtain a power supply boundary feature set. The power supply boundary feature set includes parameters such as regional power limit, voltage fluctuation range and load transfer threshold, which provide clear guidelines for power dispatch within the region. For example, in the face of the possible power supply shortage under extreme weather conditions, the power supply plan of each region can be flexibly adjusted according to the pre-set regional power limit and voltage fluctuation range to ensure the normal operation of key facilities. Finally, the power supply boundary feature set is dynamically generated by the hierarchical progressive optimization method, and a dynamic power supply distribution plan is finally obtained. This optimization method regards the entire power supply network as a multi-level system, and gradually refines the optimization strategy from macro to micro to achieve global optimization. For example, in an urban power supply system, the overall power allocation strategy can be determined first, and then the voltage regulation scheme and load balancing parameters can be optimized according to the specific conditions of each region. The result is a dynamic power supply allocation scheme that takes into account both short-term emergency response and long-term development planning. It not only improves the flexibility and adaptability of the power system, but also lays a solid foundation for the efficient management of smart grids. In short, through the above series of complex but orderly steps, from wavelet transform decomposition to the final generation of dynamic power supply allocation schemes, we can effectively improve the operating efficiency and reliability of the power system and ensure that the energy needs of social and economic development are fully guaranteed.

[0036] In a specific embodiment, the power distribution calculation of the power supply system is performed based on the load power compensation feature group to obtain a power distribution feature matrix, including: Performing multi-dimensional space mapping decomposition on the load power compensation feature group to obtain a power compensation vector set, and performing phase difference calculation on the power compensation vector set to obtain a voltage-power coupling matrix; Based on the voltage-power coupling matrix, the power supply system is tracked for node power flow to obtain a power distribution map, and a topological sensitivity analysis is performed on the power distribution map to obtain a power supply network transmission characteristic vector; The power supply network transmission characteristic vector is subjected to power partition clustering by a double-layer recursive partitioning method to obtain a power sub-network set, and the interconnection coupling degree of the power sub-network set is calculated to obtain an inter-regional power exchange matrix; Based on the inter-regional power exchange matrix, a hierarchical coordination optimization calculation is performed on the power supply system to obtain a multi-level power supply coordination parameter group, and a robustness evaluation is performed on the multi-level power supply coordination parameter group to obtain a power distribution elasticity coefficient set; The power distribution elasticity coefficient set is dynamically integrated through an adaptive weight fusion mechanism to obtain a power distribution feature matrix, wherein the power distribution feature matrix includes node power distribution ratios, voltage control strategy parameters and load balancing adjustment data.

[0037] Specifically, the process of calculating the power distribution of the power supply system based on the load power compensation feature group to obtain the power distribution feature matrix is ​​a complex and meticulous step, which aims to ensure the efficient and stable operation of the power system through precise data processing and optimization algorithms. First, the load power compensation feature group is decomposed by multidimensional space mapping, which is the first step of the whole process. At this stage, the load power compensation feature group is converted into a multidimensional spatial representation, namely a power compensation vector set. Each compensation vector represents a power adjustment strategy under a specific dimension, such as a load adjustment scheme for different regions or different types of users. Then, by calculating the phase difference of these power compensation vector sets, a voltage-power coupling matrix can be obtained. The phase difference calculation helps to understand the mutual influence relationship between voltage and power between different nodes, which is crucial for the subsequent power distribution analysis. Based on the voltage-power coupling matrix, the next step is to track the node power flow of the power supply system to construct a detailed power distribution map. This step involves in-depth analysis of the flow of power between nodes in the power grid and how they interact with each other. For example, in an urban power supply network that includes residential areas, commercial centers, and industrial parks, by tracing the power flow of nodes, we can clearly see the power exchange pattern between regions, including the power flow during peak hours and its impact on the overall power grid. In order to further optimize this power distribution, it is necessary to perform topological sensitivity analysis on the power distribution map to obtain the power supply network transmission characteristic vector. Topological sensitivity analysis can reveal the impact of the power grid structure on the efficiency and stability of power transmission, and help identify potential bottlenecks or vulnerabilities, such as the overload risk of certain key transmission lines. Subsequently, a two-layer recursive partitioning method is used to perform power partitioning clustering on the power supply network transmission characteristic vector to form a set of power subnetworks. This method recursively divides the power grid into smaller, easier-to-manage subnetworks, each of which has relatively independent but interrelated power transmission characteristics. For example, in the above-mentioned urban power supply network case, the power grid can be divided into multiple subnetworks in this way, each of which focuses on meeting the power demand in a specific area (such as an industrial area, commercial center, or residential area). Then, the interconnection coupling degree of the power subnetwork set is calculated to obtain the inter-regional power exchange matrix. This step aims to quantify the amount of electric energy exchange between different sub-networks and the degree of interdependence between them, which is very important for formulating effective cross-regional coordination strategies. Based on the inter-regional power exchange matrix, the power supply system is further layered and coordinated to optimize the calculation, and a multi-level power supply coordination parameter group is obtained. Here, "multi-level" refers to an optimization strategy that is gradually refined from macro to micro to ensure that the power distribution at each level can reach the optimal state. For example, in the face of the peak electricity consumption that may occur during the high temperature period in summer, multi-level coordination and optimization can be used to ensure the stability of the overall power grid and flexibly respond to high load demands in local areas.In addition, it is necessary to perform robustness evaluation on the multi-level power supply coordination parameter group to obtain a set of power distribution elasticity coefficients. Robustness evaluation is a method to test the performance of the scheme under various uncertain conditions, which ensures that basic power supply and service quality can be maintained even in extreme situations. Finally, the power distribution elasticity coefficient set is dynamically integrated through the adaptive weight fusion mechanism to finally obtain the power distribution feature matrix. The adaptive weight fusion mechanism allows the weights of various parameters to be dynamically adjusted according to real-time data to adapt to the changing load conditions and external environment. For example, in an urban power supply system composed of multiple user types, the power distribution ratio, voltage control strategy parameters and load balancing adjustment data of each region can be flexibly adjusted according to actual power demand, weather forecast and other information. The result is a power distribution feature matrix that can not only reflect the current power grid operation status but also guide future operations. It not only contains specific numerical parameters, but also provides detailed guidelines on how to optimize power grid performance. In short, through this series of complex steps, from wavelet transform decomposition to the final power distribution feature matrix generation, we can effectively improve the flexibility and reliability of the power system, ensure that the energy needs of social and economic development are fully guaranteed, and provide strong support for the efficient management of smart grids.

[0038] In a specific embodiment, the power flow analysis method is a Newton-Raphson power flow iteration method. The dynamic power supply distribution scheme is evaluated and adjusted for stability by a preset power flow analysis method to obtain a safe and stable power supply distribution scheme, including: The nonlinear power flow calculation of the dynamic power supply distribution scheme is performed by the Newton-Raphson power flow iteration method to obtain a system node voltage distribution feature set, and a power flow sensitivity analysis is performed on the system node voltage distribution feature set to obtain a power grid power transmission characteristic matrix; When the minimum singular value of the power transmission characteristic matrix of the power grid is less than a preset threshold, a continuous power flow tracking analysis is performed on the power supply system based on the power transmission characteristic matrix of the power grid to obtain a system voltage stability margin curve, and critical points are identified on the system voltage stability margin curve to obtain a characteristic map of the vulnerable area of ​​the power grid; The power control scheme is optimized by using the AC power flow optimization analysis method to optimize the characteristic map of the vulnerable area of ​​the power grid, so as to obtain a safe and stable power supply distribution scheme, wherein the safe and stable power supply distribution scheme includes an optimal power distribution strategy, a dynamic reactive power compensation configuration and a voltage regulation control coefficient.

[0039] Specifically, in realizing the flexible power supply distribution method based on load forecasting, the dynamic power supply distribution scheme is evaluated and adjusted for stability through a preset power flow analysis method to obtain a safe and stable power supply distribution scheme. This process is crucial to ensure the stable operation of the power grid. Specifically, the power flow analysis method adopts the Newton-Raphson power flow iteration method, which is an efficient and widely used method for solving complex nonlinear equations to calculate the node voltage distribution and power transmission characteristics in the power grid.

[0040] First, the Newton-Raphson power flow iteration method is used to perform nonlinear power flow calculation on the dynamic power supply distribution scheme to obtain the system node voltage distribution feature set. The Newton-Raphson method is an iterative solution technique that solves a nonlinear set of equations by continuously approaching the true solution. This method is particularly suitable for complex calculations of large-scale power systems. For example, in an urban power supply network that includes residential areas, commercial centers, and industrial parks, there are significant differences in power consumption patterns in different areas, which will lead to different voltage levels at each node. Through detailed calculations of the voltage distribution of these nodes, potential voltage instability points can be identified and provide a basis for subsequent optimization. Next, the power flow sensitivity analysis of the system node voltage distribution feature set is performed to obtain the power transmission characteristic matrix of the power grid. The power flow sensitivity analysis aims to evaluate the degree of mutual influence between the nodes in the power grid, especially when the load or power generation of a node changes, how to affect the voltage and power distribution of the entire power grid. In this process, by calculating the power transmission characteristic matrix of the power grid, the coupling relationship between different nodes and their impact on the stability of the overall power grid can be quantified. For example, in the above-mentioned example of the urban power supply network, if the load rate of a key transmission line is close to its maximum capacity, the line may become a vulnerable point of the system. By analyzing the power transmission characteristic matrix of the power grid, these vulnerable points can be accurately determined and their impact on the stability of the power grid can be evaluated. When the minimum singular value of the power transmission characteristic matrix of the power grid is less than the preset threshold, it means that the power grid may have voltage stability problems and needs further analysis and optimization. In this case, based on the power transmission characteristic matrix of the power grid, a continuous power flow tracking analysis is performed on the power supply system to obtain the system voltage stability margin curve. Continuous power flow tracking is a process that simulates the gradual transition of the power grid from a normal state to an unstable state. In this way, the turning point of the system from stability to instability can be identified. For example, during the high temperature period in summer, with the surge in the use of air conditioners, the power demand in some areas may suddenly rise; through continuous power flow tracking analysis, we can foresee the impact of this high load situation on the power grid and formulate corresponding response strategies. Subsequently, the critical point identification of the system voltage stability margin curve is performed to obtain the characteristic map of the vulnerable area of ​​the power grid. Critical point identification helps to find the areas in the system that are most prone to voltage collapse, thereby providing guidance for targeted optimization measures. In order to improve the safety and stability of the power grid, the AC power flow optimization analysis method is used to optimize the power control scheme of the characteristic map of the vulnerable area of ​​the power grid, and finally a safe and stable power supply distribution plan is obtained. AC power flow optimization analysis is an optimization method that comprehensively considers multiple objective functions and constraints, aiming to find a solution that can satisfy all constraints and achieve the optimal solution of the objective function. For example, in the above-mentioned urban power supply network case, the AC power flow optimization analysis can be used to ensure the basic power demand of each area and avoid the risk of overload of key transmission lines.At the same time, it is also necessary to ensure that the overall voltage level of the power grid is within a reasonable range and improve the safety of the system. The result of this is a power distribution scheme that includes an optimal power distribution strategy, a dynamic reactive compensation configuration, and a voltage regulation control coefficient. The optimal power distribution strategy aims to dynamically adjust the output power of the generator set according to the real-time load changes to ensure the balance of supply and demand. The dynamic reactive compensation configuration improves the voltage level and power factor of the power grid and reduces energy loss by intelligently adjusting the working state of reactive compensation equipment (such as capacitors and reactors). The voltage regulation control coefficient is used to set the parameters of the automatic voltage regulator (AVR) to ensure that the grid voltage remains within the specified range and prevent the voltage from being too high or too low to affect the user's power experience. In short, through this series of complex but orderly steps, from nonlinear power flow calculation to the final generation of a safe and stable power distribution scheme, we can effectively improve the flexibility and adaptability of the power system, ensure that the energy needs of social and economic development are fully guaranteed, and provide strong support for the efficient management of smart grids. For example, in the face of the power supply shortage that may occur under extreme weather conditions, the above method can be used to predict in advance and take corresponding measures to ensure that the power grid can still operate stably under various uncertain conditions.

[0041] In a specific embodiment, based on the safe and stable power supply allocation scheme, the dynamic power supply allocation scheme is corrected and calculated in real time to obtain a flexible power supply control parameter group, including: Performing high-order wavelet singular spectrum analysis on the safe and stable power supply distribution scheme to obtain a power supply scheme feature vector sequence, and performing phase compensation calculation on the power supply scheme feature vector sequence to obtain a real-time power supply control reference parameter group, wherein the real-time power supply control reference parameter group includes a power adjustment step coefficient, a node voltage control gain, and a load response time constant; Based on the real-time power supply control reference parameter group, the dynamic power supply allocation scheme is subjected to orthogonal decomposition processing to obtain a power supply scheme difference feature matrix, and multi-scale entropy value calculation is performed on the power supply scheme difference feature matrix to obtain a power supply correction quantization parameter set; The time-space domain features of the power supply correction quantization parameter set are fused through nonlinear adaptive filtering to obtain a power supply dynamic response feature sequence, and the power supply dynamic response feature sequence is subjected to fuzzy reasoning optimization processing to obtain a flexible power supply control parameter group, wherein the flexible power supply control parameter group includes real-time power allocation instructions, dynamic voltage regulation strategies and load response control signals.

[0042] Specifically, based on the safe and stable power supply distribution scheme, the process of performing real-time correction calculation on the dynamic power supply distribution scheme to obtain a flexible power supply control parameter group is intended to ensure the efficient and stable operation of the power system through precise data processing and advanced optimization algorithms. First, it is necessary to perform high-order wavelet singular spectrum analysis on the safe and stable power supply distribution scheme. This process converts the complex power supply distribution scheme into a feature vector sequence containing multi-scale information. High-order wavelet singular spectrum analysis is a powerful tool that combines wavelet transform and singular value decomposition. It can extract key features from raw data. These feature vector sequences reflect the changes in power supply characteristics on different time scales. For example, in a typical urban power supply network, there may be peak and valley periods of electricity consumption on a daily, weekly or even annual basis; through high-order wavelet singular spectrum analysis, we can identify and separate the change characteristics on these different time scales, so as to better understand and predict the behavior of the power grid. Next, the phase compensation calculation is performed on the feature vector sequence of the power supply scheme to eliminate the error caused by signal delay or distortion, so as to obtain a real-time power supply control reference parameter group. Phase compensation calculation helps to adjust the voltage and power relationship between nodes to ensure the synchronization and stability of the entire power grid. For example, during the peak power consumption period on working days in industrial areas, the phase compensation technology can be used to reduce the impact on the peak power consumption period in residential areas at night, thereby optimizing the load distribution of the entire power grid. The real-time power supply control benchmark parameter group includes key parameters such as power adjustment step coefficient, node voltage control gain and load response time constant, which together determine the control accuracy and response speed of the power system in actual operation. Based on the real-time power supply control benchmark parameter group, the dynamic power supply allocation scheme is further orthogonally decomposed to obtain the power supply scheme difference feature matrix. Orthogonal decomposition is a technique for simplifying complex data structures, which can help us understand the main difference components in the power supply scheme more clearly. For example, when facing the peak power consumption that may occur during the high temperature period in summer, by orthogonally decomposing the dynamic power supply allocation scheme, it is possible to identify which areas or user groups are the main factors causing load fluctuations, and formulate targeted control strategies accordingly. Subsequently, the power supply scheme difference feature matrix is ​​multi-scaled entropy value calculated to obtain a power supply correction quantization parameter set. Multi-scale entropy value calculation is a method to measure signal complexity, which can quantify the degree of uncertainty of signals in different frequency bands, thereby helping us to more accurately evaluate the effectiveness and stability of power supply schemes. In order to further improve the adaptability and flexibility of the power supply scheme, nonlinear adaptive filtering is used to fuse the time and space domain features of the power supply correction quantization parameter set to obtain the power supply dynamic response feature sequence. Nonlinear adaptive filtering is a method that can automatically adjust its parameters to adapt to the changing characteristics of the input signal. This step not only captures the changing laws of the load in different time and space dimensions, but also effectively integrates information from different frequency bands.For example, in the process of monitoring a city power grid, through nonlinear adaptive filtering, the load changes in each area can be monitored in real time, and corresponding adjustment measures can be made in time. Then, the power supply dynamic response feature sequence is subjected to fuzzy reasoning optimization processing, and finally a flexible power supply control parameter group is obtained. Fuzzy reasoning is a technology that simulates the human decision-making process, which can make reasonable judgments under uncertainty and incomplete information. In this process, by applying fuzzy reasoning optimization processing to the power supply dynamic response feature sequence, various control parameters can be dynamically adjusted according to the current actual operating status, thereby achieving more flexible and efficient power dispatch. The flexible power supply control parameter group includes key components such as real-time power allocation instructions, dynamic voltage regulation strategies, and load response control signals, which together constitute a specific guide for guiding the operation of the power system. For example, in response to the upcoming holiday peak period, the possible high load situation can be predicted in advance, and the start-up time and location of the backup power supply can be arranged accordingly, and the output power of the generator set can be flexibly adjusted through the real-time power allocation instruction; the dynamic voltage regulation strategy is used to ensure that the grid voltage level is within a reasonable range to avoid the impact of excessively high or low voltage on the user's power experience; finally, the load response control signal is used to guide users to consciously participate in peak-shifting power consumption, thereby effectively alleviating the pressure on the grid. In summary, through the above series of complex but orderly steps, from wavelet singular spectrum analysis to the final generation of flexible power supply control parameter groups, we can effectively improve the flexibility and adaptability of the power system, ensure that the energy needs of social and economic development are fully guaranteed, and provide strong support for the efficient management of smart grids.

[0043] The flexible power supply distribution method based on load prediction in the embodiment of the present invention is described above. The flexible power supply distribution system based on load prediction in the embodiment of the present invention is described below. Figure 2 In one embodiment of the present invention, a flexible power supply distribution system based on load prediction includes: An extraction module 21 is used to perform multi-dimensional feature extraction on the historical load data of the power supply system to obtain a load distribution feature matrix; An analysis module 22 is used to perform multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters; A calculation module 23 is used to predict whether there is a supply-demand imbalance in the power supply system based on the load change trend parameter, and if so, perform supply-demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation plan; The adjustment module 24 is used to evaluate and adjust the stability of the dynamic power distribution scheme through a preset power flow analysis method to obtain a safe and stable power distribution scheme; The correction module 25 is used to perform real-time correction calculation on the dynamic power supply allocation scheme based on the safe and stable power supply allocation scheme, obtain a flexible power supply control parameter group, and control the power supply system based on the flexible power supply control parameter group.

[0044] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the description in the above method embodiment, which will not be repeated here.

[0045] Reference Figure 3 The present invention also provides a computer device in an embodiment, wherein the internal structure of the computer device can be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0046] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0047] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0048] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0049] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0050] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A flexible power supply distribution method based on load prediction, characterized in that: Applied to the power supply system, including the following steps: Performing multi-dimensional feature extraction on the historical load data of the power supply system to obtain a load distribution feature matrix; Performing multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters; Predicting whether there is a supply-demand imbalance in the power supply system based on the load change trend parameter, and if so, performing supply-demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation plan; By using a preset power flow analysis method, the dynamic power supply distribution scheme is evaluated for stability and adjusted to obtain a safe and stable power supply distribution scheme; Based on the safe and stable power supply allocation scheme, the dynamic power supply allocation scheme is corrected and calculated in real time to obtain a flexible power supply control parameter group, and the power supply system is controlled based on the flexible power supply control parameter group.

2. The method for flexible power supply distribution based on load prediction according to claim 1 is characterized in that: The multi-dimensional feature extraction is performed on the historical load data of the power supply system to obtain a load distribution feature matrix, including: Sampling the historical load data of the power supply system in different time periods to obtain a load time series data set, and performing spectrum decomposition processing on the load time series data set to obtain a load frequency domain feature sequence; Based on the load frequency domain feature sequence, correlation analysis is performed on the power load in the power supply system to obtain a load correlation feature matrix, and singular value decomposition is performed on the load correlation feature matrix to obtain a load principal component feature set, wherein the load principal component feature set includes a load change gradient, a power factor distribution, and a load density coefficient; The instantaneous features of the load principal component feature set are extracted by Hilbert transform to obtain the load instantaneous feature vector, and the load instantaneous feature vector is combined in multiple dimensions to obtain a load distribution feature matrix, wherein the load distribution feature matrix includes power fluctuation features, load distribution density features and power consumption cycle features.

3. The method for flexible power supply distribution based on load prediction according to claim 1 is characterized in that: The multi-scale analysis of the load distribution characteristic matrix is ​​performed by wavelet transform decomposition to obtain load change trend parameters, including: Performing multi-resolution wavelet decomposition on the load distribution characteristic matrix through wavelet transform decomposition to obtain a multi-scale characteristic coefficient matrix, and performing singular spectrum analysis on the multi-scale characteristic coefficient matrix to obtain a load fluctuation characteristic vector; Based on the load fluctuation feature vector, a time-frequency joint analysis is performed on the power supply system to obtain a time-frequency feature tensor, and an asymmetric orthogonal decomposition is performed on the time-frequency feature tensor to obtain a load dynamic feature set; The load dynamic feature set is subjected to feature fusion processing by wavelet packet entropy value calculation to obtain a multi-dimensional trend feature matrix, and the multi-dimensional trend feature matrix is ​​subjected to adaptive threshold segmentation to obtain load change trend parameters.

4. The method for flexible power supply distribution based on load prediction according to claim 1, characterized in that: The step of performing supply and demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation scheme includes: Performing dynamic time series decomposition on the load change trend parameter to obtain a load power fluctuation sequence, and performing multi-dimensional phase compensation calculation on the load power fluctuation sequence to obtain a load power compensation feature group; Based on the load power compensation feature group, power distribution calculation is performed on the power supply system to obtain a power distribution feature matrix, and based on the power distribution feature matrix, topology optimization analysis is performed on the power supply system to obtain a power supply network topology feature set, wherein the power supply network topology feature set includes node connection weights, branch impedance distribution, and voltage gradient coefficient; Calculating the supply-demand matching degree of the power supply network topology feature set by a multi-objective constraint optimization method to obtain a supply-demand balance feature vector, and dynamically allocating weights to the supply-demand balance feature vector to obtain a power dispatch parameter group; Based on the power dispatch parameter group, the power supply system is subjected to regional load distribution calculation to obtain a regional power supply feature sequence, and the regional power supply feature sequence is subjected to adaptive boundary constraint processing to obtain a power supply boundary feature set, wherein the power supply boundary feature set includes a regional power limit, a voltage fluctuation range, and a load transfer threshold; A dynamic power supply scheme is generated for the power supply boundary feature set through a hierarchical progressive optimization method to obtain a dynamic power supply allocation scheme, wherein the dynamic power supply allocation scheme includes a power allocation strategy, a voltage regulation scheme and load balancing parameters.

5. The method for flexible power supply distribution based on load prediction according to claim 4 is characterized in that: The performing power distribution calculation on the power supply system based on the load power compensation feature group to obtain a power distribution feature matrix includes: Performing multi-dimensional space mapping decomposition on the load power compensation feature group to obtain a power compensation vector set, and performing phase difference calculation on the power compensation vector set to obtain a voltage-power coupling matrix; Based on the voltage-power coupling matrix, the power supply system is tracked for node power flow to obtain a power distribution map, and a topological sensitivity analysis is performed on the power distribution map to obtain a power supply network transmission characteristic vector; The power supply network transmission characteristic vector is subjected to power partition clustering by a double-layer recursive partitioning method to obtain a power sub-network set, and the interconnection coupling degree of the power sub-network set is calculated to obtain an inter-regional power exchange matrix; Based on the inter-regional power exchange matrix, a hierarchical coordination optimization calculation is performed on the power supply system to obtain a multi-level power supply coordination parameter group, and a robustness evaluation is performed on the multi-level power supply coordination parameter group to obtain a power distribution elasticity coefficient set; The power distribution elasticity coefficient set is dynamically integrated through an adaptive weight fusion mechanism to obtain a power distribution feature matrix, wherein the power distribution feature matrix includes node power distribution ratios, voltage control strategy parameters and load balancing adjustment data.

6. The method for flexible power supply distribution based on load prediction according to claim 1, characterized in that: The power flow analysis method is a Newton-Raphson power flow iteration method. The dynamic power supply distribution scheme is evaluated and adjusted for stability by a preset power flow analysis method to obtain a safe and stable power supply distribution scheme, including: The nonlinear power flow calculation of the dynamic power supply distribution scheme is performed by the Newton-Raphson power flow iteration method to obtain a system node voltage distribution feature set, and a power flow sensitivity analysis is performed on the system node voltage distribution feature set to obtain a power grid power transmission characteristic matrix; When the minimum singular value of the power transmission characteristic matrix of the power grid is less than a preset threshold, a continuous power flow tracking analysis is performed on the power supply system based on the power transmission characteristic matrix of the power grid to obtain a system voltage stability margin curve, and critical points are identified on the system voltage stability margin curve to obtain a characteristic map of the vulnerable area of ​​the power grid; The power control scheme is optimized by using the AC power flow optimization analysis method to optimize the characteristic map of the vulnerable area of ​​the power grid, so as to obtain a safe and stable power supply distribution scheme, wherein the safe and stable power supply distribution scheme includes an optimal power distribution strategy, a dynamic reactive power compensation configuration and a voltage regulation control coefficient.

7. The method for flexible power supply distribution based on load prediction according to claim 1, characterized in that: The method of performing real-time correction calculation on the dynamic power supply allocation scheme based on the safe and stable power supply allocation scheme to obtain a flexible power supply control parameter group includes: Performing high-order wavelet singular spectrum analysis on the safe and stable power supply distribution scheme to obtain a power supply scheme feature vector sequence, and performing phase compensation calculation on the power supply scheme feature vector sequence to obtain a real-time power supply control reference parameter group, wherein the real-time power supply control reference parameter group includes a power adjustment step coefficient, a node voltage control gain, and a load response time constant; Based on the real-time power supply control reference parameter group, the dynamic power supply allocation scheme is subjected to orthogonal decomposition processing to obtain a power supply scheme difference feature matrix, and multi-scale entropy value calculation is performed on the power supply scheme difference feature matrix to obtain a power supply correction quantization parameter set; The time-space domain features of the power supply correction quantization parameter set are fused through nonlinear adaptive filtering to obtain a power supply dynamic response feature sequence, and the power supply dynamic response feature sequence is subjected to fuzzy reasoning optimization processing to obtain a flexible power supply control parameter group, wherein the flexible power supply control parameter group includes real-time power allocation instructions, dynamic voltage regulation strategies and load response control signals.

8. A flexible power supply distribution system based on load prediction, characterized in that: Applied to power supply systems, including: An extraction module, used for performing multi-dimensional feature extraction on the historical load data of the power supply system to obtain a load distribution feature matrix; An analysis module, used for performing multi-scale analysis on the load distribution characteristic matrix through wavelet transform decomposition to obtain load change trend parameters; A calculation module, used to predict whether there is a supply-demand imbalance in the power supply system based on the load change trend parameter, and if so, perform supply-demand balance calculation on the power supply system based on the load change trend parameter to obtain a dynamic power supply allocation plan; An adjustment module, used to evaluate and adjust the stability of the dynamic power supply distribution scheme through a preset power flow analysis method to obtain a safe and stable power supply distribution scheme; A correction module is used to perform real-time correction calculation on the dynamic power supply allocation scheme based on the safe and stable power supply allocation scheme, obtain a flexible power supply control parameter group, and control the power supply system based on the flexible power supply control parameter group.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images