A step data coupling simulation method for accounting for multi-time scale devices

Abstract

The application relates to the technical field of power simulation, and discloses a step data coupling simulation method for considering multi-time scale equipment. The method comprises the following steps: classifying power system equipment through time constant analysis to obtain a fast variable equipment group and a slow variable equipment group, and establishing a step coupling state vector; calculating a step coupling correlation coefficient and a step data transmission variation coefficient; and obtaining a step coupling configuration parameter through ISCS-PSVR algorithm optimization. The application solves the technical problem that the existing multi-time scale equipment step data coupling simulation method lacks a quantitative coupling evaluation mechanism and adaptive parameter optimization capability, and improves the balance performance of simulation precision and calculation efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power simulation, and in particular to a step data coupling simulation method for considering multi-time scale devices. BACKGROUND

[0002] In the existing power system simulation technology, the step data coupling simulation method of multi-time scale devices mainly adopts a fixed step strategy to uniformly calculate microsecond-level power electronic devices and millisecond-level traditional power devices under the same simulation step, or adopts a simple layered simulation method to process the data transmission problem between different time scale devices through a preset step ratio relationship. The traditional method realizes the exchange of state variables between fast and slow variable devices through basic algorithms such as linear interpolation or zero-order hold, and performs multi-time scale coordinated simulation according to the coupling parameters set by experience.

[0003] However, the existing technology has significant deficiencies. The fixed step strategy cannot balance the high frequency precision requirement of fast variable devices and the calculation efficiency requirement of slow variable devices, resulting in stability problems such as numerical oscillation and error accumulation in the simulation process. The simple layered simulation method lacks real-time monitoring and quantitative evaluation of the coupling state between devices, and cannot dynamically adjust the simulation parameters according to the changes in system operating conditions, resulting in a decrease in coupling precision and a waste of computing resources. The basic interpolation algorithm lacks precision when dealing with the multi-physical field coupling of complex power systems, and the experience-based parameter setting method lacks scientific basis and adaptive ability.

[0004] A deeper technical problem is that the existing method lacks a quantitative evaluation mechanism for the coupling state of multi-time scale devices, and cannot establish a correlation analysis of the long-term correlation and short-term transient precision between devices, thereby leading to a lack of scientific basis for simulation parameter optimization. Due to the lack of quantitative indicators such as coupling correlation coefficient and data transmission variation coefficient, the existing technology cannot accurately identify when the step configuration needs to be adjusted, nor can it evaluate whether the current coupling precision meets the simulation requirements. The lack of such an evaluation mechanism directly affects the design and implementation of adaptive optimization algorithms, making it difficult for existing simulation systems to maintain sustained high precision and high efficiency in complex and variable power system operating scenarios. SUMMARY

[0005] The present application provides a step data coupling simulation method for considering multi-time scale devices, which solves the technical problems of lack of quantitative coupling evaluation mechanism and adaptive parameter optimization capability in the existing multi-time scale device step data coupling simulation method, and improves the balance performance of simulation precision and calculation efficiency.

[0006] In a first aspect, the present application provides a step data coupling simulation method for considering multi-time scale devices, which comprises:

[0007] S1 step: classifying power system equipment through time constant analysis to obtain a fast-changing equipment group and a slow-changing equipment group, and establishing a step coupling state vector;

[0008] S2 step: performing correlation coefficient calculation processing on the fast-changing equipment group and the slow-changing equipment group according to the step coupling state vector to obtain a step coupling correlation coefficient;

[0009] S3 step: performing coefficient of variation analysis processing on the coupling interface data of the fast-changing equipment group and the slow-changing equipment group to obtain a step data transmission coefficient of variation;

[0010] S4 step: performing optimization processing on the step coupling correlation coefficient and the step data transmission coefficient of variation through an ISCS-PSVR algorithm to obtain a step coupling configuration parameter.

[0011] Optionally, the S1 step further comprises:

[0012] extracting the time constant of power electronic devices and traditional power equipment in the power system to obtain a device time scale identification matrix;

[0013] grouping and classifying microsecond-level devices and millisecond-level devices according to the device time scale identification matrix to obtain the fast-changing equipment group and the slow-changing equipment group;

[0014] performing vector construction processing based on the voltage and current state of the fast-changing equipment group and the mechanical state of the slow-changing equipment group to obtain the step coupling state vector;

[0015] mapping the step coupling state vector through coordinate conversion to obtain a state variable mapping relationship.

[0016] Optionally, the S2 step further comprises:

[0017] performing time reference unification processing on the state sequence of the fast-changing equipment group and the slow-changing equipment group to obtain a synchronous time point mapping sequence;

[0018] performing correlation analysis processing on fast-changing state sequences and slow-changing state sequences according to the synchronous time point mapping sequence to obtain the step coupling correlation coefficient;

[0019] performing threshold comparison processing based on the step coupling correlation coefficient to obtain a coupling state evaluation result;

[0020] performing continuous monitoring processing on the coupling state evaluation result according to a time sequence to obtain state deviation trend data.

[0021] Optionally, the S3 step further comprises:

[0022] extracting voltage current power transfer data between the fast varying device group and the slow varying device group, forming the coupling interface data;

[0023] calculating a ratio of a standard deviation to a mean of the coupling interface data in a single simulation iteration period, determining the step data transfer coefficient of variation;

[0024] comparing the step data transfer coefficient of variation with a preset variation threshold, determining a transient synchronization state of the coupling interface, generating a transient accuracy evaluation index;

[0025] establishing a mapping relationship between the transient accuracy evaluation index and the state deviation trend data, forming a coupling accuracy correlation degree.

[0026] Optionally, the S4 step further comprises:

[0027] constructing a multi-objective optimization function containing the step coupling correlation degree coefficient and the step data transfer coefficient of variation, combining the state variable mapping relationship to establish an input feature vector;

[0028] initializing population individual coding of the ISCS-PSVR algorithm according to the input feature vector, setting a search range of the fast varying device step parameter and the slow varying device step parameter based on the device time scale identification matrix;

[0029] performing iterative optimization processing of the ISCS-PSVR algorithm with the input feature vector as the evaluation basis of the fitness function, calculating the fitness value of the population individual based on the coupling accuracy correlation degree and the transient accuracy evaluation index;

[0030] outputting the step configuration scheme corresponding to the population individual with the optimal fitness value from the optimization result guided by the input feature vector, determining the step coupling configuration parameter.

[0031] Optionally, the initializing population individual coding of the ISCS-PSVR algorithm according to the input feature vector comprises:

[0032] extracting step configuration dimension information in the input feature vector, determining the coding length and coding structure of the population individual;

[0033] randomly generating initial population individual numerical values according to the coding structure of the population individual, forming a population individual coding matrix;

[0034] mapping the coding numerical values in the population individual coding matrix into specific numerical values of the fast varying device step parameter and the slow varying device step parameter;

[0035] Verify rationality of the fast-changing device step length parameter and the slow-changing device step length parameter based on the device time scale identification matrix, and generate an effective population individual set.

[0036] Optionally, the input feature vector is evaluated as a fitness function, and the iteration optimization processing of the ISCS-PSVR algorithm is executed, the fitness value of the population individual is calculated based on the coupling accuracy correlation degree and the transient accuracy evaluation index, and the fitness value of the population individual is calculated.

[0037] The step length configuration corresponding to each individual in the effective population individual set is substituted into the multi-objective optimization function, and the simulation error value and the calculation efficiency value of the single population individual are calculated.

[0038] The simulation error value of the single population individual is weighted according to the coupling accuracy correlation degree, and the calculation efficiency value of the single population individual is corrected in combination with the transient accuracy evaluation index, to form a comprehensive evaluation index of the single population individual.

[0039] All individuals in the effective population individual set are sorted based on the comprehensive evaluation index of the single population individual, and the fitness value ranking of each population individual is determined.

[0040] The population individual with a high fitness value ranking is selected for crossover and mutation operation, the population individual coding matrix is updated, and the next generation population individual is generated.

[0041] In the technical scheme provided in the application, the power system devices are classified and processed through time constant analysis to obtain fast-changing device groups and slow-changing device groups, and a step length coupling state vector is established. This technical feature realizes systematic identification and unified modeling of devices with different time scales, and lays a data foundation for subsequent accurate coupling analysis. According to the step length coupling state vector, the correlation coefficient of the fast-changing device group and the slow-changing device group is calculated to obtain the step length coupling correlation coefficient technical feature. Through quantitative analysis, an evaluation mechanism for long-term coupling stability between devices is established, solving the problem of lack of quantitative evaluation of coupling states in the prior art. The coupling interface data of the fast-changing device group and the slow-changing device group are analyzed by the coefficient of variation to obtain the step length data transmission coefficient of variation technical feature, which realizes accurate monitoring of short-term transient transmission accuracy and makes up for the deficiency of traditional methods that cannot capture transient distortion. The step length coupling correlation coefficient and the step length data transmission coefficient of variation are optimized by the ISCS-PSVR algorithm to obtain the step length coupling configuration parameter technical feature, and an intelligent parameter optimization system based on quantitative indexes is established, which significantly improves the self-adaptation ability and overall performance of the simulation system.

[0042] In the specific application field of digital real-time simulation of power systems, the ISCS-PSVR algorithm as the core algorithm features an improved sparrow search strategy that is specially designed for the time scale difference problem of power equipment, a power grid topology perception mechanism in the algorithm can guide the search direction according to the network structure characteristics of the power system, and the introduction of the support vector regression model enables the algorithm to accurately predict the influence of different step length configurations on the simulation performance of the power system. The algorithm is functionally deeply coupled with the step length coupling correlation coefficient, the step length data transmission variation coefficient and other technical features, the fitness evaluation process of the algorithm directly depends on these quantitative indicators, and the optimization result in turn guides the dynamic adjustment of the coupling parameters, and the mutual support relationship ensures that the entire technical solution can achieve the optimal balance between precision and efficiency in complex power system simulation scenarios, and provides a more reliable and efficient simulation tool for the safety and stability analysis of new power systems. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor based on these drawings.

[0044] Figure 1 An embodiment of the step length data coupling simulation method for considering multi-time scale equipment in the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0045] The embodiments of the present application provide a step length data coupling simulation method for considering multi-time scale equipment. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "includes" or "has" and any variation thereof is intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0046] For the sake of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 The embodiment of the step length data coupling simulation method for considering multi-time scale equipment in the embodiments of the present application includes:

[0047] S1 step: classifying power system equipment through time constant analysis to obtain fast-changing equipment group and slow-changing equipment group, and establishing step coupling state vector;

[0048] S2 step: calculating the correlation coefficient of the fast-changing equipment group and the slow-changing equipment group according to the step coupling state vector, to obtain the step coupling correlation coefficient;

[0049] S3 step: performing coefficient of variation analysis on the coupling interface data of the fast-changing equipment group and the slow-changing equipment group, to obtain the step data transmission coefficient of variation;

[0050] S4 step: optimizing the step coupling correlation coefficient and the step data transmission coefficient of variation through the ISCS-PSVR algorithm, to obtain the step coupling configuration parameter.

[0051] It can be understood that the execution subject of the present application can be a step data coupling simulation system for considering multi-time scale equipment, and can also be a terminal or a server, which is not limited here. The server is taken as an example for illustration in the embodiments of the present application.

[0052] Specifically, in the S1 step, time constant analysis is performed on various types of equipment in the power system. The analysis process extracts the switching frequency characteristics of power electronic devices and the mechanical time constant of traditional power equipment, and establishes a device time scale identification matrix. The matrix contains the characteristic time parameters of each device, based on which IGBT, current transformer and other devices with microsecond-level response are classified into a fast-changing equipment group, and synchronous generators, speed regulators and other devices with millisecond-to-second-level response are classified into a slow-changing equipment group. Based on the state characteristics of the two types of equipment groups, the system constructs a step coupling state vector, which integrates the voltage and current transient information of fast-changing equipment and the mechanical angle speed information of slow-changing equipment.

[0053] The S2 step performs correlation calculation on the state sequence of the fast-changing equipment group and the slow-changing equipment group. The process unifies the time base through processing, maps the states of devices with different time scales to a synchronous time point sequence, and eliminates the influence of time scale difference on data analysis. The correlation coefficient calculation uses the Pearson correlation algorithm to quantify the linear correlation degree between the state sequences of fast and slow changing equipment, and generates a step coupling correlation coefficient. The coefficient reflects the degree of state synchronization between multi-time scale equipment in the long-term simulation process. When the coefficient value is lower than a preset threshold, it indicates that there is significant state deviation accumulation between the devices.

[0054] S3 step extracts the voltage, current and power transfer data of the coupling interface for the data transfer process between the fast-varying device group and the slow-varying device group, and forms a coupling interface data set. The data set reflects the transient interaction characteristics of devices of different time scales within a single simulation iteration period. The coefficient of variation analysis quantifies the degree of fluctuation in the data transfer process by calculating the ratio of the standard deviation to the mean of the coupling interface data, and generates the step data transfer coefficient of variation. The coefficient is combined with the transient accuracy evaluation index to construct a quantitative description system for the transient behavior of the coupling interface of multi-time scale devices.

[0055] In S4 step, the ISCS-PSVR algorithm integrates the step coupling correlation coefficient and the step data transfer coefficient of variation to construct a multi-objective optimization function. The algorithm takes the two types of coefficients as the core components of the input feature vector, and represents different step configuration schemes through population individual coding. During the optimization process, the algorithm determines the search boundaries of the fast-varying device step parameter and the slow-varying device step parameter according to the device time scale identification matrix, ensuring the physical reasonableness of the optimization results. The input feature vector is used as the basis for fitness evaluation, and the coupling accuracy correlation and the transient accuracy evaluation index are combined to calculate the comprehensive performance index of each population individual corresponding to the step configuration.

[0056] In the technical solution, the step coupling state vector provides a data source for the correlation coefficient calculation, and the long-term synchronization characteristics reflected by the correlation coefficient and the short-term fluctuation characteristics reflected by the coefficient of variation together constitute the input space of the ISCS-PSVR algorithm. The algorithm dynamically adjusts the step configuration to meet the long-term stability and short-term accuracy requirements through iterative optimization. For example, the microsecond-level switching action of the wind power converter and the second-level frequency response of the synchronous generator form a multi-time scale coupling. The traditional fixed step method cannot balance the high-frequency switching accuracy of the converter and the low-frequency stability of the generator. The scheme monitors the long-term synchronization state of the two through the correlation coefficient and captures the transient fluctuation of the converter output power through the coefficient of variation. The ISCS-PSVR algorithm optimizes the microsecond step of the converter and the millisecond step of the generator accordingly, significantly reducing the computational complexity while ensuring simulation accuracy. The technical solution solves the technical problem of the lack of systematic coupling evaluation mechanism in existing multi-time scale simulation methods, and establishes a quantitative correlation coefficient and coefficient of variation evaluation system.

[0057] In a specific embodiment, S1 step further comprises:

[0058] Performing time constant extraction processing on power electronic devices and traditional power devices in the power system to obtain a device time scale identification matrix;

[0059] Grouping and classifying the microsecond-level devices and millisecond-level devices according to the device time scale identification matrix to obtain a fast-varying device group and a slow-varying device group;

[0060] a step coupling state vector is obtained through a vector construction process based on the voltage and current states of the fast-changing device group and the mechanical states of the slow-changing device group;

[0061] The step coupling state vector is mapped through coordinate conversion to obtain a state variable mapping relationship.

[0062] Specifically, the time scale attributes of various devices are determined by analyzing their dynamic response characteristics. For power electronic devices, the switching time of IGBT, the reverse recovery time of diode, the control loop response time of the converter, and other key parameters are extracted, which are usually in the order of nanoseconds to microseconds. For traditional power devices, the electrical time constant of the synchronous generator, the mechanical time constant, the excitation system time constant, the governor response time, and other parameters are extracted, which are usually in the order of milliseconds to seconds. By establishing a mapping relationship between device parameters and time constants, a device time scale identification matrix is constructed, which represents the device number in rows and the time constant type in columns, and the matrix element value is the characteristic time parameter of the corresponding device.

[0063] Based on the device time scale identification matrix, the system performs device grouping and classification processing, which divides the devices into different categories according to the order of magnitude difference of the time constant. When the dominant time constant of a device is less than 100 microseconds, the system classifies it as a fast-changing device group, typical representatives including IGBT switching devices, PWM controllers, high-frequency transformers, and other power electronic devices. When the dominant time constant of a device is greater than 1 millisecond, the system classifies it as a slow-changing device group, typical representatives including synchronous generator bodies, turbine speed regulation systems, excitation regulators, and other traditional power devices. This grouping process is realized through a threshold judgment algorithm, ensuring that each device is accurately assigned to the corresponding device group according to its time scale characteristics, laying the foundation for subsequent differentiated simulation step configuration.

[0064] The vector construction process integrates the state information of the fast-changing device group and the slow-changing device group to form a unified step coupling state vector. For the fast-changing device group, the process extracts the voltage and current transient states of the device, including the on-off state of the switching device, the output voltage amplitude and phase of the converter, the high-frequency component of the current, and other information. For the slow-changing device group, the process extracts the mechanical state parameters of the device, including the mechanical angle, angular velocity, electromagnetic torque, and mechanical power of the generator rotor. The state vector construction adopts a block matrix form, placing the fast-changing device state in the first half of the vector and the slow-changing device state in the second half of the vector. Through this structured organization method, the state information of different time scales can be processed and analyzed in a unified framework.

[0065] The step-coupling state vector is mapped by coordinate conversion to establish the correlation between the state variables of different devices. The mapping process converts three-phase alternating current into two-phase direct current by Park transformation, and realizes the conversion between abc coordinate system and αβ coordinate system by Clarke transformation. Through these coordinate conversion operations, the data incompatibility problem caused by different reference coordinate systems of different devices is eliminated. The establishment process of the state variable mapping relationship also includes the normalization of physical quantities, which unifies the voltage, current, power and other physical quantities of different devices to the same reference value system, ensuring that the high-frequency small signal of fast-changing devices and the low-frequency large signal of slow-changing devices can be compared and operated in the same dimension.

[0066] The time constant extraction provides data for correlation analysis, the device grouping result directly affects the formulation of step configuration strategy, the state vector construction provides data source for coupling interface analysis, and the coordinate conversion mapping ensures the comparability of the state of devices with different time scales. Taking a photovoltaic power station as an example, the IGBT switching frequency of the photovoltaic inverter is 10 kHz, and the corresponding time constant is 100 microseconds, which is classified as a fast-changing device group; the magnetizing inrush time constant of the grid-connected transformer is 100 milliseconds, which is classified as a slow-changing device group. The system extracts the DC bus voltage and AC output current of the inverter as fast-changing states, and extracts the flux and excitation current of the transformer as slow-changing states, and constructs step-coupling state vectors containing these parameters. Through dq coordinate transformation, the three-phase alternating current output by the inverter is converted into direct current, and a unified mapping relationship is established with the excitation parameters of the transformer.

[0067] In a specific embodiment, the S2 step further comprises:

[0068] The state sequences of the fast-changing device group and the slow-changing device group are subjected to time reference unification processing to obtain a synchronous time point mapping sequence;

[0069] According to the synchronous time point mapping sequence, the fast-changing state sequence and the slow-changing state sequence are subjected to correlation analysis processing to obtain a step-coupling correlation coefficient;

[0070] Based on the step-coupling correlation coefficient, a threshold comparison processing is performed to obtain a coupling state evaluation result;

[0071] The coupling state evaluation result is subjected to continuous monitoring processing according to the time sequence to obtain state deviation trend data.

[0072] Specifically, the time reference unification process for fast-changing device group and slow-changing device group state sequence eliminates the time asynchronization problem caused by different device sampling frequency differences by establishing a common time reference. The fast-changing device group collects state data at a microsecond-level time step to form a high-density state sequence, while the slow-changing device group collects state data at a millisecond-level time step to form a low-density state sequence. The time reference unification process maps the two types of state sequences onto a unified time axis through an interpolation algorithm, with the sampling period of the fast-changing device as the basic unit, and the state value of the slow-changing device at the fast-changing device sampling time through linear interpolation or cubic spline interpolation algorithm. The construction process of the synchronous time point mapping sequence selects the common sampling time points of the two types of device groups as the synchronization nodes, at which the fast-changing device state directly takes the original sampling value, and the slow-changing device state takes the interpolation result, forming a time-aligned state data pair.

[0073] The correlation analysis process is based on the synchronous time point mapping sequence, and calculates the linear correlation degree between the fast-changing state sequence and the slow-changing state sequence. The analysis process adopts the Pearson correlation coefficient algorithm, and quantifies the correlation strength by calculating the ratio of the covariance of the two state sequences to the product of their respective standard deviations. In the correlation calculation process, the system takes the voltage and current state sequence of the fast-changing device as the first variable group, and takes the mechanical state sequence of the slow-changing device as the second variable group, and calculates the local correlation coefficient in the specified time window through the sliding window technology. The generation process of the step coupling correlation degree coefficient comprehensively considers the physical coupling relationship between electrical quantities and mechanical quantities. When the electrical state change of the fast-changing device and the mechanical state change of the slow-changing device show high synchronism, the correlation degree coefficient is close to 1, when the change trends of the two are opposite, the correlation degree coefficient is close to -1, and when there is no obvious correlation between the two, the correlation degree coefficient is close to 0.

[0074] The threshold comparison process compares the calculated step coupling correlation degree coefficient with the preset threshold value, which is determined according to the stability requirements of the power system and the simulation accuracy requirements. When the absolute value of the correlation degree coefficient is greater than the threshold value, the system determines that there is a strong coupling relationship between the fast-changing device group and the slow-changing device group, and a smaller simulation step size is needed to ensure the coupling accuracy; when the absolute value of the correlation degree coefficient is less than the threshold value, the system determines that the coupling relationship between the two is weak, and a larger simulation step size can be used to improve the calculation efficiency. The coupling state evaluation result contains the coupling strength level, the coupling type identification, the recommended step size configuration and other information.

[0075] The continuous monitoring process organizes and stores the coupling state evaluation results in time series to form a dynamic evolution record of the coupling state. The monitoring process uses a circular buffer technique to maintain a fixed-length historical data window, and the new evaluation results continuously update the window content, and the expired data is automatically deleted. The generation of state deviation trend data is achieved by trend analysis on the continuous coupling state evaluation results, calculating the first-order difference and second-order difference of the correlation coefficient, and identifying the growth, decay or oscillation pattern of the coupling strength. When a sustained downward trend in the correlation coefficient is detected, the system warns of possible coupling failure problems; when the correlation coefficient fluctuates sharply, the system identifies possible numerical instability in the system.

[0076] In a specific embodiment, the S3 step further comprises:

[0077] Extracting voltage and current power transfer data between the fast-varying device group and the slow-varying device group to form coupling interface data;

[0078] Calculating the ratio of the standard deviation to the mean of the coupling interface data within a single simulation iteration period to determine the step data transfer variation coefficient;

[0079] Comparing the step data transfer variation coefficient with the preset variation threshold to determine the transient synchronization state of the coupling interface, and generating a transient accuracy evaluation index;

[0080] Establishing a mapping relationship between the transient accuracy evaluation index and the state deviation trend data to form a coupling accuracy correlation degree.

[0081] Specifically, for the data transfer process between the fast-varying device group and the slow-varying device group, key transfer information is extracted by monitoring the electrical parameter changes of the coupling interface. The extraction process of coupling interface data focuses on the voltage, current and power transfer characteristics of the interaction nodes of the two device groups. This process identifies the connection points between the output terminals of fast-varying devices and the input terminals of slow-varying devices, and collects the transient electrical quantities of these nodes. For voltage transfer data, the system records the amplitude and phase changes during the transfer of fast-varying device output voltage to slow-varying devices; for current transfer data, the system records the waveform distortion and frequency component changes of the current flowing between devices at different time scales; for power transfer data, the system records the transfer efficiency and loss distribution of active power and reactive power at the coupling interface. The formation process of coupling interface data organizes these electrical parameters in time series to establish a multi-dimensional data set containing voltage amplitude sequence, current effective value sequence and power instantaneous value sequence.

[0082] The calculation process of the step length data transmission variation coefficient is based on the statistical characteristic analysis of the coupling interface data in a single simulation iteration period. The calculation process selects the simulation iteration period as the analysis window, and in this time window, the system calculates the mean and standard deviation of the coupling interface data, and quantifies the dispersion degree of the data by the ratio of the standard deviation to the mean. The determination process of the variation coefficient reflects the signal distortion degree when the high-frequency dynamic characteristics of the fast-changing device are transmitted to the slow-changing device. When the high-frequency signal output by the fast-changing device is transmitted to the slow-changing device through the coupling interface, due to the huge difference in time constants of the two, signal attenuation, delay and waveform distortion will occur. The size of the variation coefficient value directly reflects the signal quality degradation degree in the transmission process. The larger the variation coefficient, the more serious the distortion in the signal transmission process, and the lower the coupling accuracy.

[0083] The comparison process compares the calculated step length data transmission variation coefficient with the preset variation threshold value. The threshold value is determined according to the requirements of the power system on the coupling accuracy and the acceptable error range of the simulation results. When the variation coefficient is less than the preset threshold value, the system determines that the transient synchronization state of the coupling interface is good, and the data transmission between the fast-changing device and the slow-changing device basically maintains the integrity of the signal; when the variation coefficient is greater than the preset threshold value, the system determines that there is significant transient distortion in the coupling interface, and the simulation parameters need to be adjusted or a more refined coupling algorithm needs to be used. The generation process of the transient accuracy evaluation index comprehensively considers the numerical value of the variation coefficient, the degree of exceeding the threshold value, the duration and other factors, and forms a comprehensive evaluation result including the accuracy level, the risk degree and the recommended measures.

[0084] The mapping relationship establishment process analyzes the correlation between the transient accuracy evaluation index and the state deviation trend data, and identifies the causal relationship and mutual influence mechanism between them. The mapping process establishes a mathematical relationship between the transient accuracy index and the long-term state deviation through a regression analysis algorithm. When the transient accuracy index shows that there is high-frequency distortion in the coupling interface, the system predicts the cumulative impact of the distortion on the long-term simulation stability. The formation process of the coupling accuracy correlation degree quantifies the correlation strength between the short-term transient accuracy and the long-term stability, which provides a decision basis for the adaptive adjustment of the simulation parameters. When the correlation degree exceeds the critical value, the system triggers the step length optimization mechanism to maintain the overall accuracy of the simulation.

[0085] In a specific embodiment, the S4 step further comprises:

[0086] A multi-objective optimization function including the step length coupling correlation coefficient and the step length data transmission variation coefficient is constructed, and an input feature vector is established based on the state variable mapping relationship;

[0087] The population individual coding of the ISCS-PSVR algorithm is initialized according to the input feature vector, and the search ranges of the fast-changing device step parameter and the slow-changing device step parameter are set based on the device time scale identification matrix;

[0088] The input feature vector is taken as the evaluation of the fitness function, and the iterative optimization process of the ISCS-PSVR algorithm is executed, and the fitness value of the population individual is calculated based on the coupling accuracy correlation degree and the transient accuracy evaluation index;

[0089] The step length configuration scheme corresponding to the population individual with the optimal fitness value is output from the optimization result guided by the input feature vector, and the step length coupling configuration parameter is determined.

[0090] Specifically, for the construction process of the multi-objective optimization function, the step length coupling correlation coefficient and the step length data transmission mutation coefficient are taken as the core optimization objectives, and a comprehensive optimization function is established by weighted summation. The construction process of the optimization function considers the balance between simulation accuracy and computational efficiency, wherein the correlation coefficient reflects the long-term coupling stability requirement, and the mutation coefficient reflects the short-term transmission accuracy requirement, and the two are linearly combined through a weight coefficient. The combination process of the state variable mapping relationship introduces the state conversion parameters of the fast-changing device and the slow-changing device as constraint conditions into the optimization function, ensuring that the optimization result meets the physical realizability requirement. The establishment process of the input feature vector integrates multi-dimensional information such as correlation coefficient, mutation coefficient, and mapping parameter, forming a comprehensive feature description containing current simulation state, historical trend, and physical constraints. The vector provides a complete input data set for subsequent algorithm processing.

[0091] The population individual coding initialization process of the ISCS-PSVR algorithm determines the coding scheme based on the dimension structure of the input feature vector, and each population individual corresponds to a specific set of step length configuration parameters. The coding initialization process uses real number coding, and the fast-changing device step parameter and the slow-changing device step parameter are respectively coded as different components of the vector, and the device time scale identification matrix is used to determine the value range of each parameter. The setting process of the search range determines the parameter boundary according to the physical limit of the device time constant, and the search range of the fast-changing device step parameter is set to one tenth of the minimum time constant of the device to the maximum time constant of the device, and the search range of the slow-changing device step parameter is set to the minimum time constant of the device to ten times the maximum time constant of the device. The population initialization process generates initial population individuals in the specified search range through a random number generator, ensuring that the initial population has sufficient diversity to cover the entire solution space.

[0092] The iterative optimization processing process takes the input feature vector as the core evaluation basis of the fitness function, and finds the optimal step configuration through the evolution mechanism of the ISCS-PSVR algorithm. The fitness evaluation process combines the coupling accuracy correlation and the transient accuracy evaluation index to evaluate the comprehensive performance of each population individual. The evaluation process predicts the simulation performance indicators corresponding to the current step configuration through the support vector regression model, including the numerical stability, the calculation complexity, and the coupling error and other key parameters. The improved search strategy of the ISCS algorithm adopts an adaptive mutation mechanism, dynamically adjusts the mutation probability and mutation step according to the fitness distribution of the current population, increases the mutation strength when the population convergence degree is high to maintain the search diversity, and reduces the mutation strength when the population dispersion degree is large to accelerate the convergence. In the iteration process, the algorithm continuously improves the population quality through selection, crossover and mutation operations until the convergence condition is met or the maximum iteration number is reached.

[0093] The output processing process identifies the population individual with the optimal fitness value from the optimization result, extracts the corresponding step configuration scheme as the final optimization result of the system. The output process not only considers the absolute size of the fitness value, but also evaluates the stability and robustness of the solution, and verifies the reliability of the optimal solution through multiple independent runs. The determination process of the step coupling configuration parameter converts the encoding value of the optimal population individual into specific simulation parameter settings, including the basic step of the fast-changing device, the step adjustment coefficient, the basic step of the slow-changing device, and the coupling synchronization period and other key parameters. The conversion process also includes parameter rationality verification to ensure that the step configuration obtained by optimization meets the numerical stability condition and the physical constraint condition.

[0094] The technical features are deeply coupled with the power system simulation function, the construction of the optimization function is directly aimed at the specific needs of the multi-time scale simulation of the power system, and the search strategy of the algorithm is specially designed to handle the time scale difference problem of the power equipment. Taking a microgrid system as an example, the system includes a photovoltaic inverter, an energy storage converter, a diesel generator and other devices, among which the switching frequency of the inverter is 10 kHz corresponding to a 100 microsecond time scale, and the electromechanical transient process of the generator corresponds to a 100 millisecond time scale. The multi-objective optimization function constructed by the system takes the coupling correlation coefficient between the inverter and the generator as the long-term stability index, and takes the variation coefficient of the power transmission process as the short-term accuracy index. The ISCS-PSVR algorithm dynamically optimizes the 100 microsecond basic step of the inverter and the 5 millisecond basic step of the generator according to the input feature vector of the current operating condition, automatically adjusts the step configuration to maintain the coupling accuracy when the photovoltaic output fluctuates, and increases the step to reduce the calculation burden when the load is stable.

[0095] In a specific embodiment, the process of initializing the population individual encoding of the ISCS-PSVR algorithm according to the input feature vector in the execution step can specifically include the following steps:

[0096] extracting step configuration dimension information in the input feature vector, determining the coding length and coding structure of the population individual;

[0097] randomly generating initial population individual values according to the coding structure of the population individual, forming a population individual coding matrix;

[0098] mapping the coding values in the population individual coding matrix into specific values of the fast-varying device step parameter and the slow-varying device step parameter;

[0099] verifying the rationality of the fast-varying device step parameter and the slow-varying device step parameter based on the device time scale identification matrix, generating an effective population individual set.

[0100] Specifically, the number and type of optimization variables are determined by analyzing the structural composition of the feature vector. This extraction process identifies components in the feature vector related to step configuration, including fast-varying device step configuration dimension, slow-varying device step configuration dimension, coupled synchronization period dimension, step adjustment coefficient dimension, and other key parameters. The determination process of the coding length sets the gene length of the population individual according to the number of identified configuration dimensions, with each configuration dimension corresponding to one or more gene bits in the coding vector. The design process of the coding structure adopts a segmented coding method, dividing the coding vector of the population individual into multiple functional segments, with fast-varying device related parameters occupying the first half of the coding vector, slow-varying device related parameters occupying the second half of the coding vector, and coupling parameters occupying the middle part of the coding vector. This structured coding method ensures the independence and operability between different types of parameters. The formation process of the population individual coding matrix generates the coding values of the initial population through a random number generation algorithm. This process uses a uniform distribution random generation strategy to ensure that the initial population has good diversity. The random generation process independently generates values for each coding position, with the coding value range set to a standardized interval between 0 and 1, providing a unified numerical basis for subsequent parameter mapping. The construction process of the population individual coding matrix arranges the coding vectors of multiple population individuals in rows to form a matrix structure, with the number of rows corresponding to the population size and the number of columns corresponding to the coding length. This matrix provides data structure support for subsequent evolutionary operations. The initialization process of the coding matrix also includes boundary condition checking to ensure that the generated coding values meet the basic constraint conditions of the algorithm.

[0101] The encoding numerical value mapping process converts the normalized numerical values in the population individual encoding matrix into specific step parameter numerical values, which realizes the conversion from the encoding space to the parameter space by using a linear transformation algorithm. The mapping process of the fast-varying device step parameter maps the corresponding components of the encoding vector to the step value range of the microsecond level, and the mapping relationship determines the upper and lower bounds of the mapping interval according to the time constant characteristics of the fast-varying device. The mapping process of the slow-varying device step parameter maps the corresponding components of the encoding vector to the step value range of the millisecond level, and the setting of the mapping interval takes into account the dynamic response characteristics and numerical stability requirements of the slow-varying device. The mapping process also includes parameter unit conversion and numerical precision adjustment to ensure that the mapping results meet the numerical requirements and physical meaning of simulation calculation. The reasonableness verification process performs physical feasibility test on the step parameters obtained by mapping based on the device time scale identification matrix. The verification process judges the reasonableness of the parameter configuration by comparing the relationship between the step parameter and the time constant of the device. The verification process checks whether the step parameter of the fast-varying device is less than the minimum time constant of the corresponding device, whether the step parameter of the slow-varying device is within a reasonable numerical range, and whether the proportional relationship between the step of the fast-varying device and the step of the slow-varying device meets the numerical stability condition. The generation process of the effective population individual set selects the population individuals that pass the reasonableness verification, eliminates the individuals that do not meet the physical constraint conditions, and ensures that all candidate solutions in the subsequent optimization process have actual operability. The screening process also includes a repeatability check to avoid the appearance of identical individuals in the population and maintain the diversity level of the population.

[0102] Taking a DC microgrid as an example, the system contains photovoltaic DC-DC converters, energy storage bidirectional converters, load converters and other power electronic devices. The MPPT control period of the photovoltaic converter is 100 microseconds, the power control period of the energy storage converter is 1 millisecond, and the voltage control period of the load converter is 10 milliseconds. The system determines the step parameters of the three types of devices to be optimized according to the input feature vector, sets the population individual code length to 6 bits, and encodes the photovoltaic converter step parameter in the first 2 bits, the energy storage converter step parameter in the middle 2 bits, and the load converter step parameter in the last 2 bits. The randomly generated population individual code matrix is converted into specific step value through linear mapping. The photovoltaic converter step is mapped to the range of 10-50 microseconds, the energy storage converter step is mapped to the range of 0.5-2 milliseconds, and the load converter step is mapped to the range of 5-20 milliseconds. The rationality verification process checks whether the step parameters of each device meet the time constant constraints of the corresponding device, eliminates parameter combinations that may cause numerical instability, and generates an initial population containing 100 valid individuals. By establishing a coding mapping system specifically for the characteristics of power system devices, it is ensured that the optimization algorithm can effectively search within the physically feasible parameter space. The correspondence between the coding structure and the device classification ensures the pertinence of the optimization process, and the rationality verification mechanism prevents the generation of infeasible solutions, significantly improving the efficiency and reliability of the optimization algorithm.

[0103] In a specific embodiment, the execution step of executing the ISCS-PSVR algorithm to perform iterative optimization processing based on the input feature vector as the evaluation basis of the fitness function can specifically include the following steps:

[0104] Substitute the step configuration corresponding to each individual in the valid population individual set into the multi-objective optimization function to calculate the simulation error value and the calculation efficiency value of the individual population individual;

[0105] According to the coupling accuracy correlation degree, the simulation error value of the single population individual is weighted, and the calculation efficiency value of the single population individual is corrected in combination with the transient accuracy evaluation index to form a comprehensive evaluation index of the single population individual;

[0106] Rank all individuals in the valid population individual set based on the comprehensive evaluation index of the single population individual to determine the fitness value ranking of each population individual;

[0107] Select the population individual with a high fitness value ranking to perform crossover and mutation operations, update the population individual code matrix, and generate the next generation of population individuals.

[0108] Specifically, for the performance evaluation process of each individual in the effective population individual set, the step configuration parameters corresponding to the individual are substituted into the multi-objective optimization function for quantitative calculation. The calculation process evaluates the performance of the current step configuration by running a small-scale simulation test. The simulation error value is calculated by comparing the deviation of the simulation result from the theoretical reference value to quantify the precision loss. This error value reflects the ability of the current step configuration to maintain simulation accuracy. The calculation efficiency value is calculated by measuring the calculation time and resource consumption required to complete a fixed simulation duration to quantify the calculation cost. This efficiency value reflects the effect of the current step configuration in reducing computational complexity. The performance evaluation process of a single population individual considers the simulation performance of fast-changing devices and slow-changing devices, and combines the error value and efficiency value of the two types of devices into a comprehensive performance indicator of the individual through weighted averaging. The weighting process adjusts the simulation error value of a single population individual according to the coupling precision correlation degree. This process applies the correlation degree as a weight coefficient to the error value calculation. When the coupling precision correlation degree is high, it indicates that the coupling relationship between fast-changing devices and slow-changing devices is strong, and the system increases the weight of the simulation error value to emphasize the precision requirement. When the correlation degree is low, it indicates that the coupling relationship between devices is weak, and the system reduces the error value weight to allow moderate precision sacrifice. The correction process adjusts the calculation efficiency value of a single population individual in combination with the transient precision evaluation index. This process dynamically adjusts the importance of the efficiency value according to the numerical value of the transient precision index. When the transient precision index shows that there is significant distortion on the coupling interface, the system reduces the weight of the calculation efficiency value to prioritize simulation accuracy. When the transient precision index shows good coupling quality, the system increases the efficiency value weight to pursue computational performance optimization. The formation process of the comprehensive evaluation index obtains a single evaluation value through linear combination of the weighted error value and the corrected efficiency value. This index comprehensively reflects the balance performance of the individual in the precision and efficiency dimensions.

[0109] The sorting process ranks all individuals in the effective population individual set based on the comprehensive evaluation index of a single population individual. This sorting process uses descending order arrangement, with individuals having larger comprehensive evaluation index values arranged in the front and individuals having smaller values arranged in the back. Through this sorting mechanism, individuals with excellent performance and individuals with poor performance are identified. The determination process of the fitness value ranking converts the sorting position into a fitness value, with individuals ranking higher obtaining higher fitness values and individuals ranking lower obtaining lower fitness values. The allocation process of the fitness value adopts a linear decreasing allocation strategy to ensure that the fitness difference between individuals accurately reflects the degree of performance difference, providing accurate probability basis for subsequent selection operations.

[0110] The selection operation process selects the population individuals with good performance as the parent individuals for genetic operation according to the fitness value ranking, and the selection process adopts a roulette wheel selection algorithm, and the probability of selection of an individual is proportional to the fitness value thereof. The crossover operation process pairs the selected parent individuals to generate new offspring individuals through gene recombination, and the crossover process adopts a multi-point crossover strategy to exchange genes at multiple positions of the encoding vector, so as to ensure that the offspring individuals can inherit the excellent characteristics of the parent individuals. The mutation operation process randomly disturbs the encoding of the offspring individuals, and the mutation process adopts an adaptive mutation strategy to dynamically adjust the mutation probability and mutation strength according to the convergence degree of the current population. The updating process of the population individual encoding matrix replaces the individuals with poor performance with the newly generated offspring individuals, and the optimal individuals are ensured to be passed to the next generation through an elite reservation strategy, and new genetic materials are introduced to maintain the population diversity. The generation process of the next generation population individuals completes a complete evolution iteration, and provides an improved candidate solution set for continuous optimization search.

[0111] The technical features are deeply coupled with the power system simulation optimization function, the performance evaluation process is directly designed to evaluate the indicators according to the specific needs of the power system simulation, and the weighted correction mechanism dynamically adjusts the optimization target according to the coupling characteristics of the power equipment. Taking an AC-DC hybrid microgrid as an example, the system includes a synchronous generator on the AC bus side, a photovoltaic array on the DC bus side, a bidirectional converter connecting the two buses and the like. The electromechanical transient time constant of the generator is 200 milliseconds, the MPPT control time constant of the photovoltaic is 50 microseconds, and the power control time constant of the bidirectional converter is 1 millisecond. When the system evaluates the step length configuration scheme corresponding to a population individual, the generator step length is set to 10 milliseconds, the photovoltaic control step length is set to 10 microseconds, and the converter step length is set to 0.5 milliseconds. The error value and efficiency value of the configuration are calculated by running a simulation test including a power disturbance scene. When it is detected that the coupling precision degree of power transmission between the AC-DC buses is high, the system increases the weight coefficient of the simulation error value, and when the transient precision evaluation index shows that there is high-frequency oscillation in the output of the converter, the system reduces the importance of the calculation efficiency value. Finally, the comprehensive evaluation index of the individual is 0.82. By performing similar evaluation and sorting on all 100 population individuals, the system selects the top 50 individuals for crossover and mutation operation to generate the next generation population to continue optimization search.

[0112] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the foregoing embodiments of the present application have been described in detail, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A step-size data coupling simulation method for devices with multiple time scales, characterized in that, The method includes: Step S1: Classify power system equipment through time constant analysis to obtain fast-changing equipment groups and slow-changing equipment groups, and establish a step-size coupling state vector, including: extracting time constants for power electronic devices and traditional power equipment in the power system to obtain equipment time scale identification matrices; grouping and classifying microsecond-level and millisecond-level equipment according to the equipment time scale identification matrix to obtain the fast-changing equipment group and the slow-changing equipment group; constructing vectors based on the voltage and current states of the fast-changing equipment group and the mechanical states of the slow-changing equipment group to obtain the step-size coupling state vector; and mapping the step-size coupling state vector through coordinate transformation to obtain the state variable mapping relationship. Step S2: Based on the step-size coupling state vector, calculate the correlation coefficient for the fast-changing equipment group and the slow-changing equipment group to obtain the step-size coupling correlation coefficient. Through unified time reference processing, map the equipment states at different time scales to a synchronous time point sequence, eliminating the impact of time scale differences on data analysis. The correlation coefficient calculation uses the Pearson correlation algorithm to quantify the linear correlation between the fast-changing and slow-changing equipment state sequences, generating the step-size coupling correlation coefficient. Based on the step-size coupling correlation coefficient, perform threshold comparison processing to obtain the coupling state evaluation result. Continuously monitor the coupling state evaluation result according to the time series to obtain state deviation change trend data. Step S3: Perform coefficient of variation analysis on the coupling interface data of the fast-changing equipment group and the slow-changing equipment group to obtain the step-size data transmission coefficient of variation. The calculation process of the step-size data transmission coefficient of variation is based on the statistical characteristic analysis of the coupling interface data within a single simulation iteration cycle. By selecting the simulation iteration cycle as the analysis window, the mean and standard deviation of the coupling interface data are calculated within the analysis window. The ratio of the standard deviation to the mean is used to quantify the dispersion of the data. The determination process of the step-size data transmission coefficient of variation reflects the degree of signal distortion when the high-frequency dynamic characteristics of the fast-changing equipment are transmitted to the slow-changing equipment. Compare the magnitude of the step-size data transmission coefficient of variation with the preset variation threshold to determine the transient synchronization state of the coupling interface and generate a transient accuracy evaluation index. Establish a mapping relationship between the transient accuracy evaluation index and the state deviation change trend data to form a coupling accuracy correlation. The coupling interface data is based on the voltage, current, and power transfer characteristics of the interaction nodes of the two types of equipment groups. It identifies the connection point between the output end of the fast-changing equipment and the input end of the slow-changing equipment, and collects the transient electrical quantities of the connection point. For voltage transfer data, it records the amplitude and phase changes during the transfer of the output voltage of the fast-changing equipment to the slow-changing equipment. For current transfer data, it records the waveform distortion and frequency component changes when the current flows between the equipment at different time scales. For power transfer data, it records the transfer efficiency and loss distribution of active and reactive power at the coupling interface. Step S4: Optimize the step size coupling correlation coefficient and the step size data transfer variation coefficient using the ISCS-PSVR algorithm to obtain step size coupling configuration parameters. This includes: constructing a multi-objective optimization function containing the step size coupling correlation coefficient and the step size data transfer variation coefficient; establishing an input feature vector based on the state variable mapping relationship; initializing the population individual encoding of the ISCS-PSVR algorithm based on the input feature vector; setting the search range for fast-changing device step size parameters and slow-changing device step size parameters based on the device timescale identifier matrix; performing iterative optimization processing of the ISCS-PSVR algorithm using the input feature vector as the evaluation basis for the fitness function; calculating the fitness value of the population individuals based on the coupling accuracy correlation and the transient accuracy evaluation index; outputting the step size configuration scheme corresponding to the population individual with the optimal fitness value from the optimization results guided by the input feature vector, and determining the step size coupling configuration parameters.

2. The step-size data coupling simulation method for multi-timescale devices according to claim 1, characterized in that, The initialization of the population individual encoding for the ISCS-PSVR algorithm based on the input feature vector includes: Extract the step size configuration dimension information from the input feature vector to determine the encoding length and encoding structure of individual populations; Initial population individual values ​​are randomly generated based on the encoding structure of the population individuals, forming a population individual encoding matrix; The encoded values ​​in the population individual encoding matrix are mapped to the specific values ​​of the fast-change device step size parameter and the slow-change device step size parameter; The rationality of the step size parameters of the rapidly changing device and the step size parameters of the slowly changing device is verified based on the device time scale identifier matrix, and an effective population set is generated.

3. The step-size data coupling simulation method for multi-timescale devices according to claim 2, characterized in that, The step of using the input feature vector as the evaluation criterion for the fitness function and performing iterative optimization processing of the ISCS-PSVR algorithm, calculating the fitness value of individuals in the population based on the coupling accuracy correlation and the transient accuracy evaluation index, includes: Substitute the step size configuration corresponding to each individual in the effective population set into the multi-objective optimization function to calculate the simulation error value and computational efficiency value of a single population individual. The computational efficiency value is quantified by measuring the computation time and computational resource consumption required to complete a fixed simulation duration to quantify the computational cost. The simulation error value of the individual population is weighted according to the coupling accuracy correlation degree, and the computational efficiency value of the individual population is corrected by combining the transient accuracy evaluation index to form a comprehensive evaluation index of the individual population. All individuals in the effective population set are ranked based on the comprehensive evaluation index of the individual in the single population, and the fitness value ranking of each population individual is determined. Select individuals with high fitness values ​​from the population and perform crossover and mutation operations to update the population's individual coding matrix, thereby generating the next generation of population individuals.

Images