Hybrid micro-grid load control method, system and device and storage medium

Abstract

The invention discloses a hybrid micro-grid load control method, system and device and a storage medium, and the method comprises the steps: carrying out the feature extraction of historical operation data in a hybrid micro-grid, and obtaining a power factor distribution set of multiple types of loads; performing time sequence analysis on the power factor distribution set, and performing fusion processing on an increment information sequence obtained based on an analysis result to obtain load dynamic power characteristic description data; performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain a first digital twinborn model; performing secondary correction on the first digital twinborn model based on the load power characteristic subset to obtain a second digital twinborn model; and inputting the real-time operation data into the second digital twinborn model to obtain second reactive power demand data, and executing a load control instruction of the hybrid microgrid generated by the second reactive power demand data. According to the method provided by the invention, the operation stability of the hybrid microgrid is improved.

Description

Technical Field

[0001] This application relates to the technical field of power system control, and in particular to a hybrid microgrid load control method, system, device, and storage medium. Background Art

[0002] As a core component unit of the energy Internet, the hybrid microgrid plays a key role in improving energy utilization efficiency, promoting the consumption of green electricity, and ensuring the stability of the power grid by integrating distributed renewable energy, energy storage devices, and various types of loads.

[0003] The existing hybrid microgrid load control methods adopt a technical route of a fixed model plus historical experience parameters. Specifically, the operation data of the power grid is directly input into a model with fixed parameters for reactive power demand analysis. However, this will lead to a large deviation in the calculation of reactive power demand, which does not conform to the actual situation, resulting in the load control strategy generated based on the reactive power demand being difficult to accurately respond to the real-time needs of the hybrid microgrid system, causing problems such as voltage fluctuations and misoperation of reactive power compensation devices, and ultimately affecting the stability of the hybrid microgrid operation. Summary of the Invention

[0004] This application provides a hybrid microgrid load control method, system, device, and storage medium to solve the technical problem of how to improve the existing hybrid microgrid load control method and achieve the stable operation of the hybrid microgrid.

[0005] To solve the above technical problem, an embodiment of this application provides a hybrid microgrid load control method, including: Performing feature extraction on the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution set of each type of load; Performing time series analysis on the power factor distribution set to obtain the incremental information sequence of the power factor change of each load type in different time periods, and performing fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factors of multiple types of loads; Constructing a load power simulation model of the hybrid microgrid, and performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; In the process of load optimization of the actual hybrid microgrid, inputting the real-time operation data of each type of load obtained into the first digital twin model to obtain the first reactive power demand data corresponding to each load type; Judge the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjust the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; Extract features from the load power representation result to obtain a load power characteristic subset including the amplitude of power factor sudden drop and reactive power distortion rate, and perform secondary correction on the first digital twin model based on the load power characteristic subset to obtain a second digital twin model; Input the real-time operation data into the second digital twin model to obtain second reactive power demand data, and execute the load control instruction of the hybrid microgrid generated by the second reactive power demand data.

[0006] As one of the preferred solutions, extracting features from the historical operation data of each type of load in the hybrid microgrid to obtain a power factor distribution set of each type of load, including: Obtain the original data set of each type of load, where the original data set includes inductive loads, electronic device loads, capacitive loads, and nonlinear loads; Calculate the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the nonlinear load respectively to obtain a power factor set corresponding to each load type; Perform smoothing processing on the power factor set based on the mean filtering algorithm, group the smoothed power factor set, and calculate the average power factor of each group to obtain a power factor distribution set of multi-type loads.

[0007] As one of the preferred solutions, the incremental information sequence includes the change rate, fluctuation amplitude, and duration of the power factor; Performing time series analysis on the power factor distribution set to obtain an incremental information sequence of the power factor change of each load type in different time periods, and performing fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the multi-type load power factor, including: Based on the autoregressive integrated moving average model, perform modeling analysis on the power factor change of each load type in the power factor distribution set at a preset time interval to obtain an analysis result; Extract the incremental information with the power factor change amplitude exceeding the preset threshold from the analysis result, and arrange the incremental information in chronological order to obtain a dynamically updated incremental information sequence; Perform fusion processing on the incremental information sequences of each type of load according to the weighted average algorithm to obtain load dynamic power characteristics; Perform time-frequency decomposition on the load dynamic power characteristics based on the wavelet transform algorithm to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

[0008] As one of the preferred solutions, constructing the load power simulation model of the hybrid microgrid, and performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain the corrected first digital twin model, including: Initialize the load power simulation model of the hybrid microgrid based on digital twin technology, set the initial parameters of the load power simulation model to construct a basic simulation model including the voltage and current phase coupling relationship; Input the load dynamic power characteristic description data into the basic simulation model for parameter correction. During the parameter correction process, correct the linear parameters of the basic simulation model according to the Kalman filter algorithm, and correct the nonlinear parameters of the basic simulation model according to the particle swarm optimization algorithm to obtain the first digital twin model reflecting the change of power factor; Among them, the linear parameters include the phase difference between the fundamental voltage and current, and the nonlinear parameters include the harmonic component amplitude.

[0009] As one of the preferred solutions, judging the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjusting the representation weight of the power factor in the first digital twin model based on the matching degree to obtain the load power representation result, including: Establish the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, voltage deviation threshold, and system reactive power reserve margin; Compare the reactive power demand data with the stability constraints. If the comparison result deviation exceeds the preset matching degree threshold, iteratively optimize the representation weight based on the genetic algorithm; where the representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid; Update the first digital twin model based on the optimized representation weight, and calculate the load power representation result including weight adjustment based on the updated first digital twin model.

[0010] As one of the preferred solutions, perform feature extraction on the load power representation result to obtain a load power characteristic subset including the power factor sudden drop amplitude and reactive power distortion rate, and perform secondary correction on the first digital twin model based on the load power characteristic subset to obtain the second digital twin model, including: Perform spectral analysis on the load power characterization result based on the fast Fourier transform algorithm, extract the mutation amplitude of the phase difference between voltage and current and the distortion rate of each harmonic component in the reactive power from the analysis result, and generate a load power characteristic subset containing time-domain mutation characteristics and frequency-domain harmonic characteristics; Input the load power characteristic subset into the first digital twin model, establish an error function between the output value and the actual measurement value of the first digital twin model, and iteratively optimize the error function based on the least squares method to make the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converge to a preset precision threshold, and obtain a second digital twin model adapted to the dynamic load characteristics.

[0011] As one preferred solution, the step of inputting the real-time operation data into the second digital twin model to obtain the second reactive power demand data and executing the load control instruction of the hybrid microgrid generated by the second reactive power demand data includes: Based on the second digital twin model, calculate the reactive power distribution characteristics of each node in the hybrid microgrid in real time, use the support vector machine algorithm to map the reactive power distribution characteristics to the system stability index, and generate a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible regulation scheme of key loads; Send the load control instruction to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

[0012] Another embodiment of this application provides a hybrid microgrid load control system, including: An acquisition module, configured to extract features from the historical operation data of each type of load in the hybrid microgrid to obtain a power factor distribution set of each type of load; An analysis module, configured to perform time series analysis on the power factor distribution set to obtain an incremental information sequence of the power factor change of each load type in different time periods, and perform fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factors of multiple types of loads; A construction module, configured to construct a load power simulation model of the hybrid microgrid, and correct the parameters of the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; A calculation module, configured to input the real-time operation data of each type of load obtained during the load optimization of the actual hybrid microgrid into the first digital twin model to obtain the first reactive power demand data corresponding to each load type; A matching module, configured to determine the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjust the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; A calibration module, configured to perform feature extraction on the load power representation result to obtain a load power characteristic subset including the amplitude of power factor sudden drop and the reactive power distortion rate, and perform secondary calibration on the first digital twin model based on the load power characteristic subset to obtain a second digital twin model; A generation module, configured to input the real-time operation data into the second digital twin model to obtain second reactive power demand data, and execute the load control instruction of the hybrid microgrid generated by the second reactive power demand data.

[0013] As one of the preferred solutions, the acquisition module is specifically configured to: Acquire the original data sets of each type of load, where the original data sets include inductive loads, electronic device loads, capacitive loads, and nonlinear loads; Calculate the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the nonlinear load respectively to obtain a power factor set corresponding to each load type; Perform smoothing processing on the power factor set based on the mean filtering algorithm, group the smoothed power factor set, and calculate the average power factor of each group to obtain a power factor distribution set of multi-type loads.

[0014] As one of the preferred solutions, the incremental information sequence includes the change rate, fluctuation amplitude, and duration of the power factor; The analysis module is specifically configured to: Perform modeling analysis on the power factor changes of each load type in the power factor distribution set at a preset time interval based on the autoregressive integrated moving average model to obtain an analysis result; Extract the incremental information with the power factor change amplitude exceeding a preset threshold from the analysis result, and arrange the incremental information in chronological order to obtain a dynamically updated incremental information sequence; Perform fusion processing on the incremental information sequences of each type of load according to the weighted average algorithm to obtain the load dynamic power characteristics; Perform time-frequency decomposition on the load dynamic power characteristics based on the wavelet transform algorithm to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

[0015] As one of the preferred solutions, the construction module is specifically configured to: Initialize the load power simulation model of the hybrid microgrid based on digital twin technology, and set the initial parameters of the load power simulation model to construct a basic simulation model including the voltage and current phase coupling relationship; Input the load dynamic power characteristic description data into the basic simulation model for parameter correction. During the parameter correction process, correct the linear parameters of the basic simulation model according to the Kalman filter algorithm, and correct the nonlinear parameters of the basic simulation model according to the particle swarm optimization algorithm to obtain the first digital twin model reflecting the power factor change; Among them, the linear parameters include the phase difference between the fundamental voltage and current, and the nonlinear parameters include the harmonic component amplitudes.

[0016] As one of the preferred solutions, the matching module is specifically used for: Establish the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, the voltage deviation threshold, and the system reactive power reserve margin; Compare the reactive power demand data with the stability constraints. If the comparison result deviation exceeds the preset matching degree threshold, iteratively optimize the representation weight based on the genetic algorithm; where the representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid; Update the first digital twin model based on the optimized representation weight, and calculate the load power representation result including weight adjustment based on the updated first digital twin model.

[0017] As one of the preferred solutions, the calibration module is specifically used for: Perform spectrum analysis on the load power representation result based on the fast Fourier transform algorithm, extract the phase difference mutation amplitude between voltage and current and the distortion rate of each harmonic component in reactive power in the analysis result, and generate a load power characteristic subset including time-domain mutation characteristics and frequency-domain harmonic characteristics; Input the load power characteristic subset into the first digital twin model, establish an error function between the output value of the first digital twin model and the actual measurement value, and iteratively optimize the error function based on the least square method to make the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converge to the preset accuracy threshold, and obtain the second digital twin model adapting to the dynamic load characteristics.

[0018] As one of the preferred solutions, the generation module is specifically used for: Based on the second digital twin model, the reactive power distribution characteristics of each node of the hybrid microgrid are calculated in real time. The support vector machine algorithm is used to map the reactive power distribution characteristics to the system stability index, and a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible regulation scheme of key loads is generated. The load control instruction is sent to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

[0019] Another embodiment of the present application provides a hybrid microgrid load control device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the above-mentioned hybrid microgrid load control method is implemented.

[0020] Another embodiment of the present application provides a computer-readable storage medium storing a computer program. When the device where the computer-readable storage medium is located executes the computer program, the above-mentioned hybrid microgrid load control method is implemented.

[0021] Compared with the prior art, the beneficial effects of the embodiments of the present application are at least one of the following: 1) By extracting features and performing time series analysis on historical operation data, the present application constructs a load dynamic power characteristic description data that includes the mapping relationship between time-domain dynamic characteristics and load states. This technology breaks through the limitation of traditional fixed models relying on historical average parameters, can accurately capture the non-linear fluctuations of the power factor of inductive loads with the load rate and the instantaneous reactive power impact of non-linear loads, upgrades the model input from static parameters to a dynamic feature sequence of hybrid microgrid load control devices, and fundamentally solves the modeling problem of the time-varying characteristics of multiple types of loads.

[0022] 2) Relying on a double-layer optimization architecture of primary parameter correction + secondary feature calibration, the present application realizes the dynamic adaptation of the digital twin model to the changes in the operating points of the microgrid. Among them, the primary calibration establishes a basic mapping relationship between fundamental wave and harmonic characteristics by integrating historical incremental information of multiple types of loads; the secondary calibration performs targeted parameter fine-tuning for the extreme working condition characteristics (such as voltage phase mutation, high-frequency harmonic pollution) exposed during real-time operation, forming a closed-loop calibration system of basic modeling - real-time correction, and significantly improving the reliability of the control strategy. Description of the Drawings

[0023] Figure 1 It is a schematic flow chart of a hybrid microgrid load control method in one embodiment of the present application; Figure 2It is a schematic diagram of a hybrid microgrid load control system in one embodiment of the present application; Figure 3 It is a schematic diagram of a hybrid microgrid load control device in one embodiment of the present application.

[0024] Among them, 11 is an acquisition module; 12 is an analysis module; 13 is a construction module; 14 is a calculation module; 15 is a matching module; 16 is a calibration module; 17 is a generation module; 21 is a processor; 22 is a memory. Detailed implementation manners

[0025] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present application.

[0026] In the description of the present application, the terms "first", "second", "third", etc. are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of the present application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0027] In the description of the present application, it should be noted that, unless otherwise clearly defined and limited, the terms "installed", "connected", "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected, or indirectly connected through an intermediate medium, and it may be the communication inside two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are only for the purpose of illustration, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation to the present application. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0028] In the description of this application, it should be noted that unless otherwise defined, all technical and scientific terms used in this application have the same meaning as those commonly understood by those skilled in the technical field to which this application belongs. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0029] An embodiment of this application provides a method for controlling the load of a hybrid microgrid. Specifically, please refer to Figure 1 , Figure 1 which is shown as a schematic flowchart of the method for controlling the load of a hybrid microgrid in one of the embodiments of this application, and it includes steps S1 - S7: S1: Extract features from the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution sets of each type of load; Preferably, in an embodiment of this application, the extracting features from the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution sets of each type of load includes: Obtain the original data set of each type of load, where the original data set includes inductive loads, electronic device loads, capacitive loads, and non - linear loads; Calculate the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the non - linear load respectively to obtain the power factor sets corresponding to each load type; Based on the mean filtering algorithm, smooth the power factor set, group the smoothed power factor set, and calculate the average power factor of each group to obtain the power factor distribution sets of multiple types of loads.

[0030] It should be noted that a hybrid microgrid is a small - scale power system that organically integrates distributed renewable energy sources (such as solar energy, wind energy, etc.), energy storage devices (such as batteries), and various types of loads (such as inductive loads, electronic device loads, etc.). It can operate independently or be connected to the large - scale power grid, and plays an important role in improving energy utilization efficiency, promoting the consumption of green electricity, and ensuring the stability of the power grid. Reactive power is the electromagnetic energy periodically exchanged between energy - storage elements (such as inductors, capacitors) and the power source in an AC circuit. The average value of its instantaneous power is zero, so it does not consume actual electrical energy, but is crucial for maintaining the establishment of the electromagnetic field and energy conversion in the circuit. Numerically, it is equal to the vector difference between apparent power and active power, and the unit is var (Var). The dynamic balance of reactive power directly affects the voltage stability of the system and energy utilization efficiency.

[0031] Historical operation data refers to the records of operation parameters of various loads in the hybrid microgrid over a past period of time, such as data like voltage, current, power, etc. These data reflect the operation states and characteristics of the loads at different times. The power factor is an index to measure the efficiency of electrical energy utilization by electrical equipment, and it is the ratio of active power to apparent power. The closer the power factor is to 1, the higher the efficiency of the equipment in using electrical energy. The power factor distribution set contains the distribution of power factors of various types of loads in the hybrid microgrid, and can reflect the power factor characteristics of different types of loads under different working conditions.

[0032] For inductive loads, the current phase lags behind the voltage phase. For example, devices such as motors and transformers consume a certain amount of reactive power during operation, resulting in a decrease in the power factor. Electronic equipment loads include electronic devices such as computers, servers, and switching power supplies. Their load characteristics are relatively complex and may generate harmonics, which can affect the power factor. Capacitive loads are those where the current phase leads the voltage phase, such as capacitors. Capacitive loads can provide reactive power and improve the power factor to a certain extent. For non-linear loads, the relationship between current and voltage is non-linear, such as rectifiers and frequency converters. Non-linear loads generate harmonics, which degrade the power quality of the power grid and also affect the power factor.

[0033] It can be understood that in the process of load optimization in the hybrid microgrid, it is crucial to accurately understand the power factor distribution of various loads. Different types of loads have different power factor characteristics, and these characteristics change with time and working conditions. By extracting features from historical operation data to obtain the power factor distribution set, it can provide basic data for subsequent load optimization. Specifically, these data can help establish a more accurate load model, predict the reactive power demand of the load, and thus formulate a more reasonable reactive power compensation strategy to improve the operation efficiency and stability of the hybrid microgrid.

[0034] Specifically, in this embodiment, various sensors installed in the hybrid microgrid (such as voltage sensors, current sensors, etc.) are used to collect the operation data of inductive loads, electronic equipment loads, capacitive loads, and non-linear loads in real time, including information such as voltage, current, and phase difference. These data will be stored in the database as the original data set for subsequent processing.

[0035] In this embodiment, the initial power factor values of various loads are calculated respectively. Among them, for inductive loads, electronic equipment loads, and capacitive loads, the initial power factor values can be calculated based on the collected voltage, current, and the phase difference between them. The calculation process is expressed as: Among them, is the fundamental wave phase difference between voltage and current (this value is positive for inductive loads and negative for capacitive loads), is the effective value of the fundamental wave current, is the effective value of the total current.

[0036] For nonlinear loads, due to the nonlinear relationship between their current and voltage, more complex methods are needed to calculate the power factor. The current and voltage signals can be first subjected to Fourier transform to analyze their harmonic components, and then the initial value of the power factor can be calculated based on the fundamental wave components. Through the above calculations, a set of power factors corresponding to each load type is obtained.

[0037] Select a suitable window size N and traverse the set of power factors for each load type. For each power factor data point, calculate the average value of the N data points before and after it, and use this average value as the smoothed power factor of the power factor data point. Group the smoothed power factor set according to preset rules (such as time intervals, load conditions, etc.). For example, the power factors can be divided into different groups according to different time periods of a day (such as day and night), calculate the average value of the power factors in each group, and use these average values as the representative power factors of the current group. Finally, combine the average power factors of all groups to obtain a set of power factor distributions for multi-type loads.

[0038] S2: Perform time series analysis on the set of power factor distributions to obtain an incremental information sequence of the power factor changes of each load type in different time periods, and perform fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factors of multi-type loads; Preferably, in an embodiment of the present application, the incremental information sequence includes the change rate, fluctuation amplitude, and duration of the power factor; The performing time series analysis on the set of power factor distributions to obtain an incremental information sequence of the power factor changes of each load type in different time periods, and performing fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factors of multi-type loads includes: Based on the autoregressive integrated moving average model, perform modeling analysis on the power factor changes of each load type in the set of power factor distributions at a preset time interval to obtain an analysis result; Extract the incremental information with the power factor change amplitude exceeding a preset threshold from the analysis result, and arrange the incremental information in chronological order to obtain a dynamically updated incremental information sequence; Perform fusion processing on the incremental information sequences of each type of load according to the weighted average algorithm to obtain load dynamic power characteristics; Perform time-frequency decomposition on the load dynamic power characteristics based on the wavelet transform algorithm to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

[0039] It should be noted that the incremental information sequence is a sequence composed of the change amounts of the power factor in different time periods (such as the difference between adjacent moments), including three core characteristics: change rate (change amount per unit time), fluctuation amplitude (difference between the maximum and minimum values of the change), and duration (duration of a specific change state), which are used to describe the dynamic evolution process of the power factor.

[0040] The autoregressive integrated moving average model (ARIMA) is a classic time series modeling method. By integrating the autoregressive (AR), differencing (I), and moving average (MA) models, it captures the short-term correlation and trend of the data and is applicable to the analysis and prediction of non-stationary time series. In this embodiment, it is used to model the dynamic change law of the power factor.

[0041] The wavelet transform algorithm is a time-frequency analysis technique that can decompose a time series into components of different frequencies while retaining the localization information in the time and frequency domains, and can effectively identify the instantaneous changes and periodic characteristics in the data. Here, it is used to analyze the time-frequency distribution of the load dynamic power characteristics.

[0042] The load dynamic power characteristic description data is structured data formed by fusing the power factor change information of multiple types of loads, including the mapping relationship between time, power factor, and load status, which is used to accurately describe the dynamic characteristics of the reactive power demand of the load and provide core input parameters for the subsequent digital twin model.

[0043] It can be understood that the loads in the hybrid microgrid have significant time-varying and non-linear characteristics. For example: the power factor of inductive loads (such as motors) changes dynamically with the load rate (the power factor may drop suddenly from 0.8 to 0.6 under light load); the power factor of non-linear loads (such as frequency converters) fluctuates instantaneously due to harmonic interference (the fluctuation amplitude can reach ±30% of the rated value); the reactive power compensation effect of capacitive loads is affected by voltage fluctuations and shows a periodic change law.

[0044] Traditional fixed models cannot capture these dynamic characteristics, resulting in a large deviation in the calculation of reactive power demand (typical error > 25%). Through this embodiment, on the one hand, the incremental information such as the change rate, fluctuation amplitude, and duration of the power factor is made explicit to solve the problem of "fuzzy load characteristics"; on the other hand, the dynamic characteristics of different types of loads are integrated to form a unified characteristic description, providing high-precision input for the digital twin model; on the one hand, the mapping relationship between the load status and the power factor is revealed through time-frequency decomposition, enabling the model to quickly respond to instantaneous changes (such as fluctuations at the 10ms level) and improving the real-time performance of reactive power compensation.

[0045] Specifically, in this embodiment, the power factor distribution set is analyzed in time series and fused to obtain the load dynamic power characteristic description data. First, the power factor data of each type of load needs to be sorted in chronological order, and the change pattern over time is analyzed by the autoregressive integrated moving average model (ARIMA). Specifically, first check whether the data is stable. If there is a trend or periodic fluctuation, it is stabilized by differential processing, and then the model is used to mine the short-term correlation and trend in the data. For example, it is found that the power factor change of a certain type of load in a specific period is affected by the data of the previous few hours. The prediction error is controlled at a low level by training the model.

[0046] Then, the incremental information of power factor changes exceeding the preset threshold (such as the significant change threshold set according to the load characteristics) is extracted, and the rate of these changes (the amount of change per unit time), fluctuation amplitude (the size of the change interval) and duration (the length of time the change state is maintained) are recorded, and they are arranged in chronological order to form a dynamically updated incremental information sequence, such as recording information such as "the power factor of the nonlinear load in a certain period of time dropped by 0.1 within 10 minutes, the amplitude reached 0.15, and lasted for 30 minutes".

[0047] Then, weights are set according to the degree of influence of various types of loads on the stability of the microgrid (for example, nonlinear loads have higher weights due to their greater harmonic influence), and the incremental information sequences of various types of loads are fused through a weighted average algorithm to obtain a dynamic power characteristic that comprehensively reflects the overall load changes. For example, the change rates of different loads are calculated according to the weights to obtain the overall change rate.

[0048] Finally, the wavelet transform algorithm is used to perform time-frequency decomposition on the fused dynamic power characteristics, and the power factor changes are decomposed into components of different time scales, such as long-term trends (such as daily cycle changes), short-term fluctuations (minute-level changes) and instantaneous shocks (second-level mutations). Each component is associated with the timestamp, actual power factor value and load status (such as "equipment startup" and "light load operation") to form load dynamic power characteristic description data containing the mapping relationship between time, power factor and load status. For example, it is clear that "when the electronic equipment is started at 3 pm, the power factor fluctuates by 0.12 within 5 minutes, corresponding to the equipment startup status", which provides accurate dynamic input parameters for the subsequent digital twin model, so that it can capture the changing laws of reactive power demand of multiple types of loads and support real-time reactive power optimization control of the microgrid.

[0049] S3: constructing a load power simulation model of the hybrid microgrid, and performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; Preferably, in an embodiment of the present application, building the load power simulation model of the hybrid microgrid and performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain the corrected first digital twin model includes: Initializing the load power simulation model of the hybrid microgrid based on digital twin technology and setting the initial parameters of the load power simulation model to build a basic simulation model including the voltage and current phase coupling relationship; Inputting the load dynamic power characteristic description data into the basic simulation model for parameter correction. During the parameter correction process, correcting the linear parameters of the basic simulation model according to the Kalman filtering algorithm and correcting the nonlinear parameters of the basic simulation model according to the particle swarm optimization algorithm to obtain the first digital twin model reflecting the power factor change; Wherein, the linear parameters include the phase difference between the fundamental voltage and current, and the nonlinear parameters include the harmonic component amplitudes.

[0050] It should be noted that digital twin technology is to build a model in the virtual space that completely corresponds to the load characteristics of the actual hybrid microgrid, enabling this virtual model to simulate the operating state of the real load in real time. The load power simulation model is a virtual model used to simulate the power consumption characteristics of various loads in the hybrid microgrid and can describe the mutual relationship between voltage and current, such as how the current responds when the voltage changes.

[0051] The Kalman filtering algorithm is a method that can accurately extract effective information from data containing noise, like "denoising" the data, and is particularly suitable for processing parameter correction of linear systems (such as the stable change relationship between fundamental voltage and current). The particle swarm optimization algorithm is an optimization algorithm that imitates the foraging behavior of bird flocks. By continuously searching in the parameter space with "a group of particles", it finds the optimal nonlinear parameters (such as the harmonic magnitude generated by the load) to solve complex nonlinear problems.

[0052] Linear parameters are stable and predictable parameters that describe the phase difference between voltage and current of the load under the fundamental wave (normal 50Hz frequency). The change law of such parameters is linear and can be represented by a simple formula. Nonlinear parameters are complex and irregular parameters that describe the harmonic component amplitudes generated by the load. The change of such parameters is not linear. For example, the harmonic magnitude will suddenly change with the load state and is difficult to describe with a simple formula.

[0053] It can be understood that by analyzing historical data in the first two steps, the present application has obtained the dynamic laws of various load power factors changing with time (such as change speed, fluctuation amplitude). However, these laws need to be "embedded" into a model that can perform real-time calculations in order to be used to predict the reactive power demand of the current load. Traditional fixed models only assume that load parameters remain unchanged (such as a fixed power factor). However, in actual loads, there are both linear fundamental wave characteristics (such as the stable inductance effect of motors) and nonlinear harmonic characteristics (such as random harmonics generated by frequency converters). Different methods must be used to correct these two types of parameters respectively in order to make the model accurately reflect the real situation and avoid reactive power calculation deviations caused by fixed parameters (such as the error of traditional models exceeding 25%), providing a reliable model basis for subsequent real-time optimization.

[0054] First, use digital twin technology to build an initial load power simulation model. This model first presets some basic voltage and current relationships (such as assuming a simple case where the load is a pure resistor), forming a framework that can initially calculate power. Next, input the load dynamic power characteristic description data obtained from the previous analysis into this model and start adjusting the model parameters.

[0055] For the linear parameters in the model (the phase difference between the fundamental wave voltage and current), use the Kalman filter algorithm for correction. The Kalman filter algorithm will continuously compare the phase difference predicted by the model with the phase difference in the actual data, making the linear part (fundamental wave characteristics) of the model more and more accurate. For the nonlinear parameters in the model (the harmonic component amplitudes), such as the magnitudes of the 3rd and 5th harmonics generated by frequency converters, since the changes in these parameters are irregular, the present application uses the particle swarm optimization algorithm for optimization. This algorithm generates a large number of particles, each particle representing a possible harmonic amplitude parameter, and allows the particles to continuously try within the allowed range. Eventually, a set of parameters is found that makes the harmonic amplitude calculated by the model closest to the actually measured harmonic amplitude.

[0056] By adjusting the linear and nonlinear parameters respectively through the above two algorithms, the initial load power simulation model is trained into a first digital twin model that can accurately reflect the real load power factor changes. This model can not only handle the stable phase difference of linear loads such as motors but also capture the random harmonic effects of nonlinear loads such as frequency converters, making the reactive power demand output by the model closer to the actual situation and providing key support for the stable operation of the microgrid.

[0057] S4: During the load optimization process of the actual hybrid microgrid, input the real-time operation data of each type of load obtained into the first digital twin model to obtain the first reactive power demand data corresponding to each load type; It should be noted that the real-time operation data are the operation parameters of various types of loads in the hybrid microgrid collected in real time through sensors (such as voltage transformers and current transformers), including but not limited to voltage amplitude, current waveform, phase difference, real-time value of active power, etc., which reflect the current actual working state of the load.

[0058] The first digital twin model is a load power simulation model calibrated based on historical data, integrating the dynamic power characteristics of multiple types of loads (such as the fundamental wave phase difference change law of linear loads and the harmonic component characteristics of nonlinear loads), and can simulate the real-time power behavior of the load by inputting real-time data. The reactive power demand data are the reactive power values required under the current operating state calculated through the model for various types of loads, and are the core basis for judging the reactive power balance state of the microgrid and formulating compensation strategies.

[0059] It can be understood that the operating state of the hybrid microgrid has strong real-time and dynamic characteristics, and the reactive power demand of the load will change in real time with the operating conditions (such as the load rate change of inductive loads and the start and stop of nonlinear loads) and external conditions (such as the fluctuation of renewable energy output). The traditional fixed-parameter model cannot respond to this change in real time, resulting in a lag in reactive power calculation behind the actual demand (the typical delay exceeds 100 ms), which in turn causes stability problems such as voltage over-limit. By inputting the real-time operation data into the calibrated first digital twin model, on the one hand, it can capture instantaneous changes and reflect the dynamic characteristics such as the current power factor and harmonic components of the load in real time (such as the reactive power impact during the start-up of electronic equipment); on the other hand, it can provide accurate input, provide real-time and accurate reactive power demand data for subsequent stability judgment and optimization strategy generation, and realize the closed-loop real-time control of "data acquisition - model calculation - control execution".

[0060] Specifically, in the actual operation of the hybrid microgrid, sensors deployed at each load node collect the original voltage and current signals at a millisecond-level frequency, and after being preprocessed by the edge computing device, generate real-time operation data containing parameters such as voltage amplitude, fundamental wave phase difference, and harmonic distortion rate.

[0061] Through the dynamic mapping algorithm inside the model, the real-time operation parameters of various types of loads are converted into corresponding reactive power demand values, forming the first reactive power demand data classified by load type (such as "the current inductive reactive power demand of motor No. 1 is 20 kVar" and "the current nonlinear reactive power demand of the frequency converter group is -15 kVar"). These data, as the core input for the reactive power balance analysis of the microgrid, provide real-time basis for subsequent judgment of whether the stability constraints are met and whether the model parameters need to be adjusted, ensuring that the system can quickly respond based on the current operating state and avoiding optimization deviations caused by data lag.

[0062] S5: Determine the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjust the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; Preferably, in an embodiment of the present application, the determining the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjusting the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result includes: Establish the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, the voltage deviation threshold, and the system reactive power reserve margin; Compare the reactive power demand data with the stability constraints. If the comparison result deviation exceeds a preset matching degree threshold, iteratively optimize the representation weight based on the genetic algorithm; where the representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid; Update the first digital twin model based on the optimized representation weight, and calculate a load power representation result with weight adjustment based on the updated first digital twin model.

[0063] It should be noted that the upper limit of reactive power compensation capacity is the maximum reactive power that reactive power compensation devices (such as capacitors and reactors) in the microgrid can provide or absorb, to avoid equipment overload; the voltage deviation threshold is the maximum amplitude of the allowable deviation of the node voltage from the rated value (such as ±5%), to ensure the normal operation of electrical equipment; the system reactive power reserve margin is the reserved reactive power adjustment space in the microgrid under the rated operating state, used to cope with sudden load changes or distributed power fluctuations.

[0064] The representation weight is a quantization parameter that reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the microgrid. The higher the weight, the greater the influence of the power factor change of this load type on the system reactive power balance.

[0065] It can be understood that although the first digital twin model completes parameter calibration based on historical data, there may be two challenges in actual operation. Real-time operating condition differences: The load operating state (such as sudden heavy load of inductive loads, start-stop of non-linear load clusters) may exceed the coverage range of historical data, resulting in a mismatch between the reactive power demand calculated by the model and the actual system capacity (such as the calculated demand exceeding the upper limit of the reactive power compensation device capacity); and model error accumulation: High-frequency dynamic characteristics that are not fully captured during the calibration process (such as millisecond-level harmonic fluctuations) may lead to the breakthrough of stability constraints (such as voltage deviation exceeding the threshold).

[0066] By judging the matching degree and adjusting the representation weights, the attention of the model to different load types can be dynamically corrected, making the model output more conform to the current operating conditions and avoiding the risk of system instability caused by "one-size-fits-all" calculation.

[0067] Specifically, in the microgrid central control system, stability constraint parameters are predefined first: for example, the maximum capacity of the reactive power compensation device is 100 kVar, the allowable voltage deviation is ±5%, and the reactive power reserve margin of the system needs to be maintained at ≥20 kVar. After the first digital twin model outputs the first reactive power demand data of each load type, the system summarizes it into the total reactive power demand of the whole network and compares it with the above constraints: if the total reactive power demand exceeds the upper limit of the reactive power compensation capacity, or the voltage deviation of a certain node exceeds the threshold, or the reactive power reserve margin is lower than the set value, it is determined as "mismatched" and the weight optimization mechanism is triggered; if all indicators are within the constraints, it directly enters the subsequent process without adjustment.

[0068] When optimization is triggered, the genetic algorithm is used to iteratively adjust the representation weights, which specifically includes the following steps: 1) Initialize the weight population, specifically including: allocating initial weights (such as initially all set to 0.25) to load types such as inductive, electronic equipment, capacitive, and nonlinear to form multiple "weight combination individuals"; 2) Calculate the fitness, specifically including: inputting each weight combination into the first digital twin model, calculating the adjusted reactive power demand data, and evaluating its compliance with the stability constraints (the smaller the voltage deviation and the more sufficient the reserve margin, the higher the fitness); 3) Evolution operations, specifically including: selection, crossover, and mutation; among them, the selection operation includes retaining the weight combinations with high fitness and eliminating the combinations with low fitness; the crossover operation includes performing parameter crossover on the retained weight combinations (such as exchanging the weight values of different load types) to generate new individuals; the mutation operation includes making small random adjustments to the weights of the new individuals to increase the population diversity.

[0069] 4) Termination, specifically including: after several generations of evolution, when the optimal weight combination makes the reactive power demand data fully meet the stability constraints, or the fitness no longer improves significantly, stop the optimization.

[0070] Embed the optimized representation weights into the first digital twin model. Inside the model, recalculate the contribution of the power factor change of each load type to the reactive power demand according to the new weights (such as increasing the weight of non-linear loads to prioritize the response to their harmonic effects), and generate the load power representation results with weight adjustment. This result not only reflects the current reactive power demand of the load, but also embodies the dynamic attention of the system to different load types, providing more accurate feature inputs for the subsequent secondary correction of key loads (such as non-linear loads with high weights), forming a closed-loop adaptive mechanism of "model calculation - constraint judgment - weight optimization" to ensure the controllability of the stable operation of the microgrid under complex working conditions.

[0071] S6: Extract features from the load power representation results to obtain a load power characteristic subset including the sudden drop amplitude of the power factor and the reactive power distortion rate. Based on the load power characteristic subset, perform secondary correction on the first digital twin model to obtain the second digital twin model; Preferably, in an embodiment of the present application, the extracting features from the load power representation results to obtain a load power characteristic subset including the sudden drop amplitude of the power factor and the reactive power distortion rate, and performing secondary correction on the first digital twin model based on the load power characteristic subset to obtain the second digital twin model includes: Perform spectrum analysis on the load power representation results based on the fast Fourier transform algorithm, extract the sudden change amplitude of the phase difference between voltage and current and the distortion rate of each harmonic component in the reactive power in the analysis results, and generate a load power characteristic subset including time-domain sudden change characteristics and frequency-domain harmonic characteristics; Input the load power characteristic subset into the first digital twin model, establish an error function between the output value of the first digital twin model and the actual measured value, and perform iterative optimization on the error function based on the least squares method to make the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converge to a preset accuracy threshold, and obtain the second digital twin model adapted to the dynamic load characteristics.

[0072] It should be noted that the sudden drop amplitude of the power factor represents the amplitude of the sudden sharp drop of the power factor within a short time due to reasons such as starting, stopping, and load mutation during the operation of the load (such as suddenly dropping from 0.9 to 0.6), reflecting the instantaneous strong demand of the load for reactive power.

[0073] The reactive power distortion rate represents the proportion of harmonic components in reactive power (such as reactive power fluctuations caused by 3rd and 5th harmonics), which is used to measure the impact degree of nonlinear loads on the power quality of the microgrid. The higher the distortion rate, the more serious the harmonic pollution. The Fast Fourier Transform algorithm (FFT) is an efficient signal processing algorithm that can convert time-domain signals (such as voltage and current waveforms) into frequency-domain signals, separating the fundamental wave and each harmonic component for analyzing the frequency characteristics of the signal. Spectrum analysis is to decompose the frequency components of a signal through tools such as FFT, identify the harmonic components and their amplitudes and phases therein, and is the key means to extract the nonlinear characteristics of the load. The error function is a function that measures the difference between the output value of the digital twin model (such as reactive power demand) and the actual measured value. The smaller the value, the higher the model accuracy. The least squares method is a mathematical optimization method that finds the optimal model parameters by minimizing the sum of the squares of the error function, making the model output as close as possible to the actual data.

[0074] It can be understood that the first digital twin model has been initially adapted to the load power distribution under the current operating conditions, but there is still room for optimization in two aspects. The first is the insufficient coverage of extreme conditions, including sudden drops in power factor (such as the instantaneous low power factor during motor startup) and reactive power distortion (such as high-order harmonics generated by inverters), which belong to high-frequency and transient characteristics and may not be fully captured during the historical data calibration stage and need to be specifically extracted through real-time characterization results. The second is that during the stability constraint matching process, the weight adjustment focuses on overall balance and does not deeply correct the model's simulation ability for detailed features such as harmonic components and phase mutations, resulting in possible amplification of errors under complex conditions.

[0075] By extracting the characteristic subset containing the drop amplitude and distortion rate and performing secondary calibration, the model's ability to depict "extreme dynamic characteristics" can be strengthened, converging the error to a lower level and providing further accuracy guarantee for accurately generating reactive power optimization schemes.

[0076] Specifically, after obtaining the load power characterization results, first use the Fast Fourier Transform algorithm to perform spectrum analysis on it: convert the voltage and current signals collected in real time from the time domain (waveforms changing with time) to the frequency domain (amplitudes and phases of each harmonic), and separate the fundamental wave (50Hz) and harmonic components (such as 150Hz for the 3rd harmonic, 250Hz for the 5th harmonic, etc.). During this process, two key features are mainly extracted: time-domain mutation features and frequency-domain harmonic features. Among them, the time-domain mutation features are used for the mutation amplitude of the phase difference between voltage and current in a short time to judge whether a sudden drop in power factor occurs and its severity; the frequency-domain harmonic features are used to calculate the proportion of each harmonic component in reactive power (i.e., the reactive power distortion rate). For example, the reactive power distortion rate caused by the 3rd harmonic is 15%, and the 5th harmonic is 10%, forming the distribution characteristics of harmonic components.

[0077] Integrate these two types of features into a "load power characteristic subset", which includes parameters such as the moment of voltage dip occurrence, amplitude, duration, and the distortion rate of each harmonic, and input it into the first digital twin model. An error evaluation mechanism of "output value - measured value" is established inside the model. Compare the reactive power demand calculated by the model with the measured value of the sensor, and construct an error function (reflecting the overall deviation between the two).

[0078] Iteratively optimize the error function based on the least squares method. By continuously fine-tuning the parameters related to voltage dips and distortions in the model (such as the harmonic generation coefficient of non-linear loads and the transient inductance change rate of inductive loads), the value of the error function is gradually reduced. The specific process includes: if the calculated reactive power distortion rate of the model is 10% lower than the actual value when the frequency converter starts, then increase the parameter weight of the corresponding harmonic component; if the calculated magnitude of the power factor dip is too small when the motor starts, then correct the change coefficient of its transient phase difference.

[0079] Continue to iterate until the error function converges to the preset accuracy threshold. At this time, the simulation accuracy of the model for power factor dips and reactive power distortions is significantly improved, and it can accurately reflect the reactive power demand characteristics under complex working conditions such as non-linear load shocks and transient changes of inductive loads, forming a second digital twin model adapted to dynamic load characteristics. As the final accurate calculation tool, this model lays a foundation for generating reliable load optimization schemes in the follow-up, ensuring that the microgrid can still maintain voltage stability and reactive power balance when facing extreme challenges such as high-frequency fluctuations and harmonic pollution.

[0080] S7: Input the real-time operation data into the second digital twin model to obtain the second reactive power demand data, and execute the load control instruction of the hybrid microgrid generated from the second reactive power demand data.

[0081] Preferably, in an embodiment of the present application, the step of inputting the real-time operation data into the second digital twin model to obtain the second reactive power demand data and executing the load control instruction of the hybrid microgrid generated from the second reactive power demand data includes: Based on the second digital twin model, calculate the reactive power distribution characteristics of each node of the hybrid microgrid in real time. Use the support vector machine algorithm to map the reactive power distribution characteristics to the system stability index, and generate a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible regulation scheme of key loads; Send the load control instruction to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

[0082] It should be noted that the second reactive power demand data are the reactive power demand data of each node obtained by inputting the real-time operation data into the second digital twin model, which reflects the actual demand for reactive power of each part of the hybrid microgrid at the current moment. The support vector machine algorithm is a machine learning algorithm that can establish a mapping relationship between the reactive power distribution characteristics and the system stability index, can handle complex non-linear problems, and find a load optimization scheme that meets the stability requirements.

[0083] The switching strategy of the reactive power compensation device includes determining when to put into or cut off the reactive power compensation equipment (such as capacitors and reactors) to adjust the reactive power balance of the microgrid. The specific adjustment amount of the reactive power output of the distributed power source is the value for adjusting the reactive power output by the distributed power source (such as solar and wind power generation equipment), and the reactive power distribution of the microgrid is optimized by changing its reactive power output. The sliding window standard deviation algorithm is an algorithm used to analyze the fluctuation of data within a certain time window, and the standard deviation is calculated to measure the fluctuation amplitude of the reactive power and judge whether it meets the stability requirements.

[0084] It can be understood that through the processing of the previous steps, the second digital twin model that can accurately reflect the dynamic characteristics of the hybrid microgrid is obtained. However, the ultimate goal of the model is to optimize the reactive power of the microgrid and ensure its stable operation. By inputting the real-time operation data into the model to obtain accurate reactive power demand data, and then generating an optimization scheme, the reactive power distribution can be adjusted in real time according to the current actual situation of the microgrid, avoiding problems such as voltage fluctuations and increased equipment losses caused by reactive power imbalance, and improving the operation efficiency and stability of the microgrid.

[0085] Specifically, in this embodiment, first, the operation data (such as voltage, current, power, etc.) of each node of the hybrid microgrid collected in real time are input into the second digital twin model. The model calculates in real time the reactive power distribution of each node, that is, how much reactive power each node needs to maintain normal operation.

[0086] Next, the support vector machine algorithm is used to map the calculated reactive power distribution characteristics to the system stability index. This algorithm will learn the relationship between the reactive power distribution and the system stability, and generate an optimization scheme including the switching strategy of the reactive power compensation device and the adjustment amount of the reactive power output of the distributed power source according to the current reactive power distribution situation. For example, if the reactive power of a certain node is insufficient, the scheme may recommend putting into the corresponding capacitor for reactive power compensation; if the reactive power is excessive near a certain distributed power source, the scheme may require the power source to reduce its reactive power output.

[0087] Then, according to the generated optimization scheme, corresponding control instructions are generated and sent to the controllers of the reactive power compensation device and the distributed power source to perform the adjustment operation of the reactive power.

[0088] After performing the regulation operation, continuously collect the real-time operation data of each node in the hybrid microgrid. Use the sliding window standard deviation algorithm to analyze this data and calculate the fluctuation amplitude of the reactive power within a certain time window. Compare the calculated fluctuation amplitude with the preset rated value. If the fluctuation amplitude is higher than the rated value, it indicates that the current optimization scheme fails to effectively control the fluctuation of reactive power and does not meet the stability constraint. At this time, it is necessary to iteratively optimize the scheme, re-adjust the switching strategy of the reactive power compensation device and the reactive power output adjustment amount of the distributed power source, perform the regulation operation again and verify until the fluctuation amplitude of the reactive power at each node is lower than the preset rated value, and output the load optimization scheme that meets the stability constraint to ensure the stable and efficient operation of the hybrid microgrid.

[0089] To further explain the effectiveness and feasibility of this application, the following embodiments are used to further illustrate this application.

[0090] Specifically, through various sensors installed in the microgrid, such as voltage sensors and current sensors, continuously collect the operation data of various types of loads in the past month to form an original data set. For the induction load motor, calculate the initial power factor according to the collected voltage, current and the phase difference between them; for the computer server of the electronic equipment load, considering its load characteristics and possible harmonics, first perform Fourier transform analysis on the current and voltage signals to analyze the harmonic components, and then calculate the initial power factor according to the fundamental wave component; for the capacitor bank of the capacitive load, calculate the initial power factor based on the characteristic that the current phase leads the voltage phase; for the frequency converter of the nonlinear load, also use a complex harmonic analysis method to calculate the initial power factor, so as to obtain the power factor set corresponding to each load type.

[0091] Select the mean filtering algorithm with a window size of 10 to smooth the power factor set and eliminate the noise interference in the data. Then group the smoothed power factor set according to different time periods (daytime, night) and load conditions (full load, half load, light load) in a day, calculate the average power factor of each group, and finally obtain the power factor distribution set of multiple types of loads.

[0092] Adopt the autoregressive integrated moving average model to model and analyze the power factor changes of each load type in the power factor distribution set at a preset time interval of 1 hour. Extract the incremental information with the power factor change amplitude exceeding 5% (preset threshold) in the analysis results, including the change rate, fluctuation amplitude and duration of the power factor, and arrange these incremental information in chronological order to obtain a dynamically updated incremental information sequence.

[0093] According to the weighted average algorithm, different weights are set for different types of loads, and the incremental information sequences of various types of loads are fused to obtain the load dynamic power characteristics. Then, the wavelet transform algorithm is used to perform time-frequency decomposition on the load dynamic power characteristics to obtain the load dynamic power characteristic description data containing the mapping relationship between time, power factor, and load status.

[0094] Based on the digital twin technology, the load power simulation model of the hybrid microgrid is initialized, the initial parameters of the model are set, and the basic simulation model containing the coupling relationship between voltage and current phase is constructed. The load dynamic power characteristic description data is input into the basic simulation model for parameter correction. During the correction process, the Kalman filter algorithm is used to correct the linear parameters such as the phase difference between the fundamental voltage and current, and the particle swarm optimization algorithm is used to correct the nonlinear parameters such as the harmonic component amplitude to obtain the first digital twin model reflecting the power factor change.

[0095] The operation data of each type of load in the hybrid microgrid, such as voltage, current, phase difference, etc., are collected in real time and input into the first digital twin model to obtain the first reactive power demand data corresponding to each load type.

[0096] The stability constraints of the hybrid microgrid are established, including the upper limit of reactive power compensation capacity being 500 kVar, the voltage deviation threshold being ±5%, and the system reactive power reserve margin being 10%. The first reactive power demand data is compared with the stability constraints, and it is found that the comparison result deviation in some time periods exceeds 10% (the preset matching degree threshold). At this time, the genetic algorithm is used to iteratively optimize the representation weight of the power factor in the first digital twin model, and this representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid. The first digital twin model is updated based on the optimized representation weight, and the load power representation result including weight adjustment is calculated.

[0097] Based on the fast Fourier transform algorithm, the spectrum analysis of the load power representation result is carried out, the mutation amplitude of the phase difference between voltage and current and the distortion rate of each harmonic component in the reactive power are extracted to generate a load power characteristic subset containing time-domain mutation characteristics and frequency-domain harmonic characteristics. The load power characteristic subset is input into the first digital twin model, the error function between the model output value and the actual measured value is established, and the least squares method is used to iteratively optimize the error function so that the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converges to 3% (the preset accuracy threshold) to obtain the second digital twin model adapted to the dynamic load characteristics.

[0098] Based on the second digital twin model, the reactive power distribution characteristics of each node in the hybrid microgrid are calculated in real time. The support vector machine algorithm is used to map the reactive power distribution characteristics to the system stability index, and load control instructions including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible control scheme of critical loads are generated. These load control instructions are sent to the corresponding execution units in the hybrid microgrid, such as the switching switches of capacitor banks, the controllers of distributed power sources, etc., to adjust the reactive power output and load power consumption of the hybrid microgrid in real time to ensure the stable operation of the microgrid.

[0099] Through the above embodiments, the load control method of the hybrid microgrid can effectively cope with the dynamic changes of various types of loads in the hybrid microgrid and improve the operation efficiency and stability of the microgrid.

[0100] Another embodiment of this application provides a hybrid microgrid load control system. Specifically, please refer to Figure 2 , Figure 2 which shows a schematic diagram of the hybrid microgrid load control system in one of the embodiments of this application. It includes: an acquisition module 11, an analysis module 12, a construction module 13, a calculation module 14, a matching module 15, a calibration module 16, and a generation module 17. Among them, The acquisition module 11 is used to extract features from the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution set of each type of load; The analysis module 12 is used to perform time series analysis on the power factor distribution set to obtain an incremental information sequence of the power factor change of each load type in different time periods, and perform fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factor of various types of loads; The construction module 13 is used to construct a load power simulation model of the hybrid microgrid, and perform parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; The calculation module 14 is used to input the real-time operation data of each type of load obtained during the load optimization of the actual hybrid microgrid into the first digital twin model to obtain the first reactive power demand data corresponding to each load type; The matching module 15 is used to judge the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjust the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; A calibration module 16, configured to extract features from the load power characterization result to obtain a subset of load power characteristics including the amplitude of power factor dip and the reactive power distortion rate, and perform secondary correction on the first digital twin model based on the subset of load power characteristics to obtain a second digital twin model; A generation module 17, configured to input the real-time operation data into the second digital twin model to obtain second reactive power demand data, and execute the load control instruction of the hybrid microgrid generated from the second reactive power demand data.

[0101] Preferably, in an embodiment of the present application, the acquisition module 11 is specifically configured to: Acquire the original data set of each type of load, where the original data set includes inductive loads, electronic device loads, capacitive loads, and nonlinear loads; Calculate the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the nonlinear load respectively to obtain a set of power factors corresponding to each load type; Perform smoothing processing on the set of power factors based on the mean filter algorithm, group the smoothed set of power factors, and calculate the average power factor of each group to obtain a set of power factor distributions of multi-type loads.

[0102] As one preferred solution, the incremental information sequence includes the change rate, fluctuation amplitude, and duration of the power factor; Preferably, in an embodiment of the present application, the analysis module 12 is specifically configured to: Perform modeling analysis on the power factor changes of each load type in the set of power factor distributions at a preset time interval based on the autoregressive integrated moving average model to obtain an analysis result; Extract the incremental information with the power factor change amplitude exceeding the preset threshold in the analysis result, and arrange the incremental information in chronological order to obtain a dynamically updated incremental information sequence; Perform fusion processing on the incremental information sequences of each type of load according to the weighted average algorithm to obtain the load dynamic power characteristics; Perform time-frequency decomposition on the load dynamic power characteristics based on the wavelet transform algorithm to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

[0103] Preferably, in an embodiment of the present application, the construction module 13 is specifically configured to: Initialize the load power simulation model of the hybrid microgrid based on digital twin technology, and set the initial parameters of the load power simulation model to construct a basic simulation model including the voltage and current phase coupling relationship; Input the load dynamic power characteristic description data into the basic simulation model for parameter calibration. During the parameter calibration process, correct the linear parameters of the basic simulation model according to the Kalman filtering algorithm, and correct the non-linear parameters of the basic simulation model according to the particle swarm optimization algorithm, so as to obtain the first digital twin model reflecting the power factor change; Among them, the linear parameters include the phase difference between the fundamental wave voltage and current, and the non-linear parameters include the harmonic component amplitudes.

[0104] Preferably, in an embodiment of the present application, the matching module 15 is specifically used for: Establish the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, voltage deviation threshold, and system reactive power reserve margin; Compare the reactive power demand data with the stability constraints. If the comparison result deviation exceeds the preset matching degree threshold, iteratively optimize the representation weight based on the genetic algorithm; where the representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid; Update the first digital twin model based on the optimized representation weight, and calculate the load power representation result including weight adjustment based on the updated first digital twin model.

[0105] Preferably, in an embodiment of the present application, the calibration module 16 is specifically used for: Perform spectrum analysis on the load power representation result based on the fast Fourier transform algorithm, extract the phase difference mutation amplitude between voltage and current and the distortion rate of each harmonic component in reactive power in the analysis result, and generate a load power characteristic subset including time-domain mutation characteristics and frequency-domain harmonic characteristics; Input the load power characteristic subset into the first digital twin model, establish an error function between the output value of the first digital twin model and the actual measurement value, and iteratively optimize the error function based on the least squares method, so that the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converges to the preset accuracy threshold, and obtain a second digital twin model adapted to the dynamic load characteristics.

[0106] Preferably, in an embodiment of the present application, the generation module 17 is specifically used for: Based on the second digital twin model, calculate the reactive power distribution characteristics of each node of the hybrid microgrid in real time, and map the reactive power distribution characteristics to the system stability index by using the support vector machine algorithm, and generate a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of distributed power source reactive power output, and the flexible regulation scheme of key loads; Send the load control instruction to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

[0107] Another embodiment of the present application provides a hybrid microgrid load control device. Specifically, please refer to Figure 2 , Figure 2 which shows a schematic diagram of the hybrid microgrid load control device in one of the embodiments of the present application. Refer to Figure 3 , which is a structural block diagram of the hybrid microgrid load control device provided by the embodiment of the present application. The hybrid microgrid load control device provided by the embodiment of the present application includes a processor 21, a memory 22, and a computer program stored in the memory 22 and configured to be executed by the processor 21. When the processor 21 executes the computer program, the steps in the embodiment of the above-mentioned hybrid microgrid load control method are implemented, such as Figure 1 the steps S1 to S7 described in

[0108] Exemplarily, the computer program can be divided into one or more modules. The one or more modules are stored in the memory 22 and executed by the processor 21 to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program in the hybrid microgrid load control device. For example, the computer program can be divided into an acquisition module 11, an analysis module 12, a construction module 13, a calculation module 14, a matching module 15, a calibration module 16, and a generation module 17.

[0109] The hybrid microgrid load control device may include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art can understand that the schematic diagram is only an example of the hybrid microgrid load control device, and does not constitute a limitation on the hybrid microgrid load control device. It may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the hybrid microgrid load control device may further include input / output devices, network access devices, buses, etc.

[0110] The processor 21 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor 21 is the control center of the hybrid microgrid load control device, and connects all parts of the hybrid microgrid load control device using various interfaces and lines.

[0111] The memory 22 can be used to store the computer programs and / or modules. The processor 21 realizes various functions of the hybrid microgrid load control device by running or executing the computer programs and / or modules stored in the memory 22, and by calling the data stored in the memory 22. The memory 22 mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.). In addition, the memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0112] Among them, if the modules integrated in the hybrid microgrid load control device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present application, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0113] Those of ordinary skill in the art can understand that to implement all or part of the processes in the above-mentioned embodiment methods, it can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned method embodiments. Among them, the storage medium can be a magnetic disk, optical disk, read-only memory (Read-Only Memory, ROM), or random access memory (Random Access Memory, RAM), etc.

[0114] Correspondingly, the embodiments of the present application provide a computer-readable storage medium, and the computer-readable storage medium includes a stored computer program. Among them, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the steps in the hybrid microgrid load control method in the above-mentioned embodiment, for example Figure 1 the steps S1 - step S7 described above.

[0115] Compared with the prior art, the beneficial effects of the embodiments of the present application are at least one of the following: 1) By extracting features and performing time series analysis on historical operation data, the present application constructs load dynamic power characteristic description data including the mapping relationship between time-domain dynamic characteristics and load states. This technology breaks through the limitation of traditional fixed models relying on historical average parameters, can accurately capture the non-linear fluctuations of the power factor of inductive loads with the load rate and the instantaneous reactive power impact of non-linear loads, upgrades the model input from static parameters to the dynamic characteristic sequence of the hybrid microgrid load control device, and fundamentally solves the modeling problem of the time-varying characteristics of multi-type loads.

[0116] 2) Relying on the double-layer optimization architecture of primary parameter calibration + secondary feature calibration, this application realizes the dynamic adaptation of the digital twin model to the changes in the operating points of the microgrid. Among them, the primary calibration establishes the basic mapping relationship between the fundamental wave and harmonic characteristics by integrating the historical increment information of multiple types of loads; the secondary calibration conducts targeted parameter fine-tuning for the extreme working condition characteristics (such as voltage phase mutation and high-frequency harmonic pollution) exposed during real-time operation, forming a closed-loop calibration system of basic modeling - real-time correction, significantly improving the reliability of the control strategy.

[0117] The above embodiments only represent several implementation manners of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several deformations and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of the patent of this application shall be subject to the appended claims.

Claims

1. A hybrid microgrid load control method, characterized in that, Including: Performing feature extraction on the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution set of each type of load; Performing time series analysis on the power factor distribution set to obtain the incremental information sequence of the power factor change of each load type in different time periods, and performing fusion processing on the incremental information sequence of each load type to obtain the load dynamic power characteristic description data reflecting the dynamic change law of the power factor of multiple types of loads; Constructing a load power simulation model of the hybrid microgrid, and performing parameter correction on the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; Inputting the real-time operation data of each type of load obtained from the hybrid microgrid into the first digital twin model to obtain the first reactive power demand data corresponding to each load type; Judging the matching degree between the first reactive power demand data and the stability constraint of the hybrid microgrid, and adjusting the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; Performing feature extraction on the load power representation result to obtain a load power characteristic subset including the power factor sudden drop amplitude and the reactive power distortion rate, and performing secondary correction on the first digital twin model based on the load power characteristic subset to obtain a second digital twin model; Inputting the real-time operation data into the second digital twin model to obtain the second reactive power demand data, and executing the load control instruction of the hybrid microgrid generated by the second reactive power demand data.

2. The hybrid microgrid load control method according to claim 1, wherein The performing feature extraction on the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution set of each type of load includes: Obtaining the original data set of each type of load, where the original data set includes inductive loads, electronic device loads, capacitive loads, and nonlinear loads; Respectively calculating the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the nonlinear load to obtain the power factor set corresponding to each load type; Performing smoothing processing on the power factor set based on the mean filtering algorithm, grouping the smoothed power factor set, and calculating the average power factor of each group to obtain the power factor distribution set of multiple types of loads.

3. The hybrid microgrid load control method according to claim 1, characterized in that, The incremental information sequence includes the change rate, the fluctuation amplitude, and the duration of the power factor; The performing time series analysis on the power factor distribution set to obtain the incremental information sequence of the power factor change of each load type in different time periods, and performing fusion processing on the incremental information sequence of each load type to obtain the load dynamic power characteristic description data reflecting the dynamic change law of the power factor of multiple types of loads includes: Based on the autoregressive integrated moving average model, performing modeling analysis on the power factor change of each load type in the power factor distribution set at a preset time interval to obtain an analysis result; Extracting the incremental information with the power factor change amplitude exceeding a preset threshold from the analysis result, and arranging the incremental information in chronological order to obtain a dynamically updated incremental information sequence; Fusing and processing the incremental information sequences of various types of loads according to the weighted average algorithm to obtain the load dynamic power characteristics; Performing time-frequency decomposition on the load dynamic power characteristics based on the wavelet transform algorithm to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

4. The hybrid microgrid load control method according to claim 1, wherein Constructing the load power simulation model of the hybrid microgrid, and calibrating the parameters of the load power simulation model based on the load dynamic power characteristic description data to obtain the calibrated first digital twin model, including: Initializing the load power simulation model of the hybrid microgrid based on digital twin technology, setting the initial parameters of the load power simulation model to construct a basic simulation model including the coupling relationship between voltage and current phases; Inputting the load dynamic power characteristic description data into the basic simulation model for parameter calibration. During the parameter calibration process, correcting the linear parameters of the basic simulation model according to the Kalman filter algorithm and correcting the nonlinear parameters of the basic simulation model according to the particle swarm optimization algorithm to obtain the first digital twin model reflecting the change of power factor; Wherein, the linear parameters include the phase difference between the fundamental wave voltage and current, and the nonlinear parameters include the harmonic component amplitudes.

5. The load control method for a hybrid microgrid according to claim 1, wherein Judging the matching degree between the first reactive power demand data and the stability constraints of the hybrid microgrid, and adjusting the representation weight of the power factor in the first digital twin model based on the matching degree to obtain the load power representation result, including: Establishing the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, voltage deviation threshold, and system reactive power reserve margin; Comparing the reactive power demand data with the stability constraints. If the comparison result deviation exceeds the preset matching degree threshold, iteratively optimizing the representation weight based on the genetic algorithm; wherein the representation weight reflects the influence degree of the power factor change of different types of loads on the reactive power demand of the hybrid microgrid; Updating the first digital twin model based on the optimized representation weight, and calculating the load power representation result including weight adjustment based on the updated first digital twin model.

6. The load control method for a hybrid microgrid according to claim 1, wherein Extracting the characteristics of the load power representation result to obtain a load power characteristic subset including the sudden drop amplitude of the power factor and the reactive power distortion rate, and performing secondary calibration on the first digital twin model based on the load power characteristic subset to obtain the second digital twin model, including: Performing spectrum analysis on the load power representation result based on the fast Fourier transform algorithm, extracting the sudden change amplitude of the phase difference between voltage and current and the distortion rate of each harmonic component in the reactive power in the analysis result, and generating a load power characteristic subset including time-domain sudden change characteristics and frequency-domain harmonic characteristics; Input the subset of the load power characteristics into the first digital twin model, establish an error function between the output value of the first digital twin model and the actual measurement value, and iteratively optimize the error function based on the least squares method to converge the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data to a preset precision threshold, so as to obtain a second digital twin model adapted to the dynamic load characteristics.

7. The hybrid microgrid load control method according to claim 1, characterized in that, The step of inputting the real-time operation data into the second digital twin model to obtain second reactive power demand data and executing the load control instruction of the hybrid microgrid generated by the second reactive power demand data includes: Based on the second digital twin model, calculate the reactive power distribution characteristics of each node of the hybrid microgrid in real time, and use the support vector machine algorithm to map the reactive power distribution characteristics to the system stability index, and generate a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible regulation scheme of key loads; Send the load control instruction to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

8. A hybrid microgrid load control system, characterized in that, It includes: An acquisition module for extracting features from the historical operation data of each type of load in the hybrid microgrid to obtain the power factor distribution set of each type of load; An analysis module for performing time series analysis on the power factor distribution set to obtain an incremental information sequence of the power factor change of each load type in different time periods, and performing fusion processing on the incremental information sequences of each load type to obtain load dynamic power characteristic description data reflecting the dynamic change law of the power factor of multi-type loads; A construction module for constructing a load power simulation model of the hybrid microgrid, and correcting the parameters of the load power simulation model based on the load dynamic power characteristic description data to obtain a corrected first digital twin model; A calculation module for inputting the real-time operation data of each type of load obtained in the load optimization process of the actual hybrid microgrid into the first digital twin model to obtain first reactive power demand data corresponding to each load type; A matching module for judging the matching degree between the first reactive power demand data and the stability constraint of the hybrid microgrid, and adjusting the representation weight of the power factor in the first digital twin model based on the matching degree to obtain a load power representation result; A calibration module for extracting features from the load power representation result to obtain a subset of load power characteristics including the power factor sudden drop amplitude and the reactive power distortion rate, and performing secondary correction on the first digital twin model based on the load power characteristics subset to obtain a second digital twin model; A generation module for inputting the real-time operation data into the second digital twin model to obtain second reactive power demand data and executing the load control instruction of the hybrid microgrid generated by the second reactive power demand data.

9. The hybrid microgrid load control system according to claim 8, wherein, The acquisition module is specifically used for: Obtain the original data set for each type of load, where the original data set includes inductive load, electronic device load, capacitive load, and nonlinear load; Calculate the initial power factor values of the inductive load, the electronic device load, the capacitive load, and the nonlinear load respectively to obtain a power factor set corresponding to each load type; Based on the mean filtering algorithm, smooth the power factor set, group the smoothed power factor set, and calculate the average power factor of each group to obtain a power factor distribution set for multi-type loads.

10. The hybrid microgrid load control system according to claim 8, characterized in that, The incremental information sequence includes the change rate, fluctuation amplitude, and duration of the power factor; The analysis module is specifically used for: Based on the autoregressive integrated moving average model, perform modeling analysis on the power factor changes of each load type in the power factor distribution set at a preset time interval to obtain an analysis result; Extract the incremental information with the power factor change amplitude exceeding the preset threshold in the analysis result, and arrange the incremental information in chronological order to obtain a dynamically updated incremental information sequence; According to the weighted average algorithm, fuse the incremental information sequences of each type of load to obtain the load dynamic power characteristics; Based on the wavelet transform algorithm, perform time-frequency decomposition on the load dynamic power characteristics to obtain load dynamic power characteristic description data including the mapping relationship between time, power factor, and load status.

11. The hybrid microgrid load control system according to claim 8, characterized in that, The construction module is specifically used for: Based on the digital twin technology, initialize the load power simulation model of the hybrid microgrid, and set the initial parameters of the load power simulation model to construct a basic simulation model including the voltage and current phase coupling relationship; Input the load dynamic power characteristic description data into the basic simulation model for parameter correction. During the parameter correction process, correct the linear parameters of the basic simulation model according to the Kalman filtering algorithm, and correct the nonlinear parameters of the basic simulation model according to the particle swarm optimization algorithm to obtain the first digital twin model reflecting the power factor change; Among them, the linear parameters include the phase difference between the fundamental wave voltage and current, and the nonlinear parameters include the harmonic component amplitude.

12. The hybrid microgrid load control system according to claim 8, characterized in that, The matching module is specifically used for: Establish the stability constraints of the hybrid microgrid, where the stability constraints at least include the upper limit of reactive power compensation capacity, voltage deviation threshold, and system reactive power reserve margin; Compare the reactive power demand data with the stability constraints. If the comparison result deviation exceeds the preset matching degree threshold, iteratively optimize the representation weight based on the genetic algorithm; where the representation weight reflects the influence degree of the power factor changes of different types of loads on the reactive power demand of the hybrid microgrid; Update the first digital twin model based on the optimized representation weight, and calculate the load power representation result including weight adjustment based on the updated first digital twin model.

13. The hybrid microgrid load control system according to claim 8, wherein, The calibration module is specifically used for: Perform spectral analysis on the load power characterization result based on the fast Fourier transform algorithm, extract the mutation amplitude of the phase difference between voltage and current and the distortion rate of each harmonic component in the reactive power from the analysis result, and generate a load power characteristic subset containing time-domain mutation characteristics and frequency-domain harmonic characteristics; Input the load power characteristic subset into the first digital twin model, establish an error function between the output value of the first digital twin model and the actual measurement value, and iteratively optimize the error function based on the least squares method to make the root mean square error between the reactive power demand data output by the first digital twin model and the real-time operation data converge to a preset precision threshold, and obtain a second digital twin model adapted to the dynamic load characteristics.

14. The hybrid microgrid load control system according to claim 8, wherein The generating module is specifically configured to: Based on the second digital twin model, calculate the reactive power distribution characteristics of each node in the hybrid microgrid in real time, and use the support vector machine algorithm to map the reactive power distribution characteristics to the system stability index, and generate a load control instruction including the switching strategy of reactive power compensation devices, the adjustment amount of reactive power output of distributed power sources, and the flexible control scheme of key loads; Send the load control instruction to the corresponding execution unit in the hybrid microgrid to adjust the reactive power output and load power consumption of the hybrid microgrid in real time.

15. A hybrid microgrid load control device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the hybrid microgrid load control method according to any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program. When the device where the computer-readable storage medium is located executes the computer program, it implements the hybrid microgrid load control method according to any one of claims 1 to 7.

Images