Electric load model modeling method and system based on time-sharing electric quantity

Abstract

The invention relates to the technical field of electrical load analysis, and discloses an electrical load model modeling method and system based on time-sharing electric quantity. The method comprises the following steps: based on intelligent electric meter time-sharing electric quantity acquisition and equipment start-stop frequency identification, extracting peak-valley normal-period basic load characteristics by adopting a time-sharing electric quantity empirical mode decomposition algorithm, and obtaining equipment start-stop load correlation characteristics by combining a time-sharing variational mode decomposition algorithm with an equipment start-stop state matrix; a time-sharing hyper-parameter optimization algorithm is used to carry out load prediction in each time period, and finally a start-stop frequency influence weight mechanism is established through a time-sharing load state fusion algorithm to obtain a time-sharing electric quantity load modeling result. The method solves the problems that an existing time-sharing electric quantity load modeling method lacks equipment start-stop state sensing, modeling precision between time periods is unbalanced, and time period switching continuity is poor.

Description

Technical Field

[0001] The present application relates to the technical field of power load analysis, and in particular to a method and system for modeling a power load model based on time-sharing power consumption. Background Art

[0002] Currently, electricity load modeling methods based on time-of-use electricity consumption are widely used in fields such as power system planning, demand response, and smart grid management. Existing technologies primarily rely on historical time-of-use electricity consumption data for modeling, using time series analysis methods such as ARIMA models, regression analysis, or machine learning algorithms like neural networks to predict load. These methods divide the 24-hour day into different time periods, collect actual electricity consumption data for each period, and then use this historical time-of-use electricity consumption data to identify electricity consumption patterns and build mathematical models to predict future electricity load. Traditional modeling methods typically use a unified algorithm framework when processing time-of-use electricity consumption data, applying the same processing strategy to data from each time period.

[0003] However, existing technologies have significant technical defects. First, traditional methods mainly rely on historical time-of-use electricity data for modeling, and fail to take the operating status characteristics of residents' main household electrical appliances as key input variables for modeling, resulting in the model being unable to effectively identify and quantify the specific impact and impact pattern of the start-stop behavior of high-power household appliances on time-of-use load fluctuations. Secondly, existing static modeling methods are difficult to accurately capture the superposition effect and mutual influence relationship of power demand when multiple high-power household appliances are running simultaneously or alternately, especially the dynamic coupling relationship in the scenario of concurrent use of multiple devices during peak power consumption periods. In addition, when residents' electricity consumption habits change seasonally, their lifestyles are adjusted, or new electrical appliances are added, the existing models lack a mechanism to quickly identify changes and perform adaptive parameter adjustments, resulting in a significant decrease in prediction accuracy.

[0004] Due to the lack of real-time perception of the start and stop status of household appliances, traditional modeling methods are unable to establish an accurate correlation between the start and stop behavior of equipment and time-of-use load fluctuations, which leads to significant differences in load forecasting accuracy in different time periods (peak, flat, and valley). In particular, when there are large differences in the start and stop frequencies of equipment between time periods, a single modeling strategy is difficult to meet the accuracy requirements of each time period at the same time. At the same time, the results of independent predictions for each time period often have load jumps and logical inconsistencies at the time period switching points, which seriously affects the overall modeling accuracy and practicality of the time-of-use electricity load model. Summary of the Invention

[0005] The present application provides a method and system for modeling an electricity load model based on time-sharing electricity, which is used to solve the problems in existing time-sharing electricity load modeling methods, such as lack of equipment start and stop status perception, uneven modeling accuracy between time periods, and poor continuity in time period switching.

[0006] In the first aspect, the present application provides a method for modeling an electricity load model based on time-sharing electricity quantity, and the method for modeling an electricity load model based on time-sharing electricity quantity includes: collecting time-sharing electricity quantity of residential users through smart meters to obtain a time-sharing electricity quantity sequence, and at the same time identifying the start and stop frequency of household appliances through an equipment monitoring module to obtain an equipment start and stop state matrix; according to the time-sharing electricity quantity sequence, using a time-sharing electricity quantity empirical mode decomposition algorithm to perform differential decomposition processing on the data of each time period to obtain the basic load characteristics of peak, valley and normal periods; using a time-sharing variational mode decomposition algorithm to identify the start and stop fluctuation signal of the basic load characteristics, and combining the equipment start and stop state matrix to obtain the equipment start and stop load correlation characteristics; according to the equipment start and stop load correlation characteristics, using a time-sharing hyperparameter optimization algorithm to perform load forecasting processing on the start and stop frequency of each time period respectively to obtain a time period start and stop load forecast result; integrating the start and stop load forecast results of the time period through a time-sharing load state fusion algorithm, establishing a start and stop frequency influence weight mechanism, and obtaining a time-sharing electricity load modeling result.

[0007] In a second aspect, the present application provides a system for modeling a power load model based on time-sharing power consumption, the system comprising: The acquisition module is used to collect the time-sharing power consumption of residential users through smart meters to obtain the time-sharing power sequence. At the same time, the device monitoring module is used to identify the start and stop frequency of household appliances to obtain the device start and stop state matrix; A decomposition module is used to perform differential decomposition processing on the data of each time period according to the time-sharing power sequence using the time-sharing power empirical mode decomposition algorithm to obtain the basic load characteristics of peak, valley and normal periods; an identification module for performing start-stop fluctuation signal identification processing on the basic load characteristics through a time-varying mode decomposition algorithm, and combining the equipment start-stop state matrix to obtain equipment start-stop load correlation characteristics; A prediction module is used to perform load prediction processing on the start-stop frequency of each time period based on the start-stop load correlation characteristics of the equipment using a time-sharing hyperparameter optimization algorithm to obtain a start-stop load prediction result for each time period; The integration module is used to integrate the start-stop load prediction results of the period through the time-sharing load state fusion algorithm, establish a start-stop frequency influence weight mechanism, and obtain the time-sharing power load modeling results.

[0008] In a third aspect, a device for modeling an electricity load model based on time-sharing electricity is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the device for modeling an electricity load model based on time-sharing electricity executes the above-mentioned method for modeling an electricity load model based on time-sharing electricity.

[0009] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned method for modeling a power load model based on time-sharing power consumption.

[0010] In the technical solution provided by the present application, the combination of time-sharing electricity collection of smart meters and start-stop frequency identification of the equipment monitoring module establishes a direct correlation between time-sharing electricity data and the operating status of household appliances, solving the technical defect that traditional methods rely solely on historical electricity data and cannot identify the impact of equipment behavior. The differentiated decomposition processing of the time-sharing electricity empirical mode decomposition algorithm sets adaptive parameters for the differences in load characteristics between peak, valley and normal periods, effectively extracts the basic load characteristics of each time period, and lays the foundation for subsequent precise modeling. The time-sharing variational mode decomposition algorithm identifies start-stop fluctuation signals by combining the equipment start-stop state matrix, establishes an accurate mapping relationship between equipment start-stop behavior and load changes, and significantly enhances the model's perception of equipment behavior changes. The time-sharing hyperparameter optimization algorithm constructs differentiated network structures for the start-stop frequency characteristics of each time period and performs independent parameter optimization, avoiding the problem that a single model is difficult to adapt to the complexity differences of each time period at the same time, and greatly improves the accuracy and stability of load forecasting in each time period. The time-sharing load state fusion algorithm realizes the organic integration of the prediction results of each time period by establishing a start-stop frequency influence weight mechanism, which not only maintains the load continuity between time periods, but also ensures the logical consistency of the fusion results with the actual start-stop behavior of the equipment, thereby obtaining high-precision and physically reasonable time-sharing power load modeling results.

[0011] The time-sharing power empirical mode decomposition algorithm can accurately separate transient load changes caused by the start and stop of household appliances from the basic power load through adaptive noise enhancement and differentiated mode settings. This separation capability is of great significance for understanding the laws of residents' electricity consumption behavior. The innovative design of the time-sharing variational mode decomposition algorithm combined with the device state constraint enables the algorithm to accurately identify the independent contribution of each device to the load in a complex environment with multiple devices starting and stopping concurrently, which is key to achieving refined electricity management and demand response. The time-sharing hyperparameter optimization algorithm adopts a differentiated network configuration strategy for device usage patterns in different time periods, making full use of the inherent characteristics of complex device start and stop during peak periods, relative stability during normal periods, and simplicity during valley periods. Through the reasonable matching of deep, mid-level, and shallow networks, it significantly reduces the computational complexity while ensuring prediction accuracy. The device state transfer matrix and continuity constraint mechanism in the time-sharing load state fusion algorithm effectively solve the problem of smooth transition of residents' electricity consumption behavior during time period switching. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 This is a schematic diagram of an embodiment of a method for modeling a power load model based on time-sharing power consumption in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of a system for modeling a power load model based on time-sharing power consumption in an embodiment of the present application; Figure 3 It is a schematic block diagram of the structure of a device for modeling a power load model based on time-sharing power in an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a method and system for modeling an electricity load model based on time-sharing electricity consumption. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0015] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In one embodiment of the present application, an embodiment of a method for modeling a power load model based on time-sharing power consumption includes: Step S101: Time-sharing electricity consumption of residential users is collected through smart meters to obtain a time-sharing electricity sequence. At the same time, the start and stop frequencies of household appliances are identified through the device monitoring module to obtain a device start and stop state matrix. Step S102: Based on the time-sharing electricity sequence, the time-sharing electricity empirical mode decomposition algorithm is used to perform differential decomposition processing on the data of each time period to obtain the basic load characteristics of peak, valley and normal periods; Step S103: Using the time-varying modal decomposition algorithm to identify the start-stop fluctuation signal of the basic load characteristics, and combining it with the equipment start-stop state matrix to obtain the equipment start-stop load correlation characteristics; Step S104: Based on the equipment start-stop load correlation characteristics, a time-sharing hyperparameter optimization algorithm is used to perform load forecasting on the start-stop frequency of each time period to obtain a time period start-stop load forecast result; Step S105: Integrate the start-stop load forecast results of the time period through the time-sharing load state fusion algorithm, establish a start-stop frequency impact weight mechanism, and obtain the time-sharing power load modeling result.

[0016] It is understandable that the execution subject of this application can be a power load modeling system based on time-sharing power consumption, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0017] Specifically, a 24-hour period is divided into three intervals: peak, average, and off-peak periods, according to the time-of-use electricity pricing policy. Smart meters sample residential electricity consumption at 15-minute intervals, generating 96 time-of-use electricity data points. These data points are arranged and combined in a time series to form a time-of-use electricity sequence that reflects peak, off-peak, and average load variations. Furthermore, power thresholds are set to filter high-power household appliances such as air conditioners, water heaters, and washing machines, creating a list of target monitoring devices. Current sensors monitor the start and stop status of each device on the list, recording the number of times the device starts and stops. This monitoring data is organized into a device start and stop status matrix based on device number and timestamp.

[0018] Differentiated noise parameters for each time period are calculated based on the load fluctuation variance during peak, normal, and valley periods. The noise amplitude is adaptively adjusted according to the load fluctuation characteristics of each time period, and paired white noise is added to the electricity data of the corresponding time period for noise enhancement. During the empirical mode decomposition process, the number of high-frequency modes is set during peak periods to capture severe load fluctuations, the number of medium-frequency modes is set during normal periods to extract stable load characteristics, and the number of low-frequency modes is set during valley periods to analyze the basic load change trend. The inherent modal components of each time period are calculated through integrated averaging to eliminate the influence of noise, retaining the true load change trend to form the purified modal components of the time period. The basic load modal screening is carried out based on the frequency characteristics of the purified modal components of the time period, and the modal components reflecting the basic laws of electricity consumption in each time period are extracted.

[0019] High-frequency modal components are extracted from the basic load characteristics of peak, valley and normal periods. These high-frequency components contain transient load fluctuation information caused by equipment start-up and shutdown. Differentiation penalty parameters are set according to the differences in load characteristics during peak, normal and valley periods, and time period differentiation decomposition parameters are constructed. Equipment state constraints are established based on the equipment start-up and stop state matrix, and the equipment start-up and stop states are associated with load fluctuation signals. The time-sharing variational modal decomposition algorithm takes the high-frequency load fluctuation signals and equipment state constraints of each period as input, and quantifies the matching between the equipment start-up and stop times and the load fluctuation peak times through constraint weight coefficient calculation. The algorithm uses an iterative decomposition strategy to extract variational modes, separates each variational mode by combining frequency domain filtering and time domain constraints, and adjusts the center frequency and bandwidth parameters of each mode according to the equipment state constraints.

[0020] The time-sharing hyperparameter optimization algorithm separates the time-period characteristics of equipment start-stop load correlations based on the differences in start-stop frequencies between peak, normal, and off-peak periods, extracting equipment start-stop load correlation data for each period. The algorithm constructs differentiated network structures for different time periods: a deep network structure is used to address complex load variations during peak periods, a mid-layer network structure is used to handle stable load patterns during normal periods, and a shallow network structure is used to analyze simple load patterns during off-peak periods. Hyperparameters for each time period are optimized using a successive halving strategy, with independent optimization of the learning rate, number of neurons, and dropout rate. The optimized network uses historical start-stop frequency data and corresponding load data for parameter learning. Current start-stop frequency data is then fed into the trained network for load forecasting calculations, yielding load forecast values ​​for each time period.

[0021] The time-of-use load state fusion algorithm calculates load continuity constraints for the switching points of peak, flat, and valley periods in the period start-stop load forecast results, and establishes load continuity constraints between adjacent periods. The device state transfer weight matrix is ​​constructed based on the device start-stop state matrix, and the weight of the device state impact between adjacent periods is calculated. The fusion algorithm calculates the weighted fusion weight of the forecast results for each period based on the period switching continuity constraint parameters and the device state transfer weight matrix, and assigns a time-related fusion weight coefficient to the forecast results for each period. The period start-stop load forecast results are weighted and integrated according to the fusion weight coefficient, and the weight coefficient of the load impact of the equipment start-stop frequency is comprehensively calculated. The fused load forecast sequence is verified to be logically consistent with the actual equipment start-stop state through equipment state consistency verification, and the time-of-use power load modeling results are formed through iterative adjustment and optimization.

[0022] In a specific embodiment, the process of executing step S101 may specifically include the following steps: The 24-hour time period is divided into peak period, normal period and valley period to obtain three time-of-use electricity price period intervals; Based on the three time-of-use electricity price intervals, the electricity data collected by the smart meter is sampled and processed at 15-minute intervals to obtain the time-of-use electricity data points for each period; The time-sharing electricity data points of each period are arranged and combined according to the time series to obtain a time-sharing electricity series containing peak, valley and flat load changes; According to the power threshold of household appliances, high-power devices such as air conditioners, water heaters, and washing machines are screened and processed to obtain a list of target monitoring devices; Based on the target monitoring equipment list, the start and stop status of each device is monitored through the current sensor, the number of device starts and stops is recorded, and the device start and stop status matrix is ​​obtained.

[0023] Specifically, the time-of-use pricing system divides the 24-hour period into three intervals based on time-of-use electricity pricing policies: peak hours cover peak electricity consumption, normal hours cover daily electricity consumption, and valley hours correspond to low nighttime electricity consumption. This division reflects the temporal regularity of residents' electricity consumption behavior and corresponds to different electricity price levels, forming a tiered structure in which peak hours are more expensive than normal hours, and normal hours are more expensive than valley hours.

[0024] The 15-minute interval sampling process slices the smart meter's energy consumption data into fixed time windows. The energy consumption data within each time window constitutes a sampling point. The smart meter uses its built-in metering chip to monitor current and voltage in real time, calculate the instantaneous power value, and then integrate the power value over time to obtain the cumulative energy value over 15 minutes. This sampling method ensures the temporal continuity and integrity of the data while balancing data accuracy and storage space requirements.

[0025] The permutation and combination process concatenates the time-of-day electricity data points for each time period in timestamp order, forming a two-dimensional data sequence with both time and electricity attributes. Each data point contains a timestamp and a corresponding electricity value. The timestamp identifies the time period and the specific moment in time to which the data point belongs, while the electricity value records the electricity consumption within that time window. Data points are logically connected through their temporal sequence, forming a continuous time-of-day electricity sequence that encompasses load variation information for peak, valley, and flat periods.

[0026] High-power device screening sets a power threshold based on the rated power parameters of household appliances, typically set at 1000 watts. Air conditioners typically have a rated power between 1500 and 3000 watts, water heaters between 2000 and 4000 watts, and washing machines between 500 and 1200 watts. The screening algorithm traverses the power parameters of all household appliances and adds devices with rated power exceeding the threshold to the target monitoring list, creating a list of high-power devices that require key monitoring.

[0027] Start-stop status monitoring is accomplished through current sensors installed on each device's power supply circuit. Current sensors use the Hall effect or electromagnetic induction principle to detect current changes in the circuit. When the device starts, the current jumps from zero to the operating current, and when the device stops, the current drops from the operating current to zero. The monitoring algorithm determines the device's start-stop status by setting a current threshold. When the current exceeds the start threshold, it is recorded as a start event; when the current falls below the stop threshold, it is recorded as a stop event. The start count statistics algorithm calculates the number of times a device switches from the off state to the on state during the monitoring period, while the stop count statistics algorithm calculates the number of times a device switches from the on state to the off state. The device start-stop status matrix organizes the start-stop count data for each device into a matrix based on the device number and time period. The rows of the matrix represent different devices, and the columns represent different time periods. The matrix element values ​​represent the number of starts and stops of the corresponding device within the corresponding time period.

[0028] In a specific embodiment, the process of executing step S102 may specifically include the following steps: Adaptive noise amplitude calculation is performed on the time-of-use electricity sequence based on the load fluctuation variance during peak, normal and valley periods to obtain differentiated noise parameters for each period. Based on the differentiated noise parameters of each time period, paired white noise is added to the corresponding time period electricity data for noise enhancement processing to obtain the noisy time period electricity series; The noisy time period electricity series is subjected to empirical mode decomposition according to the number of high-frequency modes in peak period, the number of medium-frequency modes in normal period, and the number of low-frequency modes in valley period, to obtain the inherent modal components of each period; Perform integrated average calculation on the inherent modal components of each time period to eliminate the influence of noise and retain the true load change trend to obtain the purified modal components of the time period; The basic load modal screening process is performed according to the frequency characteristics of the time period purified modal components, and the modal components reflecting the basic laws of electricity consumption in each time period are extracted to obtain the basic load characteristics of peak, valley and normal periods.

[0029] Specifically, the adaptive noise amplitude calculation process first calculates the load fluctuation variance of the time-of-use electricity series during peak, normal, and off-peak periods. The load fluctuation variance reflects the degree of dispersion and fluctuation intensity of the electricity data during each period. During peak periods, due to the frequent start-up and shutdown of household appliances, the load fluctuation is large, resulting in a higher calculated variance. During normal periods, the load fluctuation variance is relatively stable, with a medium level. During off-peak periods, the load fluctuation is minimal, with the lowest variance. The algorithm sets differentiated noise parameters based on the variance values ​​for each period. The noise amplitude is proportional to the load fluctuation variance for that period. Periods with greater fluctuation variance have higher noise amplitudes, while periods with smaller fluctuation variance have lower noise amplitudes. Noise enhancement uses paired white noise addition, which refers to two sets of random noise signals with the same amplitude but opposite signs. The algorithm adds positive and negative noise to the electricity data for the corresponding period, creating two noisy copies of the data. Positive noise increases the original electricity value, while negative noise decreases it. The two sets of noise have the same statistical characteristics but opposite polarity. This paired addition method ensures that the effects of noise on the data statistically cancel each other out, while also increasing the randomness and diversity of the data. The noisy period electricity series contains both the original electricity information and the added noise. The noise component helps the empirical mode decomposition algorithm better identify the true modal characteristics in the data.

[0030] The empirical mode decomposition process sets different modal number parameters according to the differences in load characteristics in each time period. During peak periods, a larger number of high-frequency modes is set, usually 8 to 10 modal components, to capture high-frequency load fluctuations caused by the frequent start and stop of high-power equipment such as air conditioners and water heaters. During normal periods, the number of medium-frequency modes is set, usually 5 to 7 modal components, to analyze the periodic operation modes of equipment such as washing machines and microwave ovens. During valley periods, a smaller number of low-frequency modes is set, usually 3 to 5 modal components, mainly to analyze the low-frequency load changes of continuously running equipment such as refrigerators and routers. The decomposition process extracts inherent modal components layer by layer through a screening algorithm. Each modal component represents an oscillation component with specific frequency characteristics in the original signal. The algorithm sequentially separates each modal component from high to low frequency from the noisy time period power series until the residual component no longer contains obvious oscillation characteristics.

[0031] The integrated averaging process averages the intrinsic modal components of each time period, obtained by adding different noises multiple times. Because paired white noise is added, positive and negative noise cancel each other out over multiple calculations, preserving and enhancing the true load variation trend. During the averaging process, the randomness of the noise components statistically approaches zero, while the determinism of the true load signal strengthens it during the averaging process. The purified modal components of each time period are the result of integrated averaging and contain the true load variation information after noise interference is removed, while maintaining the differences in load characteristics between time periods.

[0032] The basic load modal screening process selects the modal according to the frequency characteristics of the purified modal components in each time period. The algorithm calculates the main frequency and energy distribution of each modal component. The main frequency reflects the speed of the modal oscillation, and the energy distribution reflects the contribution of the mode to the original signal. Basic load modes usually correspond to low-frequency and high-energy modal components. These modes reflect the basic laws and long-term trends of residential electricity consumption. The screening algorithm sets frequency thresholds and energy thresholds. Modes with frequencies below the thresholds and energies above the thresholds are identified as basic load modes. The basic load characteristics of peak, valley and normal periods are reconstructed from the screened basic load modes, which contain the basic law information of electricity consumption in each time period and remove the interference of transient fluctuations such as equipment start and stop.

[0033] In a specific embodiment, the process of executing step S103 may specifically include the following steps: Extract high-frequency modal components from the basic load characteristics during peak, valley and normal periods to filter and process the start-stop fluctuation signal, and obtain high-frequency load fluctuation signals in each period; According to the load characteristics differences between peak period, normal period and valley period, the high-frequency load fluctuation signal of each period is processed by differential penalty parameter setting to obtain the period differential decomposition parameters; Based on the equipment start-stop state matrix, the equipment state constraint conditions are constructed to establish the constraint mechanism, and the equipment start-stop state is associated with the load fluctuation signal to obtain the equipment state constraint conditions; The high-frequency load fluctuation signals and equipment state constraints of each period are input into the variational modal decomposition algorithm for constrained variational solution processing. The optimal variational mode is solved through iterative optimization to obtain the equipment-associated variational modal components. The start-stop behavior matching and verification processing of the device-associated variational modal components is performed to identify the load change mode corresponding to the start-stop moment of the home appliance and obtain the device start-stop load association characteristics.

[0034] Specifically, the high-frequency modal component extraction process uses frequency domain analysis to identify and separate high-frequency components from the basic load characteristics of peak, valley and flat periods. High-frequency modal components refer to oscillation components with higher frequencies, which usually correspond to transient load changes caused by the start and stop operations of household appliances. The algorithm uses spectrum analysis technology to calculate the frequency distribution of basic load characteristics, sets a frequency threshold to divide the modal components into high-frequency and low-frequency categories, and the modes with frequencies exceeding the threshold are classified as high-frequency modal components. The start-stop fluctuation signal screening process further identifies signal components related to the start-stop behavior of the equipment from the high-frequency modal components, analyzes the distribution characteristics of high-frequency modes in the time domain through time-frequency analysis technology, and screens out signal segments with significant energy peaks near the start and stop moments of the equipment as start-stop fluctuation signals.

[0035] Differentiated penalty parameter settings are used to set different decomposition parameters for each period based on the significant differences in load characteristics between peak, off-peak, and off-peak hours. The penalty parameter is a key parameter in the variational mode decomposition algorithm that controls the modal bandwidth. Larger values ​​narrow the modal bandwidth and improve decomposition accuracy, while smaller values ​​widen the modal bandwidth and increase decomposition coarseness. During peak hours, due to frequent equipment startups and shutdowns and complex load fluctuations, the algorithm sets a larger penalty parameter value to achieve more refined modal decomposition results, enabling accurate identification of the transient characteristics of high-power equipment such as air conditioners and water heaters. During off-peak hours, when equipment operation is relatively stable, the algorithm sets a moderate penalty parameter value to balance decomposition accuracy and computational efficiency. During off-peak hours, when load fluctuations are minimal, the algorithm sets a smaller penalty parameter value to focus on slowly changing trends in the base load. The time-period differentiated decomposition parameters include the penalty parameter value, tolerance parameter, and number of iterations for each period, which are key parameters controlling the variational mode decomposition process.

[0036] The equipment status constraint condition construction process establishes the correlation constraint relationship between the equipment operating status and the load fluctuation signal based on the equipment start-stop state matrix. The equipment start-stop state matrix records the number of starts and stops of each device in different time periods and the specific time information. The constraint condition construction algorithm extracts the equipment start and stop times as time constraint points. The correlation constraint mechanism requires that the modal component obtained by variational mode decomposition exhibits a positive load jump feature near the equipment start time and a negative load drop feature near the equipment stop time. The constraint mechanism establishment process integrates the equipment status information into the objective function of variational mode decomposition through a weight function. The load changes corresponding to the equipment start and stop times are given higher weights, and the load changes at non-start and stop times are given lower weights. The equipment status constraint conditions include three dimensions of constraints: time constraints, amplitude constraints, and correlation constraints.

[0037] The constrained variational solution process takes the high-frequency load fluctuation signals and equipment status constraints of each time period as input data and feeds them into the variational modal decomposition algorithm for processing. The variational modal decomposition algorithm decomposes the input signal into several variational modal components with different frequency characteristics by solving the constrained optimization problem. Each modal component has a compact spectral support and a clear physical meaning. The constrained variational solution process uses the alternating direction multiplier method for iterative optimization. The algorithm alternately updates the variational mode and Lagrange multiplier until the convergence conditions are met. During the iterative optimization process, the equipment state constraints act as additional constraints to affect the shape and frequency distribution of the modal components, so that the decomposed modal components have a stronger correlation with the start-stop behavior of the equipment. The equipment-associated variational modal component is the output result of the constrained variational solution process, which contains load change information that is highly correlated with the start-stop behavior of a specific equipment.

[0038] The start-stop behavior matching verification process verifies the degree of match between the device-associated variational modal components and the start-stop times of household appliances through time alignment and correlation analysis. The algorithm calculates the energy distribution and peak position of each variational modal component and compares the time position of the modal energy peak with the start-stop times recorded in the device start-stop state matrix. The matching verification algorithm sets a time tolerance range. The match is considered successful when the time difference between the modal energy peak time and the device start-stop time is less than the tolerance range. Correlation analysis calculates the mutual correlation coefficient between the variational modal components and the device start-stop state sequence. Modes with correlation coefficients exceeding the threshold are considered to have a strong correlation with the corresponding device. The device start-stop load association features are composed of the variational modal components that have passed the matching verification. These features contain detailed information on the impact of the start-stop behavior of household appliances on time-sharing power load.

[0039] In a specific embodiment, the step of inputting the high-frequency load fluctuation signal of each time period and the equipment state constraint condition into a variational mode decomposition algorithm for constrained variational solution processing may specifically include the following steps: The constraint weight coefficient of the high-frequency load fluctuation signal in each period is calculated and processed according to the equipment status constraint conditions. The matching of the equipment start and stop time with the load fluctuation peak time is quantified to obtain the constraint weight coefficient corresponding to the start and stop state of each equipment. Based on the constraint weight coefficient, the high-frequency load fluctuation signal of each period is processed into a weighted constrained variational target. The load fluctuation signal corresponding to the start and stop time of the equipment is given a higher weight to obtain the equipment constrained variational optimization target. The iterative decomposition strategy is used to extract the variational modes of the equipment constraint variational optimization target. The variational modes are separated by combining frequency domain filtering and time domain constraints to obtain the initial variational modal components. The initial variational modal components are processed for modal parameter adjustment through an iterative optimization algorithm. The center frequency and bandwidth parameters of each mode are continuously adjusted according to the equipment state constraints to obtain the optimized variational modal components. The correlation between the optimized variational modal components and the equipment start-stop moments is verified. The correlation between each mode and the time node of the equipment start-stop state matrix is ​​calculated, and the highly correlated modes are screened to obtain the equipment-associated variational modal components.

[0040] Specifically, the constraint weight coefficient calculation process uses a time alignment algorithm to accurately match device start and stop times with peak load fluctuation times. The algorithm first identifies peak and valley load fluctuations from the high-frequency load fluctuation signal for each time period. Peaks correspond to moments of sudden load increases, while valleys correspond to moments of sudden load decreases. The time matching algorithm calculates the time interval between each device start-up time and the peak load fluctuation moment, as well as the time interval between the device stop time and the valley load fluctuation moment, as recorded in the device start-up and stop status matrix. The matching quantification process uses a Gaussian decay function to calculate the matching score corresponding to the time interval. Smaller time intervals are associated with higher matching scores, while larger time intervals are associated with lower matching scores. The constraint weight coefficient is proportional to the matching score. Device start and stop times with higher matching scores are assigned larger constraint weight coefficients, while those with lower matching scores are assigned smaller weight coefficients. The constraint weight coefficients corresponding to each device start and stop status form a weight coefficient vector, which reflects the impact of different device start and stop behaviors on the load fluctuation signal.

[0041] The weighted constrained variational objective construction process performs weighted processing on the high-frequency load fluctuation signals of each time period based on the constraint weight coefficient. The weighted processing process multiplies the constraint weight coefficient as the weight factor with the load fluctuation signal value at the corresponding moment. The load fluctuation signal corresponding to the equipment start and stop moment is amplified, and the signal at non-start and stop moments is reduced. The variational objective function contains two main components: the data fidelity term and the regularization term. The data fidelity term ensures the consistency of the decomposition result with the original signal, and the regularization term controls the smoothness and sparsity of the modal component. The constrained variational objective construction algorithm uses the weighted load fluctuation signal as the input of the data fidelity term, and at the same time adds the equipment state constraint term to the regularization term. The equipment constrained variational optimization objective finds the optimal variational modal decomposition result by minimizing the weighted sum of the weighted data fidelity error and the constrained regularization term. This objective function ensures that the modal components obtained by decomposition have a strong correlation with the equipment start and stop behavior.

[0042] The iterative decomposition strategy uses a combination of frequency domain filtering and time domain constraints to perform variational mode extraction. Frequency domain filtering decomposes the device constraint variational optimization target in the frequency domain by filtering, and decomposes the signal into multiple frequency sub-bands by designing filter groups with different center frequencies and bandwidths. Time domain constraint processing applies device state constraints on each frequency sub-band to ensure that the decomposed modal components exhibit the expected load change characteristics at the start and stop moments of the equipment. The iterative decomposition algorithm alternately performs the two steps of frequency domain filtering and time domain constraints. In each iteration, frequency domain filtering updates the frequency characteristics of each mode, and time domain constraints adjust the time characteristics of each mode. The variational mode extraction process obtains modal components that meet the dual constraints of frequency domain and time domain by solving the constrained optimization problem. The initial variational modal component is the intermediate output result of the iterative decomposition strategy, which contains load change information that is preliminarily associated with the start and stop behavior of the equipment.

[0043] The modal parameter adjustment process uses the gradient descent method to optimize the center frequency and bandwidth parameters of the initial variational modal component. The center frequency parameter determines the main frequency component of the modal component, and the bandwidth parameter controls the frequency distribution range of the modal component. The iterative optimization algorithm calculates the gradient direction of the parameter adjustment based on the device state constraints. The gradient direction points to the direction of parameter change that increases the constraint satisfaction. The parameter adjustment step size is determined using an adaptive strategy. When the constraint satisfaction improves significantly, a larger step size is used to speed up the convergence speed. When the improvement is slow, a smaller step size is used to ensure convergence stability. The modal parameter adjustment process continues until the constraint satisfaction reaches a preset threshold or the number of iterations reaches an upper limit. The optimized variational modal component is the output of the parameter adjustment process. It has optimized center frequency and bandwidth parameters and has significantly improved correlation with the device start-stop behavior compared to the initial modal component.

[0044] The device start-stop correlation verification process calculates the strength of the correlation between the optimized variational modal components and the time nodes of the device start-stop state matrix through correlation analysis. The correlation calculation algorithm uses a cross-correlation function to analyze the correlation between the modal component time series and the device start-stop state series. The magnitude of the cross-correlation function value reflects the degree of temporal synchronization between the two series. The correlation quantification process calculates the maximum cross-correlation coefficient between each optimized variational modal component and the start-stop state of each device. The larger the correlation coefficient, the higher the correlation between the modal component and the corresponding device. The screening process sets a correlation threshold and only retains modal components with a maximum cross-correlation coefficient exceeding the threshold. The high-correlation modal screening algorithm further analyzes the modal components that pass the threshold screening and identifies modal components with significant correlation to the start-stop behavior of specific devices. The device-associated variational modal component is the output of the correlation verification process and contains load variation characteristics closely related to the start-stop behavior of household appliances.

[0045] In a specific embodiment, the process of executing step S104 may specifically include the following steps: According to the start-stop frequency differences in peak, normal, and valley periods, the equipment start-stop load correlation characteristics are separated into time periods, and the equipment start-stop load correlation data of each period are extracted to obtain the start-stop load feature group by time period. Based on the start-stop load characteristic group of different time periods, a differentiated network structure is constructed to perform time period prediction network configuration processing. A deep network is set for peak period, a medium network is set for normal period, and a shallow network is set for valley period, thus obtaining a time period network structure configuration. The time-segment start-stop load feature groups are input into the corresponding time-segment network structure configuration for hyperparameter optimization. The learning rate, number of neurons, and dropout rate of the network in each time period are independently optimized using a successive halving strategy to obtain the time-segment optimized hyperparameter group. The start-stop frequency load forecasting training process is performed on the network in each time period according to the time period optimization hyperparameter group. The network parameters are learned using the historical start-stop frequency data and the corresponding load data to obtain the time period start-stop load forecasting network. The current start-stop frequency data is input into the period start-stop load forecasting network for load forecasting calculation and processing, and the load forecast values ​​for peak period, normal period and valley period are obtained respectively to obtain the period start-stop load forecast results.

[0046] Specifically, the time-period feature separation process classifies the device start-stop load correlation features based on the differences in start-stop frequencies between peak, normal, and off-peak periods. Start-stop frequency differences refer to significant differences in the number of times household appliances start and stop within different time periods. During peak periods, due to concentrated residential electricity demand, devices frequently start and stop. During normal periods, device usage is relatively stable, with a moderate number of starts and stops. During off-peak periods, only basic equipment operates with minimal starts and stops. The separation algorithm extracts data subsets corresponding to the device start-stop load correlation features based on the time period identifier. The peak period data subset contains frequent start and stop information for high-power devices such as air conditioners and water heaters; the normal period data subset contains start and stop information for intermittent devices such as washing machines and microwave ovens; and the off-peak period data subset primarily contains occasional start and stop information for continuously operating devices such as refrigerators and routers. The time-period start-stop load feature group is the output of the separation process and consists of three independent data subsets, each reflecting the correlation between device start and stop behavior and load changes within the corresponding time period. The differentiated network structure construction process designs a corresponding neural network architecture based on the start-stop frequency and load complexity of each time period. A deep network refers to a neural network structure that contains multiple hidden layers. A large number of layers can learn complex nonlinear mapping relationships and is suitable for the complex and changeable start-stop load patterns during peak hours. A mid-level network contains a moderate number of hidden layers, balancing learning ability and computational complexity, and is suitable for the relatively stable start-stop load patterns during normal hours. A shallow network contains fewer hidden layers, has a simple structure and efficient computation, and is suitable for simple start-stop load patterns during valley hours. The time period prediction network configuration process designs a dedicated network structure for each time period. The peak period network adopts a deep structure with three hidden layers, the normal period network adopts a mid-level structure with two hidden layers, and the valley period network adopts a shallow structure with a single hidden layer. The time period network structure configuration includes structural parameters such as the number of layers in each time period, the inter-layer connection method, and the activation function selection.

[0047] The hyperparameter optimization process uses a successive halving strategy to independently optimize the key hyperparameters of the network at each time step. This strategy is an efficient hyperparameter optimization method that gradually eliminates poorly performing hyperparameter combinations to find the optimal configuration. The algorithm first generates a large number of candidate hyperparameter combinations for each time step, including different learning rate values, number of neurons, and dropout rate settings. The learning rate controls the step size for network parameter updates, the number of neurons determines the network's expressiveness, and the dropout rate adjusts the strength of the network's regularization to prevent overfitting. The optimization process is divided into multiple rounds. In each round, the algorithm briefly trains all candidate hyperparameter combinations using the training data. Based on their performance on the validation set, the best-performing half of the candidate combinations are retained, while the poorest-performing half are eliminated. The successive halving process continues until a small number of candidate combinations remain. The algorithm then selects the best-performing combination from these remaining combinations as the optimized hyperparameters for the network at that time step. The time-step optimized hyperparameter set contains the optimized learning rate, number of neurons, and dropout rate parameters for each time step.

[0048] The on / off frequency load forecasting training process uses historical on / off frequency data and corresponding load data to learn parameters for each time period network. The historical on / off frequency data records the number of times each device started and stopped during different time periods over a period of time, while the corresponding load data records the actual power consumption during the same time period. Training data preprocessing uses the on / off frequency data as network input features and the load data as network output targets. Data is assigned to the corresponding time period network for training according to time period identifiers. The network parameter learning process utilizes a backpropagation algorithm. The algorithm calculates parameter gradients based on the error between the predicted and actual loads and uses gradient descent to update network weights and bias parameters. During training, network parameters are updated independently for each time period, and the learning process is controlled by optimized hyperparameters for the corresponding time period. The time period on / off load forecasting network is the output of the training process and consists of three neural networks trained for peak, normal, and off-peak periods. The load forecasting calculation process inputs the current on / off frequency data into the corresponding time period network for forward calculation. The current on / off frequency data represents the number of times each device started and stopped during the time period for load forecasting. The data format is consistent with the historical on / off frequency data used during training. The forecast calculation process assigns input data to the corresponding time period network for processing based on its time period attributes. Peak-period start / stop frequency data is fed into the peak-period network, average-period data into the average-period network, and off-peak data into the off-peak network. Each time period network outputs a load forecast value for the corresponding time period through forward propagation. The forecast value reflects the expected power consumption for that period under the given start / stop frequency conditions. The time period start / stop load forecast results include three independent load forecast values ​​for peak, average, and off-peak periods, forming a load forecast sequence covering all time periods throughout the day.

[0049] The load correlation characteristics of a household's appliance starts and stops show that the air conditioner starts and stops frequently during peak hours, the washing machine starts and stops sporadically during normal hours, and only the refrigerator starts and stops regularly during off-peak hours. A time-of-day feature separation algorithm divides the correlation features into three subsets based on these differences in start-up and stop frequencies. The peak-hour subset contains strong correlations between the air conditioner and load fluctuations, the normal-hour subset contains moderate correlations between the washing machine and load variations, and the off-peak subset contains weak correlations between the refrigerator and the underlying load. During the construction of the differentiated network structure, a three-layer deep network was designed for the peak-hour network due to the complexity of the air conditioner's starts and stops; a two-layer medium network was designed for the normal-hour network due to the regularity of the washing machine's starts and stops; and a single-layer shallow network was designed for the off-peak period due to the simplicity of the refrigerator's starts and stops. During hyperparameter optimization, a successive halving strategy was used to select the optimal hyperparameter combination from a large number of candidate combinations: a learning rate of 1 / 1000, a number of neurons of 128, and a dropout rate of 30%. This combination performed best on the validation set. During the training phase of load forecasting for start-stop frequency, the peak period network uses the number of air conditioner start-stop times in historical data as input features, and the corresponding peak period power consumption as the output target for parameter learning. The network learns the mapping relationship between air conditioner start-stop frequency and load changes. During load forecast calculations, the current peak period air conditioner start-stop frequency data is input into the trained peak period network. The network calculates the corresponding load forecast value based on the learned mapping relationship. This forecast value reflects the impact of the air conditioner start-stop behavior on peak period power consumption. This time period-based network design and training strategy solves the problem of poor model adaptability in existing technologies. By designing specialized network structures and optimization strategies for different time periods, it avoids the technical defect that a single model is difficult to simultaneously handle the differences in load characteristics of each time period. At the same time, independent hyperparameter optimization is used to ensure that the network in each time period achieves optimal performance.

[0050] In a specific embodiment, the process of executing step S105 may specifically include the following steps: Perform load continuity constraint calculations on the switching points of peak, flat, and valley periods in the period start-stop load forecast results, establish load continuity constraint conditions between adjacent periods, and obtain period switching continuity constraint parameters; According to the equipment start-stop state matrix, the state transfer matrix of the equipment state transfer relationship between each time period is constructed and processed, and the state influence weight of the equipment between adjacent time periods is calculated to obtain the equipment state transfer weight matrix; Based on the time period switching continuity constraint parameters and the equipment state transfer weight matrix, the time period start-stop load forecast results are weighted fusion weight calculation and processing are performed, and the time period forecast results are assigned time-related fusion weights to obtain the time period fusion weight coefficients; The load forecast results of each period are weighted and integrated according to the period fusion weight coefficient, and the weight coefficient of the impact of the equipment start-stop frequency on the load is comprehensively calculated to obtain the fusion load forecast sequence; The fused load forecast sequence is subjected to equipment status consistency check to verify the logical consistency between the fusion result and the actual equipment start and stop status, and iterative adjustment and optimization are performed to obtain the time-sharing power load modeling results.

[0051] Specifically, the load continuity constraint calculation process establishes constraints by analyzing the load jump amplitude at the switchover points between peak, off-peak, and off-peak periods in the time period start-stop load forecast results. A time period switchover point refers to the temporal boundary between adjacent time periods, such as the transition from peak to off-peak. Load continuity requires that the load value changes before and after the time period switchover should be smooth, avoiding unreasonable jumps. The constraint calculation algorithm first identifies the load values ​​at the switchover points in each time period's forecast results and calculates the difference between the load forecast values ​​before and after the switchover. The continuity constraint requires that the absolute value of this difference must not exceed a specific percentage of the average load for that period, typically set to 15 percent of the average load. The time period switchover continuity constraint parameters include the load difference limit threshold and constraint strength coefficient at each switchover point; these parameters specify the continuity requirements that must be met during the fusion process. The state transfer matrix construction process analyzes the transmission patterns of equipment operating states between time periods based on the equipment start-stop state matrix. The equipment state transfer relationship refers to the degree to which the equipment operating state in the previous time period affects the equipment operating state in the next time period, for example, the impact of frequent start-up and shutdown of an air conditioner during peak periods on the probability of start-up and shutdown during off-peak periods. An algorithm was constructed to count the frequency of device state transitions between adjacent time periods in historical data and calculate the probability distribution of a device transitioning from a certain state in time period A to various states in time period B. State impact weights were calculated using conditional probabilities, reflecting the impact of the device state in the previous time period on the device state and load changes in the subsequent time period. The device state transfer weight matrix is ​​a three-dimensional data structure, with the first dimension representing the device type, the second dimension representing the source time period, and the third dimension representing the target time period. The matrix element values ​​represent the state transfer weights of the corresponding devices between corresponding time periods.

[0052] The weighted fusion weight calculation process assigns fusion weights to the forecast results for each time period based on the time period switching continuity constraint parameters and the equipment state transfer weight matrix. Time-dependent fusion weights determine the contribution of a time period's forecast result to the predicted value at that time, based on the distance between the forecast time and the time period's center. The weight calculation algorithm uses a Gaussian decay function, assigning higher weights to times closer to the time period's center and lower weights to times farther away. The equipment state transfer weight matrix regulates the weight distribution between adjacent time periods. When the equipment state transfer weight is larger, the influence of the previous time period on the next increases, and the weight distribution is tilted toward the previous time period. The continuity constraint parameters influence the weight calculation through a penalty function. When the forecast result violates the continuity constraint, the weight of the corresponding time period is reduced. The time period fusion weight coefficient is the output of the weight calculation process and contains the fusion weight distribution scheme for each time period at different times. The weighted summation integration process linearly combines the time period's start-stop load forecast results according to the time period fusion weight coefficients. Weighted summation multiplies the forecast value for each time period by the corresponding fusion weight and then accumulates them to obtain the comprehensive forecast value at each time. The integration algorithm traverses each moment in the forecast time series, extracts the forecast values ​​for each time period at that moment and the corresponding fusion weights, and performs a weighted summation calculation. The weight coefficient of the impact of equipment start-up and shutdown frequency on load is included in the integrated calculation as an additional adjustment factor. This coefficient reflects the importance of equipment start-up and shutdown behavior on load changes. The integrated calculation process multiplies the time period fusion result with the equipment start-up and shutdown impact weight to form a fused forecast value that takes into account both time period characteristics and equipment behavior. The fused load forecast sequence is the output of the integrated processing and contains the load forecast time series that has been fused during time periods and adjusted for equipment impact.

[0053] The device state consistency check verifies the rationality of the fusion results by comparing the degree of match between the fused load forecast sequence and the actual device start and stop states. Logical consistency means that the predicted load changes should maintain a logical correspondence with the actual device start and stop behavior: the device start time corresponds to a load increase, and the device stop time corresponds to a load decrease. The verification algorithm extracts the start and stop times recorded in the device start and stop state matrix and searches the fused load forecast sequence for the corresponding time. The consistency verification calculates the degree of match between the device start and stop direction and the load change direction. A high degree of match indicates that the fusion result maintains good logical consistency with the actual device behavior. The iterative adjustment optimization algorithm adjusts the fusion weight coefficient based on the consistency check results. When a moment of poor consistency is found, the algorithm reduces the fusion weight of the dominant period at that moment and increases the weight of other periods. The time-of-use load modeling result is the output of the consistency check and iterative optimization, including the verified and adjusted load forecast sequence and the corresponding confidence assessment.

[0054] The above describes the method for modeling a power load model based on time-sharing electricity in the embodiment of the present application. The following describes the system for modeling a power load model based on time-sharing electricity in the embodiment of the present application. Figure 2 In one embodiment of the present application, a system for modeling a power load model based on time-sharing power consumption includes: The acquisition module is used to collect the time-sharing power consumption of residential users through smart meters to obtain the time-sharing power sequence. At the same time, the device monitoring module is used to identify the start and stop frequency of household appliances to obtain the device start and stop state matrix; The decomposition module is used to perform differential decomposition processing on the data of each time period based on the time-sharing power sequence using the time-sharing power empirical mode decomposition algorithm to obtain the basic load characteristics of peak, valley and normal periods; An identification module is used to identify and process the start-stop fluctuation signal of the basic load characteristics through a time-varying modal decomposition algorithm, and obtain the equipment start-stop load correlation characteristics by combining the equipment start-stop state matrix; A prediction module is used to perform load prediction processing on the start-stop frequency of each time period based on the start-stop load correlation characteristics of the equipment using a time-sharing hyperparameter optimization algorithm to obtain a start-stop load prediction result for each time period; The integration module is used to integrate the start-stop load prediction results of the period through the time-sharing load state fusion algorithm, establish a start-stop frequency influence weight mechanism, and obtain the time-sharing power load modeling results.

[0055] above Figure 2 The power load modeling system based on time-sharing electricity in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The power load modeling device based on time-sharing electricity in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0056] Reference Figure 3 In the embodiment of the present invention, a device for modeling a power load model based on time-sharing power is also provided. The device for modeling a power load model based on time-sharing power can be a server, and its internal structure can be as follows: Figure 3As shown. The power load modeling device based on time-sharing electricity consumption 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 computer-designed processor is used to provide computing and control capabilities. The memory of the power load modeling device based on time-sharing electricity consumption 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 power load modeling device based on time-sharing electricity consumption is used to store the corresponding data in this embodiment. The network interface of the power load modeling device based on time-sharing electricity consumption 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.

[0057] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the time-sharing electricity load modeling device to which the solution of the present invention is applied.

[0058] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions. When the instructions are executed on a computer, the computer executes the steps of the method for modeling a power load model based on time-sharing electricity.

[0059] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0060] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention based on a time-sharing electricity load model. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0061] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for modeling a power load model based on time-sharing electricity consumption, characterized in that: The method comprises: Smart meters are used to collect time-sharing electricity consumption from residential users to obtain a time-sharing electricity sequence. At the same time, the device monitoring module identifies the start and stop frequencies of household appliances to obtain a device start and stop state matrix. According to the time-sharing electricity series, the time-sharing electricity empirical mode decomposition algorithm is used to perform differential decomposition processing on the data of each period to obtain the basic load characteristics of peak, valley and normal periods; The basic load characteristics are processed by time-varying modal decomposition algorithm to identify the start-stop fluctuation signal, and combined with the equipment start-stop state matrix to obtain the equipment start-stop load correlation characteristics; According to the load correlation characteristics of the equipment start and stop, a time-sharing hyperparameter optimization algorithm is used to perform load forecasting on the start and stop frequencies of each period, and a load forecast result of the period start and stop is obtained; The start-stop load forecast results of the period are integrated and processed through the time-sharing load state fusion algorithm, and a start-stop frequency influence weight mechanism is established to obtain the time-sharing power load modeling results.

2. The method for modeling a power load model based on time-sharing electricity according to claim 1, characterized in that: The method uses a smart meter to collect time-sharing electricity consumption from residential users to obtain a time-sharing electricity sequence, and uses an equipment monitoring module to identify the start and stop frequencies of household appliances to obtain an equipment start and stop state matrix, including: The 24-hour time period is divided into peak period, normal period and valley period to obtain three time-of-use electricity price period intervals; Based on the three time-of-use electricity price period intervals, the electricity data collected by the smart meter is sampled and processed at 15-minute intervals to obtain the time-of-use electricity data points for each period; Arrange and combine the time-sharing electricity data points of each period according to the time series to obtain a time-sharing electricity sequence containing peak, valley and flat load changes; According to the power threshold of household appliances, high-power devices such as air conditioners, water heaters, and washing machines are screened and processed to obtain a list of target monitoring devices; Based on the target monitoring equipment list, the start and stop status of each device is monitored by a current sensor, and the number of times the device is started and stopped is recorded to obtain a device start and stop status matrix.

3. The method for modeling a power load model based on time-sharing electricity according to claim 1, characterized in that: According to the time-sharing electricity series, the time-sharing electricity empirical mode decomposition algorithm is used to perform differential decomposition processing on the data of each period to obtain the basic load characteristics of peak, valley and normal periods, including: Adaptively calculating the noise amplitude of the time-sharing electricity sequence based on the load fluctuation variance during peak, normal, and valley periods to obtain differentiated noise parameters for each period; Based on the differentiated noise parameters of each time period, paired white noise is added to the power data of the corresponding time period for noise enhancement processing to obtain a noisy time period power sequence; Performing empirical mode decomposition on the noisy time period electricity quantity sequence according to the number of high-frequency modes in peak time period, the number of medium-frequency modes in normal time period, and the number of low-frequency modes in valley time period to obtain the inherent mode components of each time period; Performing integrated average calculation processing on the inherent modal components of each time period to eliminate the influence of noise and retain the true load change trend, thereby obtaining the purified modal components of the time period; The basic load modal screening process is performed according to the frequency characteristics of the purified modal components in the time period, the modal components reflecting the basic laws of electricity consumption in each time period are extracted, and the basic load characteristics of the peak, valley and normal periods are obtained.

4. The method for modeling a power load model based on time-sharing electricity according to claim 1, characterized in that: The process of performing start-stop fluctuation signal identification processing on the basic load characteristics through a time-varying modal decomposition algorithm and combining the equipment start-stop state matrix to obtain equipment start-stop load correlation characteristics includes: Extracting high-frequency modal components from the basic load characteristics during peak, valley, and normal periods to perform start-stop fluctuation signal screening and processing, thereby obtaining high-frequency load fluctuation signals during each period; According to the load characteristic differences among peak period, normal period and valley period, differential penalty parameter setting processing is performed on the high-frequency load fluctuation signal of each period to obtain differential decomposition parameters of the period; Constructing equipment state constraint conditions based on the equipment start-stop state matrix to perform constraint mechanism establishment processing, correlating equipment start-stop state with load fluctuation signal to obtain equipment state constraint conditions; Inputting the high-frequency load fluctuation signal of each time period and the equipment state constraint condition into the variational modal decomposition algorithm for constrained variational solution processing, solving the optimal variational mode through iterative optimization, and obtaining the equipment-associated variational modal component; The device-associated variational modal component is subjected to start-stop behavior matching verification processing to identify the load change mode corresponding to the start-stop moment of the home appliance, and obtain the device start-stop load association feature.

5. The method for modeling a power load model based on time-sharing electricity according to claim 4, characterized in that: The high-frequency load fluctuation signal of each time period and the equipment state constraint condition are input into the variational modal decomposition algorithm for constrained variational solution processing, and the optimal variational mode is solved through iterative optimization to obtain the equipment-associated variational modal component, including: Calculate and process the constraint weight coefficient of the high-frequency load fluctuation signal in each time period according to the equipment state constraint condition, quantify the matching between the equipment start and stop time and the load fluctuation peak time, and obtain the constraint weight coefficient corresponding to the start and stop state of each equipment; Based on the constraint weight coefficient, the high-frequency load fluctuation signal of each time period is subjected to weighted constrained variational target construction processing, and the load fluctuation signal corresponding to the start and stop time of the equipment is given a higher weight to obtain the equipment constrained variational optimization target; An iterative decomposition strategy is used to perform variational mode extraction processing on the device constraint variational optimization target, and each variational mode is separated by combining frequency domain filtering and time domain constraints to obtain initial variational mode components; The initial variational modal components are subjected to modal parameter adjustment processing by an iterative optimization algorithm, and the center frequency and bandwidth parameters of each mode are continuously adjusted according to the device state constraints to obtain optimized variational modal components; The optimized variational modal components are verified for correlation with the equipment start-stop moments, the correlation between each mode and the time node of the equipment start-stop state matrix is ​​calculated, and the modes with high correlation are screened to obtain the equipment-associated variational modal components.

6. The method for modeling a power load model based on time-sharing electricity according to claim 1, characterized in that: The load forecasting process is performed on the start-stop frequency of each time period using a time-sharing hyperparameter optimization algorithm based on the start-stop load correlation characteristics of the equipment to obtain a load forecast result for each time period, including: According to the start-stop frequency differences among peak period, normal period and valley period, the start-stop load correlation characteristics of the equipment are separated into time period characteristics, and the start-stop load correlation data of the equipment in each time period are extracted to obtain the start-stop load characteristic group by time period; Based on the time-divided start-stop load characteristic group, a differentiated network structure is constructed to perform time-divided prediction network configuration processing, setting a deep network for the peak period, a middle network for the normal period, and a shallow network for the valley period, to obtain a time-divided network structure configuration; The time-divided start-stop load feature group is input into the corresponding time-divided network structure configuration for hyperparameter optimization processing, and the learning rate, number of neurons, and dropout rate of the network in each time period are independently optimized using a successive halving strategy to obtain a time-divided optimized hyperparameter group; Performing a start-stop frequency load forecasting training process on the network for each time period according to the time period optimization hyperparameter group, using historical start-stop frequency data and corresponding load data to learn network parameters, and obtaining a time period start-stop load forecasting network; The current start-stop frequency data is input into the period start-stop load forecasting network for load forecasting calculation processing, and the load forecast values ​​of the peak period, the normal period and the valley period are obtained respectively to obtain the period start-stop load forecast results.

7. The method for modeling a power load model based on time-sharing electricity according to claim 1, characterized in that: The time-sharing load state fusion algorithm is used to integrate the start-stop load forecast results of the time period, establish a start-stop frequency impact weight mechanism, and obtain the time-sharing power load modeling results, including: Perform load continuity constraint calculation on the switching points of the peak period, the normal period, and the valley period in the load start-stop forecast results of the period, establish load continuity constraint conditions between adjacent periods, and obtain period switching continuity constraint parameters; According to the device start-stop state matrix, a state transfer matrix is ​​constructed for the device state transfer relationship between each time period, and the state influence weight of the device between adjacent time periods is calculated to obtain a device state transfer weight matrix; Based on the time period switching continuity constraint parameter and the equipment state transfer weight matrix, a weighted fusion weight calculation process is performed on the start-stop load forecast result of the time period, and a time-related fusion weight is assigned to the forecast result of each time period to obtain a time period fusion weight coefficient; The start-stop load forecast results of the time period are weighted and integrated according to the fusion weight coefficient of the time period, and the weight coefficient of the impact of the equipment start-stop frequency on the load is comprehensively calculated to obtain a fusion load forecast sequence; The fused load forecast sequence is subjected to equipment state consistency verification, the logical consistency between the fusion result and the actual equipment start and stop state is verified, and iterative adjustment and optimization are performed to obtain the time-sharing power load modeling result.

8. A power load modeling system based on time-sharing electricity consumption, characterized in that: A method for modeling a power load model based on time-sharing power as claimed in any one of claims 1 to 7, wherein the power load model modeling system based on time-sharing power comprises: The acquisition module is used to collect the time-sharing power consumption of residential users through smart meters to obtain the time-sharing power sequence. At the same time, the device monitoring module is used to identify the start and stop frequency of household appliances to obtain the device start and stop state matrix; A decomposition module is used to perform differential decomposition processing on the data of each time period according to the time-sharing power sequence using the time-sharing power empirical mode decomposition algorithm to obtain the basic load characteristics of peak, valley and normal periods; an identification module for performing start-stop fluctuation signal identification processing on the basic load characteristics through a time-varying mode decomposition algorithm, and combining the equipment start-stop state matrix to obtain equipment start-stop load correlation characteristics; A prediction module is used to perform load prediction processing on the start-stop frequency of each time period based on the start-stop load correlation characteristics of the equipment using a time-sharing hyperparameter optimization algorithm to obtain a start-stop load prediction result for each time period; The integration module is used to integrate the start-stop load prediction results of the period through the time-sharing load state fusion algorithm, establish a start-stop frequency influence weight mechanism, and obtain the time-sharing power load modeling results.

9. A device for modeling a power load model based on time-sharing power consumption, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, the method for modeling a power load model based on time-sharing power consumption according to any one of claims 1 to 7 is 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 processor is enabled to execute the method for modeling an electricity load model based on time-sharing electricity quantity according to any one of claims 1 to 7.

Images