A joint operation and information data interaction processing method for a multi-simulation device

Abstract

The application relates to the technical field of data processing, and discloses a joint operation and information data interaction processing method for multiple simulation devices. The method comprises the following steps: collecting electrical quantity data of boundary nodes of multiple simulation devices, calculating electrical distance weight factors, and dividing the boundary nodes into three categories of strong coupling, medium coupling and weak coupling; predicting the electrical state change trend based on the historical data of the nodes; solving a data interaction strategy optimization solution set through predictive boundary state estimation; assigning time stamps to the boundary data, dynamically adjusting the transmission priority, and generating a transmission queue; and performing data exchange according to the queue by each device, and monitoring the synchronization accuracy and response performance. The application realizes intelligent hierarchical management and dynamic priority scheduling of boundary data, and solves the problems of inconsistency and insufficient real-time performance of boundary node data in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a joint operation and information data interaction processing method for multiple simulation devices. BACKGROUND

[0002] With the continuous expansion of the scale of the power system and the rapid development of new power systems, large-scale power system simulation has become an important technical means for power system planning, operation and control. In the prior art, the joint operation of multiple simulation devices mainly adopts a static system partitioning and fixed task allocation method, which divides a complex power system model into multiple relatively independent subsystems according to geographical location or electrical connection characteristics, and respectively deploys them on different digital real-time simulation devices for parallel computing. This kind of method usually establishes an inter-device connection based on Ethernet or a special communication bus, uses a periodic data exchange mechanism to realize the transmission of boundary information, and ensures the convergence of the whole system through a master device or a distributed coordination algorithm. In terms of data interaction, the prior art mainly relies on simple FIFO (First-In-First-Out) queue management and fixed priority scheduling strategy, and performs synchronous calculation and data exchange according to the preset time step. In addition, the existing multi-device coordination method also includes global clock management based on time synchronization protocol, simple task scheduling based on load monitoring, and data synchronization mechanism based on communication delay compensation.

[0003] However, the prior art has significant deficiencies, mainly manifested in the lack of a systematic solution to the inconsistency of boundary node data, and the static system partitioning and task allocation strategy cannot adapt to the dynamic changes of the operating conditions of the power system. When new energy output fluctuation, load mutation or equipment failure occurs, the computing load of each simulation device becomes seriously unbalanced, resulting in idle computing resources in some devices and overload in other devices. More critically, the prior art lacks effective boundary data prediction capability and dynamic priority adjustment mechanism, and cannot intelligently manage data interaction according to the electrical coupling strength and data importance of the boundary nodes, often resulting in data delay and synchronization error problems when processing electromagnetic transient level simulation, which seriously affects the real-time performance and accuracy of the joint simulation of multiple devices. SUMMARY

[0004] The present application provides a joint operation and information data interaction processing method for multiple simulation devices, which is used to realize intelligent hierarchical management and dynamic priority scheduling of boundary data, and solves the problems of boundary node data inconsistency and insufficient real-time performance in the prior art.

[0005] The present application provides a joint operation and information data interaction processing method for multiple simulation devices, which comprises:

[0006] S1 step, collect the boundary node voltage amplitude, phase angle and current data of multiple simulation devices, calculate the electrical distance weight factor to determine the coupling strength, and divide the boundary nodes into strong coupling boundary nodes, medium coupling boundary nodes and weak coupling boundary nodes;

[0007] S2 step, respectively based on the historical electrical quantity data of the strong coupling boundary nodes, medium coupling boundary nodes and weak coupling boundary nodes, predict the boundary node electrical state change trend in the future time step;

[0008] S3 step, the boundary node electrical state change trend is solved through the predictive boundary state estimation to obtain the boundary data interaction strategy optimization solution set;

[0009] S4 step, time stamp marks are assigned to the boundary data in the optimization solution set, and the transmission priority is dynamically adjusted through the electrical distance weight factor to generate a boundary data transmission queue;

[0010] S5 step, each simulation device performs data exchange operation according to the boundary data transmission queue, and monitors the data synchronization accuracy and system response performance index.

[0011] The technical scheme provided in the application solves the problem of lack of quantitative evaluation of the importance of boundary nodes in the prior art by collecting the voltage amplitude, phase angle and current data of the boundary nodes of multiple simulation devices and calculating the electrical distance weight factor to determine the coupling strength. The electrical distance weight factor comprehensively considers the equivalent impedance and physical distance parameters between the boundary nodes, can accurately reflect the influence degree of different boundary nodes on the overall simulation accuracy, divides the boundary nodes into three levels of strong coupling, medium coupling and weak coupling, realizes hierarchical management and differentiated processing strategies of boundary data, and ensures the transmission quality and timeliness of key boundary data by giving the strongest data processing priority and the strictest synchronization requirement to the strong coupling boundary nodes. The technical scheme based on the historical electrical quantity data of the boundary nodes with different coupling strengths respectively predicts the trend of the electrical state of the boundary nodes in the future time step, overcomes the lack of boundary data prediction capability in the prior art, accurately predicts the voltage amplitude, phase angle and current variation law of each boundary node in the future multiple time steps by establishing an improved characteristic line equation parameter matrix and boundary condition correction term calculation, provides a reliable theoretical basis for subsequent data interaction strategy formulation, and significantly reduces the synchronization error and delay problems caused by unpredictable data. The technical scheme of solving the boundary data interaction strategy optimization solution set through the predictive boundary state estimation of the trend of the electrical state of the boundary nodes solves the problem of lack of systematic optimization target in the prior art, constructs three objective functions of voltage deviation minimization, phase angle synchronization error minimization and device load balancing, solves the multi-objective optimization problem by using the Pareto frontier analytical processing method, obtains a comprehensive optimal solution set considering simulation accuracy, synchronization performance and calculation efficiency, and avoids the local optimum and performance bias problems caused by single objective optimization. The technical scheme of assigning time stamp labels to the boundary data in the optimal solution set and generating a boundary data transmission queue by dynamically adjusting the transmission priority according to the electrical distance weight factor solves the problem of lack of intelligent management of data transmission in the prior art, ensures the accurate timing relationship of data packets through nanosecond-level time stamp labels, dynamically adjusts the priority value according to the importance weight coefficient and transmission urgency of the boundary data in real time, and realizes efficient queuing management and orderly transmission of boundary data by combining the hierarchical cache architecture and intelligent discard strategy, which significantly improves the real-time performance and reliability of data exchange. The technical scheme of performing data exchange operations according to the boundary data transmission queue by each simulation device and monitoring the data synchronization accuracy and system response performance indicators solves the problem of lack of dynamic performance monitoring and feedback adjustment mechanism in the prior art, realizes the coordinated joint operation of multiple devices by establishing a distributed coordination control architecture and a device load redistribution strategy, and realizes real-time monitoring of key performance indicators such as data synchronization accuracy and system response time, which provides timely and accurate feedback information for system optimization and fault diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments description. 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 any creative effort based on these drawings.

[0013] Figure 1 An embodiment of the method for joint operation and information data interaction processing of multiple simulation devices in the embodiments of the present application. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a method for joint operation and information data interaction processing of multiple simulation devices. 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 have to be used to describe a specific order or sequence. It should be understood that the data thus used 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 terms "include" or "have" and any variations thereof are 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 only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0015] For the convenience of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 An embodiment of the method for joint operation and information data interaction processing of multiple simulation devices in the embodiments of the present application includes:

[0016] S1, collecting the boundary node voltage amplitude, phase angle and current data of multiple simulation devices, calculating the electrical distance weight factor to determine the coupling strength, and dividing the boundary nodes into strong coupling boundary nodes, medium coupling boundary nodes and weak coupling boundary nodes;

[0017] S2, predicting the boundary node electrical state change trend in the future time step based on the historical electrical quantity data of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes respectively;

[0018] S3, solving the boundary data interaction strategy optimization solution set through the predictive boundary state estimation of the boundary node electrical state change trend;

[0019] S4, assigning a time stamp to the boundary data in the optimization solution set, dynamically adjusting the transmission priority through the electrical distance weight factor to generate a boundary data transmission queue;

[0020] S5, each simulation device performs data exchange operations according to the boundary data transmission queue, and monitors data synchronization accuracy and system response performance indicators.

[0021] It can be understood that the execution subject of the present application can be a joint operation and information data interaction processing system for multiple simulation devices, and can also be a terminal or a server, which is not limited here. The server is taken as an example for description of the embodiments of the present application.

[0022] Specifically, the electrical quantity data of the boundary nodes of each digital real-time simulation device RT1000 is collected to obtain original data sets of the voltage amplitude, phase angle and current of the boundary nodes, which contain real-time state information of the electrical connection points between devices. Then, the equivalent impedance and physical distance parameters between the boundary nodes are calculated based on the original data sets, the equivalent impedance reflects the tightness of the electrical connection, and the physical distance parameter represents the spatial distribution relationship between devices. The electrical distance weight factor is obtained by multiplying the reciprocal of the equivalent impedance with the reciprocal of the physical distance coefficient. Then, the electrical distance weight factor is compared and analyzed with the preset coupling strength threshold value, when the weight factor exceeds the high threshold value, it is classified as strong coupling, when it is between the high and low threshold values, it is classified as medium coupling, and when it is lower than the low threshold value, it is classified as weak coupling. The boundary nodes are classified and marked according to the coupling strength level, and the classification results of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes are obtained.

[0023] The time series of the historical electrical quantity data of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes are constructed, and the historical data is arranged in time sequence to form continuous voltage amplitude time series, phase angle time series and current time series. Based on these time series, an improved characteristic line equation parameter matrix is established, the characteristic line equation describes the propagation law of electromagnetic waves on the transmission line, and the parameter matrix contains key parameters such as wave propagation speed and characteristic impedance. Then, the electromagnetic transient wave propagation analysis is performed on the boundary node electrical state transition law parameters through boundary condition correction term calculation, the boundary condition correction term considers the mutual influence between multiple devices, and the voltage amplitude prediction value, the phase angle prediction value and the current prediction value of each boundary node in the future time step are obtained through wave propagation analysis calculation. Finally, the boundary node state change vector is constructed according to these prediction values, the state change vector contains the change trend information of the voltage, the phase angle and the current, and the boundary node electrical state change trend is formed.

[0024] The boundary node electrical state change trend is calculated with the real-time voltage amplitude of each simulation device, and the difference between the predicted value and the actual value is compared point by point to construct a boundary node voltage amplitude deviation matrix, which reflects the voltage synchronization accuracy of each boundary point. At the same time, based on the boundary node electrical state change trend, the phase angle difference between devices is calculated and the synchronization error accumulation operation is processed, the phase angle difference represents the degree of phase shift between devices, and the cumulative operation superimposes the errors at multiple time points to obtain the phase angle synchronization error cumulative value. In addition, according to the boundary node electrical state change trend, the calculation load of each simulation device is processed by variance analysis, and by statistical analysis of the dispersion degree of the load of each device, the load imbalance metric value between devices is calculated. The boundary node voltage amplitude deviation matrix, the phase angle synchronization error cumulative value and the load imbalance metric value between devices are calculated by weighting coefficient calculation for Pareto frontier analysis, and the Pareto frontier analysis is a method for finding non-dominated solutions in multi-objective optimization. By weighing the relationship between the three objective functions, the boundary data interaction strategy optimization solution set is obtained.

[0025] The boundary data in the boundary data interaction strategy optimization solution set is assigned a nanosecond-level timestamp, and the timestamp marker ensures the time sequence of the data packet, and the timestamp marker boundary data packet is obtained. Then, based on the electrical distance weight factor, the importance weight coefficient of each boundary data is calculated and combined with the data transmission urgency to perform priority operation processing, the importance weight coefficient reflects the influence degree of the data on the overall simulation accuracy, and the transmission urgency considers the timeliness requirement of the data, and the combination of the two calculates the boundary data dynamic priority value. Then, the boundary data dynamic priority value and the data packet size parameter are analyzed for transmission scheduling, the transmission scheduling strategy is developed by considering the priority and the data packet size, and the boundary data transmission sorting strategy is obtained. Finally, according to the boundary data transmission sorting strategy, the timestamp marker boundary data packet is processed for queue construction, and the ordered queue is established according to the transmission priority and the time sequence, and the boundary data transmission queue is obtained.

[0026] The distributed coordination control architecture is established according to the boundary data transmission queue, and the running state of each simulation device is monitored and processed. The distributed coordination control architecture adopts a master-slave structure, the master device is responsible for global coordination, and the slave device is responsible for specific execution. The device calculation load rate, communication load rate and boundary data synchronization error parameters are obtained through real-time monitoring. Then, based on these parameters, load balancing analysis and operation processing are performed. By comparing the load conditions of each device, the devices with excessively high or low load are identified, a load redistribution scheme is developed, and a device-to-device load redistribution strategy is obtained. Then, the device-to-device load redistribution strategy is coordinated and matched with the boundary data transmission queue to ensure the coordination and consistency of data transmission and load distribution, and data exchange execution instructions for each simulation device are generated. According to the data exchange execution instructions, the data synchronization accuracy and system response time of each simulation device are monitored and processed. Through continuous monitoring of the data consistency and response delay between devices, the data synchronization accuracy and system response performance indicators are obtained.

[0027] In a specific embodiment, the S1 step includes:

[0028] The boundary nodes of each simulation device are subjected to electrical quantity data acquisition processing to obtain the original data set of the voltage amplitude, phase angle and current of the boundary nodes;

[0029] Based on the original data set, the equivalent impedance and physical distance parameters between each boundary node are calculated to obtain the electrical distance weight factor;

[0030] The electrical distance weight factor is compared and analyzed with the preset coupling strength threshold to obtain the coupling strength level of each boundary node;

[0031] According to the coupling strength level, the boundary nodes are classified and marked to obtain the classification results of the strongly coupled boundary nodes, the moderately coupled boundary nodes and the weakly coupled boundary nodes.

[0032] Specifically, in the electrical quantity data acquisition processing of the boundary nodes of each simulation device, the data acquisition module built-in each simulation device is used to monitor the boundary nodes in real time. The boundary nodes refer to the electrical connection points connecting different simulation devices, which bear the function of electrical information transmission between devices. The data acquisition module continuously acquires the voltage amplitude, phase angle and current of the boundary nodes according to the preset sampling frequency. The voltage amplitude reflects the voltage size of the node, the phase angle represents the phase information of the voltage waveform, and the current represents the current intensity through the node. The synchronous sampling technology is adopted in the acquisition process to ensure the consistency of the data collected by each device in time. After analog-to-digital conversion and filtering processing, the collected data form the original data set containing time stamp, node identification, voltage amplitude, phase angle and current value. These original data sets constitute the basic data source for subsequent processing.

[0033] The process of calculating the equivalent impedance and physical distance parameters between each boundary node based on the original data set involves complex electrical parameter calculation. The equivalent impedance refers to the parameter of simplifying a complex electrical network into an equivalent impedance, which is calculated by analyzing the voltage and current relationship between the boundary nodes. The calculation process extracts the voltage and current data of adjacent boundary nodes in the original data set, and then applies the extended form of Ohm's law to calculate the impedance value between the nodes. That is, the resistance component is obtained by dividing the voltage difference by the current difference, and the reactance component is obtained by calculating the phase angle difference. The modulus and amplitude angle of the equivalent impedance are obtained by combining the two. The physical distance parameter is determined according to the physical layout and connection mode of the simulation device, including the straight-line distance between devices, cable length, and signal propagation path length. Then, the reciprocal of the equivalent impedance is multiplied by the reciprocal of the physical distance coefficient, which is a weight coefficient set according to the distance. The closer the distance, the larger the coefficient. By this calculation method, the electrical distance weight factor is obtained, which comprehensively reflects the electrical closeness and physical connection strength between the boundary nodes.

[0034] Comparing and analyzing the electrical distance weight factor with the preset coupling strength threshold is a key step in classification. The preset coupling strength threshold includes two values: strong coupling threshold and weak coupling threshold. These two thresholds are determined based on experience and theoretical analysis of power system simulation. In the comparison and analysis process, the electrical distance weight factor of each boundary node is compared with the two threshold values. When the weight factor is greater than the strong coupling threshold, it indicates that there is strong electrical interaction between the boundary node and the adjacent node. When the weight factor is less than the weak coupling threshold, it indicates that the electrical interaction is weak. When the weight factor is between the two thresholds, it indicates that there is moderate electrical interaction. Through this threshold comparison method, each boundary node is assigned a corresponding coupling strength level identifier. The coupling strength level directly reflects the importance of the node in multi-device joint simulation and the urgency of data interaction.

[0035] The classifying and marking process of the boundary nodes according to the coupling strength levels is a link of the whole classifying process, and the classifying and marking process converts the coupling strength levels obtained in the previous step into specific node class identification. The processing process establishes a node classifying database, the database includes fields of node number, device identification, coupling strength level and classifying result, etc., the strongly coupled boundary nodes are marked as high-priority processing class, the nodes in this class enjoy the highest transmission priority and the most stringent synchronization requirement in the data interaction process, the moderately coupled boundary nodes are marked as moderate-priority processing class, and the weakly coupled boundary nodes are marked as low-priority processing class. The classifying and marking also includes color coding and graphical identification, the strongly coupled nodes are marked with red color, the moderately coupled nodes are marked with yellow color, and the weakly coupled nodes are marked with green color, and the visualized marking facilitates the running personnel to intuitively identify the boundary nodes of different classes. The obtained classifying result contains detailed classifying information of all the boundary nodes, and forms a hierarchical node management system.

[0036] In a specific embodiment, the S2 step comprises:

[0037] The time series construction process is performed on the historical electrical quantity data of the strongly coupled boundary nodes, the moderately coupled boundary nodes and the weakly coupled boundary nodes, and the voltage amplitude time series, the phase angle time series and the current time series of the boundary nodes of different classes are obtained.

[0038] The improved characteristic line equation parameter matrix is established based on the voltage amplitude time series, the phase angle time series and the current time series, and the boundary node electrical state transition law parameters are obtained.

[0039] The electromagnetic transient wave propagation analysis processing is performed on the boundary node electrical state transition law parameters through the boundary condition correction term calculation, and the voltage amplitude prediction value, the phase angle prediction value and the current prediction value of each boundary node in the future time step are obtained.

[0040] The boundary node state change vector is constructed according to the voltage amplitude prediction value, the phase angle prediction value and the current prediction value, and the boundary node electrical state change trend is obtained.

[0041] Specifically, in the time series construction process of the historical electrical quantity data of the strong coupling boundary node, the medium coupling boundary node and the weak coupling boundary node, the electrical quantity records of the boundary nodes of different coupling levels are extracted from the historical database of each simulation device. The historical electrical quantity data contains the continuous records of the voltage amplitude, phase angle and current of the boundary nodes in the past period of time. The time series construction process reorganizes these historical data in chronological order to form an ordered data sequence with time as the horizontal axis and electrical quantity value as the vertical axis. The construction process corrects the time synchronization of the historical data to eliminate the clock deviation between different devices, and then resamples the data according to a unified time interval to ensure that each sequence has the same time resolution. For the strong coupling boundary node, the constructed time series has the highest sampling density and the longest historical span. The time series of the medium coupling and weak coupling boundary nodes correspondingly reduces the sampling density, but maintains sufficient data length to meet the prediction requirements. The obtained voltage amplitude time series reflects the change law of voltage with time, the phase angle time series reflects the evolution trend of voltage phase, and the current time series reflects the time domain characteristics of current intensity. These three types of time series constitute the data basis for subsequent parameter calculation.

[0042] The process of establishing the improved characteristic line equation parameter matrix based on the voltage amplitude time series, the phase angle time series and the current time series involves complex mathematical modeling. The characteristic line equation is a partial differential equation describing the propagation law of electromagnetic waves in transmission media. The improved characteristic line equation adds boundary condition correction and multi-device coupling effect to the traditional equation. The parameter matrix establishment process analyzes the statistical characteristics of each time series, calculates the mean, variance, autocorrelation function and cross-correlation function. These statistical parameters reflect the change law and mutual relationship of electrical quantities. Then, the least squares method is used to fit the time series data to determine the key parameters in the characteristic line equation, including wave propagation speed, characteristic impedance, attenuation coefficient and phase constant. The wave propagation speed is calculated by analyzing the propagation delay of voltage and current waveforms between different boundary nodes. The characteristic impedance is determined by the ratio of voltage amplitude to current amplitude. The attenuation coefficient reflects the degree of energy loss of the signal in the propagation process. The phase constant describes the phase characteristics of wave propagation. These parameters form the parameter matrix, with rows corresponding to different boundary nodes and columns corresponding to different characteristic line equation parameters. The obtained boundary node electrical state transfer law parameters completely describe the electrical behavior mode of each boundary node.

[0043] The boundary condition correction term is a correction factor introduced to consider the mutual influence among multiple devices, and the calculation of the correction term is based on the operating state and coupling strength of the adjacent devices. The electromagnetic transient wave propagation analysis process solves the improved characteristic line equation using a numerical calculation method. In the calculation process, the state of the boundary node at the current time is taken as the initial condition, combined with the boundary condition correction term and the electrical state transition law parameters, and the state evolution of the boundary node in the future time step is obtained through iterative calculation. The analysis process considers the reflection, refraction and transmission of waves. When electromagnetic waves propagate to the boundary of a device, part of the energy is reflected back to the original device, and part of the energy is transmitted to the adjacent device. The boundary condition correction term quantifies these complex boundary effects. The calculated voltage amplitude prediction value describes the voltage amplitude change of the boundary node at each future time, the phase angle prediction value gives the expected evolution trajectory of the voltage phase, and the current prediction value predicts the change trend of the boundary current.

[0044] The construction of the boundary node state change vector according to the voltage amplitude prediction value, the phase angle prediction value and the current prediction value is the output link of the entire prediction process. The state change vector is a multi-dimensional vector containing multiple electrical quantity prediction values, and each component of the vector corresponds to the prediction value of the voltage amplitude, the phase angle and the current. The construction process arranges the three types of prediction values in chronological order to form a prediction value sequence, and then combines the prediction value sequences of different electrical quantities into a multi-dimensional vector structure. The vector construction also includes the calculation of the change trend. By analyzing the first-order difference and the second-order difference of the prediction value sequence, the change rate and the change acceleration of the electrical quantity are obtained. These derivative information reflects the dynamic characteristics of the boundary node state. The state change vector also contains uncertainty information. By calculating the confidence interval and the error range of the prediction value, the reliability of the prediction result is quantified. The obtained boundary node electrical state change trend is a comprehensive description containing prediction value, change rate, change acceleration and uncertainty information. This trend information accurately describes the evolution law of the electrical behavior of each boundary node in the future time period.

[0045] In a specific embodiment, the S3 step comprises:

[0046] The boundary node electrical state change trend is calculated with the real-time voltage amplitude of each simulation device to obtain a boundary node voltage amplitude deviation matrix;

[0047] The phase angle difference between each device is calculated based on the boundary node electrical state change trend, and the synchronization error accumulation value is obtained by performing synchronization error accumulation operation processing;

[0048] The load variance of each simulation device is analyzed based on the boundary node electrical state change trend to obtain a load imbalance metric value between devices;

[0049] The boundary node voltage amplitude deviation matrix, the phase angle synchronization error cumulative value and the inter-device load imbalance measure value are subjected to Pareto frontier analysis processing through weighted coefficient calculation to obtain a boundary data interaction strategy optimization solution set.

[0050] Specifically, in the deviation calculation processing of the boundary node electrical state change trend and the real-time voltage amplitude of each simulation device, the real-time voltage amplitude data at the current time is obtained from each simulation device. The real-time voltage amplitude is the instantaneous voltage measurement value of the boundary node of each device, reflecting the current operating state of the device. The deviation calculation processing compares the voltage amplitude prediction value in the boundary node electrical state change trend obtained in the previous step with the real-time measured voltage amplitude point by point, and the calculation process adopts two calculation methods of absolute deviation and relative deviation. The absolute deviation is obtained by subtracting the measured value from the prediction value to obtain the absolute value of the deviation, and the relative deviation is obtained by dividing the absolute deviation by the measured value to obtain the percentage representation of the deviation. The calculated deviation value is arranged in a matrix according to the boundary node number and the device identifier. The rows of the matrix correspond to different boundary nodes, and the columns correspond to different simulation devices. Each matrix element contains the voltage deviation value of the boundary node on the corresponding device. The boundary node voltage amplitude deviation matrix not only contains the numerical information of the deviation, but also records the time variation characteristics and spatial distribution pattern of the deviation, providing a quantitative index of voltage synchronization accuracy for subsequent multi-objective optimization.

[0051] The calculation of the phase angle difference between devices based on the boundary node electrical state change trend and the synchronization error accumulation operation processing is a key link to solve the phase angle synchronization problem. The phase angle difference reflects the phase difference of voltage waveforms between different devices, and the phase angle synchronization error directly affects the accuracy of multi-device joint simulation. The calculation process extracts the phase angle prediction value of the boundary node of each device from the boundary node electrical state change trend, and then calculates the phase angle difference between adjacent devices. The phase angle difference calculation needs to consider the periodicity of the phase angle. When the phase angle difference exceeds 180 degrees, angle normalization processing is needed. The synchronization error accumulation operation processing performs time integration operation on the phase angle difference values at multiple time points. The integration process uses the trapezoidal integration method to numerically integrate the time series of the phase angle difference to obtain the cumulative value of the phase angle synchronization error. The cumulative operation also considers the weight distribution of the error. The phase angle error of the strongly coupled boundary node is given a higher weight coefficient, and the weight coefficients of the moderately coupled and weakly coupled boundary nodes are correspondingly reduced. The phase angle synchronization error cumulative value is a scalar value. The smaller the value, the higher the phase angle synchronization accuracy between devices, and the larger the value, the more serious the synchronization error. This provides a quantitative basis for the phase angle synchronization objective function in multi-objective optimization.

[0052] The variance analysis processing of the computing load of each simulation device according to the trend of the electrical state change of the boundary node is an important step for evaluating the load balance. The computing load reflects the computing resources consumed by each simulation device when processing the boundary data. The load imbalance will cause some devices to be overloaded while other devices are idle. The variance analysis processing predicts the computing load demand of each device according to the trend of the electrical state change of the boundary node. The load prediction considers factors such as the complexity of data processing, communication overhead and computing time. The variance analysis is a method for evaluating the dispersion degree of data in statistics. The uniformity of load distribution is quantified by calculating the mean and variance of the computing load of each device. The arithmetic mean of the computing load of all devices is calculated, and then the square of the deviation of each device load from the mean is calculated. The load variance is obtained by dividing the sum of all deviation squares by the number of devices. The smaller the variance value, the more uniform the load distribution of each device. The larger the variance value, the more serious the load imbalance. The load imbalance metric value between devices is the standardized result of variance calculation. This metric value provides a quantitative evaluation standard for the load balance objective function in subsequent multi-objective optimization.

[0053] The Pareto frontier analysis processing of the boundary node voltage amplitude deviation matrix, the phase angle synchronization error cumulative value and the load imbalance metric value between devices through weighted coefficient calculation is the core algorithm link of multi-objective optimization. The Pareto frontier analysis is a classical method for finding non-dominated solution set in multi-objective optimization. The weighted coefficient calculation standardizes and combines the values of the three objective functions with different dimensions. The weighted coefficients are determined according to the importance and priority of each objective function. The weight coefficients of the three objectives of voltage deviation, phase angle synchronization and load balance reflect the importance of simulation accuracy, synchronization performance and computing efficiency. The non-dominated sorting algorithm is used in the Pareto frontier analysis processing to identify the Pareto optimal solution. The non-dominated solution refers to the solution that cannot improve the values of other objective functions without deteriorating the values of any objective functions. In the analysis process, all candidate solutions are sorted in layers according to the dominance relationship. The first layer contains all non-dominated solutions, the second layer contains solutions that are dominated by the first layer but are non-dominated among themselves, and so on to form the Pareto frontier hierarchy. The obtained boundary data interaction strategy optimization solution set contains multiple Pareto optimal solutions. Each solution represents a different trade-off scheme between voltage accuracy, phase angle synchronization and load balance.

[0054] In a specific embodiment, the S4 step includes:

[0055] The nanosecond-level timestamp assignment processing is performed on each boundary data in the boundary data interaction strategy optimization solution set to obtain the timestamp marked boundary data packet.

[0056] The importance weight coefficient of each boundary data is calculated based on the electrical distance weight factor, and the priority operation processing is performed combined with the urgency of data transmission to obtain the dynamic priority value of the boundary data.

[0057] The boundary data dynamic priority value is analyzed and processed with the data packet size parameter for transmission scheduling, to obtain a boundary data transmission sequencing strategy;

[0058] The boundary data transmission sequencing strategy is used to queue the boundary data packets marked with time stamps, to obtain a boundary data transmission queue.

[0059] Specifically, in the nanosecond-level time stamp assignment processing of each boundary data in the boundary data interaction strategy optimization solution set, each boundary data is extracted from the boundary data interaction strategy optimization solution set obtained in the previous step. The boundary data contains voltage, phase angle, and current information of different boundary nodes and corresponding transmission strategy parameters. The nanosecond-level time stamp assignment processing uses a high-precision clock source to generate a time mark. The time stamp precision reaches the nanosecond level to meet the strict requirements of electromagnetic transient simulation on time synchronization. The time step of electromagnetic transient simulation is usually in the microsecond level, so the time mark precision of the data packet must reach at least the nanosecond level to ensure timing accuracy. The assignment process assigns a unique time stamp identifier to each boundary data. The time stamp contains absolute time information and relative time sequence information. The absolute time information records the specific time when the data is generated, and the relative time sequence information records the relative position of the data in the entire transmission sequence. The time stamp mark also contains the survival period information of the data packet. Data packets that exceed the survival period will be automatically discarded to avoid the influence of expired data on simulation accuracy. The obtained time stamp marked boundary data packet contains original boundary data and accurate time mark information.

[0060] The importance weight coefficient of each boundary data is calculated based on the electrical distance weight factor, and the priority operation processing is combined with the transmission urgency, which is the core link of dynamic priority management. The importance weight coefficient reflects the influence degree of boundary data on the overall simulation accuracy, and the transmission urgency reflects the timeliness requirement of data. The electrical distance weight factor is extracted and calculated, and the greater the value of the electrical distance weight factor, the stronger the coupling strength of the corresponding boundary node, and the higher the importance of the data. The importance weight coefficient is obtained by normalizing the electrical distance weight factor. The normalization process divides the weight factor of all boundary nodes by the maximum value of the weight factor, so that the importance weight coefficient is distributed between 0 and 1. The calculation of the transmission urgency is based on the timestamp information of the data packet and the time difference of the current time. The smaller the time difference, the fresher the data, and the higher the transmission urgency. The greater the time difference, the older the data, and the lower the transmission urgency. The priority operation processing is a weighted sum of the importance weight coefficient and the transmission urgency. The weighting coefficient is determined according to the specific requirements of the simulation application. For applications with high real-time requirements, the weight of transmission urgency is greater, and for applications with high accuracy requirements, the weight of importance weight coefficient is greater. The obtained dynamic priority value of boundary data is a value between 0 and 1. The greater the value, the higher the transmission priority of the boundary data.

[0061] The transmission scheduling analysis processing of the dynamic priority value of boundary data and the data packet size parameter is a key step in the development of transmission strategy. The data packet size parameter reflects the transmission overhead of boundary data, and the transmission scheduling analysis needs to balance between priority and transmission efficiency. The data packet size parameter is obtained by counting the number of bytes of boundary data, including the total number of bytes of data header, payload data and check information. The larger the data packet, the more network bandwidth it occupies, and the longer the transmission time. The transmission scheduling analysis processing uses a bandwidth efficiency evaluation algorithm. This algorithm divides the dynamic priority value of boundary data by the data packet size parameter to obtain a transmission efficiency indicator, which reflects the priority benefit under unit transmission overhead. The analysis process also considers network congestion and available bandwidth resources. When the network is congested, high-priority small data packets are transmitted first. When the network is idle, large data packets can be transmitted to improve bandwidth utilization. The transmission scheduling analysis also includes data packet merging and fragmentation strategy. Merging and transmitting multiple small data packets can reduce transmission overhead, and fragmenting and transmitting large data packets can avoid long-term occupation of network resources. The obtained boundary data transmission sorting strategy includes the transmission order, transmission opportunity and transmission method of each boundary data packet. This strategy considers multiple factors such as priority, transmission efficiency and network status.

[0062] The queue construction processing of the timestamp marked boundary data packet according to the boundary data transmission ordering strategy is a link of data management, and the queue construction processing organizes discrete boundary data packets into an ordered transmission queue, and the queue structure includes the storage, ordering and scheduling functions of the data packet. The construction process determines the position of each data packet in the queue according to the boundary data transmission ordering strategy, and the ordering strategy includes multiple dimensions such as priority ordering, time ordering and size ordering. The queue construction adopts a multi-level queue structure, high-priority data packets enter a fast queue, medium-priority data packets enter a standard queue, and low-priority data packets enter a slow queue, and different levels of queues adopt different scheduling algorithms and service strategies. The queue management also includes a flow control mechanism, which starts flow control when the queue length exceeds the threshold, suspends the enqueuing of low-priority data packets, and processes high-priority data packets first. The queue construction processing also realizes dynamic adjustment function, dynamically adjusts the queue parameters and scheduling strategy according to the network condition and simulation demand, and guarantees the adaptability and robustness of the queue. The obtained boundary data transmission queue is a dynamic management data structure, which includes all the timestamp marked boundary data packets to be transmitted and their transmission control information.

[0063] In a specific embodiment, the process of performing the queue construction processing of the timestamp marked boundary data packet according to the boundary data transmission ordering strategy can specifically include the following steps:

[0064] Based on the boundary data transmission ordering strategy, a hierarchical cache architecture is established, and a cache level allocation processing of the timestamp marked boundary data packet is performed, to obtain high-speed cache boundary data packets and regular cache boundary data packets;

[0065] The time stamps of the high-speed cache boundary data packets and the regular cache boundary data packets are subjected to clock deviation correction operation processing, to obtain corrected timestamp boundary data packets;

[0066] The corrected timestamp boundary data packets are subjected to ordering algorithm processing according to the time stamp sequence, to obtain a time sequence arranged boundary data packet sequence;

[0067] According to the time sequence arranged boundary data packet sequence, a queue overflow protection mechanism is established, and an intelligent discard strategy configuration processing is performed, to obtain a boundary data transmission queue.

[0068] Specifically, in the process of establishing a hierarchical cache architecture based on the boundary data transmission scheduling strategy and assigning cache levels to the timestamp-labeled boundary data packets, the hierarchical cache architecture is a multi-level storage management structure, containing cache levels with different access speeds and capacity sizes. The high-speed cache has faster access speed but smaller capacity, and the regular cache has larger capacity but relatively slower access speed. The establishment process first analyzes the transmission priority and access frequency of each timestamp-labeled boundary data packet according to the boundary data transmission scheduling strategy obtained in the previous step. Data packets with high transmission priority need to be accessed quickly, and data packets with high access frequency need to be long-term resident in the cache. The cache level assignment process uses a threshold judgment method to compare the dynamic priority value with the pre-set high-speed cache threshold. When the priority value exceeds the high-speed cache threshold, the corresponding timestamp-labeled boundary data packet is assigned to the high-speed cache level, and when the priority value is below the threshold, it is assigned to the regular cache level. The assignment process also considers the size limit of the data packet. Even if the priority is high, a data packet that is too large can only be assigned to the regular cache to avoid wasting high-speed cache resources. High-speed cache boundary data packets mainly include real-time electrical quantity data and emergency control signals of strongly coupled boundary nodes, and regular cache boundary data packets mainly include historical data and state information of moderately coupled and weakly coupled boundary nodes. The assignment result forms a hierarchical cache management system, and different levels of cache use different management strategies and access mechanisms.

[0069] The clock bias correction operation processing of the timestamps of high-speed cache boundary data packets and regular cache boundary data packets is a key step to solve the clock synchronization problem of multiple devices. Clock bias refers to the difference in clock reference between different simulation devices, which can cause the timestamps of boundary data packets to be inaccurate and affect the timing relationship of data. The clock bias correction operation processing first calculates the time bias between each simulation device and the standard clock source, which usually uses a network time protocol server or a GPS clock as a reference. The bias calculation measures the clock offset of each device by periodically sending clock synchronization messages, taking into account factors such as network transmission delay and clock drift. The correction operation uses a linear interpolation method to modify the timestamps. The correction formula adds the clock bias value of the corresponding device to the original timestamp to obtain the corrected timestamp. The correction process modifies the timestamps of high-speed cache boundary data packets and regular cache boundary data packets. Since high-speed cache data packets have more stringent time requirements, their clock correction accuracy is also higher. The timestamps of the corrected boundary data packets are unified to the standard clock reference, eliminating the clock differences between different devices and ensuring the consistency and accuracy of the boundary data packets in the time dimension.

[0070] The sorting algorithm processing of the corrected timestamp boundary data packets in chronological order is an important link to establish correct data timing. The sorting algorithm processing rearranges the discrete boundary data packets according to their chronological order of timestamps to form a time-continuous data sequence. The sorting algorithm adopts a quick sorting method. The basic idea of the quick sorting method is to select a reference element, divide the array into two parts smaller than the reference and larger than the reference, and then recursively sort the two parts. In the present application, the comparison standard of the sorting is the corrected timestamp value of the boundary data packet. The data packet with a smaller timestamp is arranged in front, and the data packet with a larger timestamp is arranged in back. The sorting process also considers the processing of the data packets with the same timestamp. When multiple data packets have the same timestamp, the secondary sorting is performed according to the priority of the data packets, and the data packet with a higher priority is arranged in front. The sorting processing sorts the boundary data packets in the cache and the regular cache respectively. The data packets in the cache have a smaller quantity and a faster sorting speed, and the data packets in the regular cache have a larger quantity but a relatively lower requirement for the sorting speed. The obtained time-sequenced boundary data packet sequence is a data set arranged in strict chronological order, which accurately reflects the evolution process of the boundary data in the time dimension.

[0071] The queue overflow protection mechanism and intelligent discard strategy configuration processing according to the time-sequenced boundary data packet sequence are the safety guarantee link of the queue management. The queue overflow protection mechanism is a protective measure taken when the cache queue capacity is close to saturation to prevent the addition of new data packets from causing queue overflow and data loss. The protection mechanism establishment includes capacity monitoring, threshold setting and trigger condition configuration. The capacity monitoring tracks the usage of the queue in real time, the threshold setting determines the critical point of triggering the protection measure, and the trigger condition configuration defines the specific conditions for starting the protection mechanism. The intelligent discard strategy configuration processing is a data management strategy when the queue overflows. The strategy intelligently selects the discarded data packets according to the importance, timeliness and size of the data packets. The discard strategy adopts a multi-level discard algorithm. First, the expired data packets are discarded, then the data packets with the lowest priority are discarded, and finally the data packets occupying the largest space are discarded. The strategy configuration also includes a data recovery mechanism. For important data packets that are discarded, the identification information is recorded and the data is tried to be reacquired when the conditions allow. The configuration parameters of the intelligent discard strategy are differentiated according to different types of boundary data packets. The data packets of the strongly coupled boundary nodes have a more stringent protection level, and the data packets of the weakly coupled boundary nodes are more easily discarded. The obtained boundary data transmission queue is a data structure with perfect protection mechanism and intelligent management function. The queue can guarantee the safety and integrity of the key data under various abnormal conditions.

[0072] In a specific embodiment, the S5 step comprises:

[0073] The distributed coordination control architecture is established according to the boundary data transmission queue, and the running state of each simulation device is monitored to obtain device computing load rate, communication load rate and boundary data synchronization error parameters;

[0074] Based on the device computing load rate, the communication load rate and the boundary data synchronization error parameters, load balancing analysis and operation processing are performed to obtain the inter-device load redistribution strategy;

[0075] The inter-device load redistribution strategy is matched with the boundary data transmission queue to obtain the data exchange execution instruction of each simulation device;

[0076] According to the data exchange execution instruction, the data synchronization accuracy and system response time of each simulation device are monitored to obtain the data synchronization accuracy and system response performance index.

[0077] Specifically, in the process of establishing the distributed coordination control architecture according to the boundary data transmission queue and monitoring the running state of each simulation device, the distributed coordination control architecture is a multi-level control management structure, which includes a hierarchical organization form of master nodes and slave nodes. The master node is responsible for global decision and coordination management, and the slave node is responsible for specific execution and state feedback. In the establishment process, the data exchange requirements and processing capacity of each simulation device are analyzed according to the boundary data transmission queue obtained in the previous step. The number, size and priority distribution of data packets in the queue reflect the work load intensity of each device. The distributed coordination control architecture adopts a master-slave topology structure, and the simulation device with the strongest computing capacity and the most stable network connection is selected as the master node, and the remaining devices are selected as the slave nodes. The master node and each slave node establish a control communication link. The running state monitoring process is realized by regularly collecting the running parameters of each simulation device. The monitoring parameters include CPU usage, memory occupancy, network bandwidth usage and data processing delay, etc. The device computing load rate is calculated by dividing the current CPU usage by the maximum CPU processing capacity, which reflects the degree of use of the computing resources of the device. The communication load rate is calculated by dividing the current network bandwidth usage by the maximum available bandwidth, which reflects the occupancy of the communication resources of the device. The boundary data synchronization error parameter is calculated by comparing the deviation of the boundary data received by each device from the standard reference value. The larger the deviation value is, the more serious the synchronization error is, and the smaller the deviation value is, the higher the synchronization accuracy is.

[0078] The load balancing analysis and operation processing based on the device computing load rate, communication load rate and boundary data synchronization error parameters is the core algorithm of dynamic load management. The load balancing analysis and operation aims to identify the devices with excessive load and insufficient load, and to develop a reasonable load redistribution scheme. The analysis and operation processing first normalizes the load parameters of each device, converting the load indicators of different dimensions into standardized values between 0 and 1. The standardization process uses the maximum and minimum normalization method, which subtracts the minimum value from the original load value and divides it by the difference between the maximum and minimum values. The load balancing degree evaluation is achieved by calculating the variance of each device load. The smaller the variance value, the more uniform the load distribution. The larger the variance value, the more serious the load imbalance. The load imbalance value calculation process first calculates the arithmetic mean of all device loads, then calculates the square of the difference between each device load and the mean value, and finally divides the sum of all square differences by the number of devices to obtain the load variance. The load redistribution strategy development uses a load transfer algorithm that identifies high-load devices with load rates exceeding the upper threshold and low-load devices with load rates below the lower threshold, and then calculates the amount of tasks that need to be transferred from high-load devices to low-load devices. The task transfer amount calculation considers the divisibility of tasks, transfer cost and communication delay between devices, and preferentially transfers divisible and low-cost computing tasks. The final inter-device load redistribution strategy contains specific task transfer scheme, transfer timing and transfer path, which ensures the load distribution of each device to be balanced.

[0079] The coordination and matching processing of the inter-device load redistribution strategy and the boundary data transmission queue is the key link of unified scheduling management. The coordination and matching processing needs to consider the demand of load redistribution and the requirement of boundary data transmission at the same time, to avoid conflicts and interference between them. The coordination and matching processing first analyzes the task transfer plan in the load redistribution strategy, identifies which tasks need to be transferred from which device to which device at what time, and then analyzes the data transmission plan in the boundary data transmission queue to determine the transmission time and transmission path of each data packet. The matching algorithm uses a time window scheduling method, which divides time into multiple consecutive time windows, and uniformly arranges task transfer and data transmission activities in each time window. The conflict detection mechanism identifies the situation of task transfer and data transmission at the same time in the same time window, and coordinates when a conflict is detected, with high-priority boundary data transmission being executed first and low-priority task transfer being executed later. The coordination and matching also considers the allocation of network bandwidth resources, reserving corresponding bandwidth resources for task transfer and data transmission to avoid performance degradation caused by competition for the same network resource. The final data exchange execution instruction is a comprehensive scheduling scheme containing detailed time arrangement and resource allocation, which uniformly coordinates the execution process of load redistribution and data transmission.

[0080] According to the data exchange execution instruction, the data synchronization accuracy and system response time of each simulation device are monitored and processed, which is an important part of performance evaluation and feedback regulation. The monitoring process verifies the effectiveness of coordinated control by tracking the data synchronization status and response performance of each device in real time. The data synchronization accuracy monitoring is realized by comparing the consistency of the boundary data received by each device and the boundary data sent, which includes numerical accuracy checking and timestamp accuracy checking. The numerical accuracy checking calculates the numerical deviation of the received data and the sent data, and the timestamp accuracy checking calculates the time delay of data transmission. The system response time monitoring is realized by measuring the time interval from the data exchange instruction to the execution completion, which includes instruction transmission time, data processing time and result feedback time. The monitoring process uses a sliding window statistical method to statistically analyze the statistical characteristics of data synchronization accuracy and response time in a fixed length time window, including average value, standard deviation, maximum value and minimum value, etc. The performance index calculation process statistically analyzes the original data obtained by monitoring. The data synchronization accuracy index is obtained by calculating the root mean square value of the synchronization error, and the response time performance index is obtained by calculating the percentile of the response time. The monitoring results also include performance trend analysis, which judges the effectiveness and stability of the coordinated control strategy by comparing the performance index change trend in different time periods.

[0081] In a specific embodiment, the execution step can specifically include the following steps based on the device computing load rate, communication load rate and boundary data synchronization error parameters for load balancing analysis and operation processing:

[0082] The device computing load rate, communication load rate and boundary data synchronization error parameters are standardized and normalized to obtain a device load comprehensive evaluation index;

[0083] Based on the device load comprehensive evaluation index, the load difference between each simulation device is calculated and threshold comparison analysis is performed to obtain a load imbalance device identification result;

[0084] According to the load imbalance device identification result, the computing task of the high load device is decomposed and split to obtain a redistributable computing task unit;

[0085] The redistributable computing task unit is matched and allocated with the remaining computing capacity of the low load device to obtain a device load redistribution strategy.

[0086] Specifically, in the standardization and normalization process of the device load rate, the communication load rate and the boundary data synchronization error parameter, the standardization and normalization process is a data processing method for converting parameters of different dimensions and value ranges into a unified standard, eliminating the weight deviation caused by the value difference between parameters. The standardization process first calculates the maximum and minimum values of each type of parameter by statistically analyzing the value range of each type of parameter. The value range of the device load rate is usually between 0 and 1, the value range of the communication load rate is also between 0 and 1, and the value range of the boundary data synchronization error parameter is determined according to the specific application scenario. The normalization process uses the maximum and minimum value normalization method to subtract the minimum value of each original parameter value of the same type of parameter, and then divide by the difference between the maximum value and the minimum value to obtain a standardized value between 0 and 1. The processing process also includes weight distribution calculation. According to the performance requirements of the joint operation of multiple simulation devices, the importance weights of the three types of parameters are determined. The device load usually occupies the highest weight because it directly affects the simulation speed, the communication load occupies the medium weight because it affects the data exchange efficiency, and the synchronization error occupies a relatively low but non-negligible weight because it affects the simulation accuracy. The device load comprehensive evaluation index is calculated by weighted average of the three standardized parameters according to the corresponding weights. The index comprehensively reflects the overall load level and running state of each simulation device.

[0087] The core algorithm of load imbalance identification is to calculate the load difference between each simulation device based on the device load comprehensive evaluation index and perform threshold comparison analysis processing. The load difference quantifies the unevenness of the load distribution between different devices, and the threshold comparison determines the range of devices that need to be adjusted. The load difference is calculated using the variance analysis method. First, the arithmetic mean of the device load comprehensive evaluation index of all devices is calculated. Then, the square of the difference between the evaluation index of each device and the average value is calculated. The sum of all squared differences is divided by the number of devices to obtain the load variance. The square root of the variance is the load difference. The larger the difference value, the more uneven the load distribution between devices. The smaller the difference value, the more balanced the load distribution. The threshold comparison analysis processing sets two judgment criteria, the high load threshold and the low load threshold. The high load threshold is usually set to 1.2 to 1.5 times the average load, and the low load threshold is set to 0.5 to 0.8 times the average load. The comparison analysis compares the comprehensive evaluation index of each device with the two thresholds. When the evaluation index of a device exceeds the high load threshold, it is identified as a high load device. When the evaluation index is lower than the low load threshold, it is identified as a low load device. Devices between the two thresholds are considered to have normal load. The load imbalance device identification result includes a high load device list, a low load device list and the load exceeding degree of each device, providing accurate target positioning for subsequent task redistribution.

[0088] The decomposition and splitting of the computing tasks of the high-load device according to the identification result of the load imbalance device is a key step of task reorganization. The decomposition and splitting of the computing tasks decompose the complex simulation computing tasks into multiple relatively independent sub-task units, so that part of the sub-tasks can be transferred to other devices for execution. The decomposition and splitting process first analyzes the type and structure of the computing tasks currently executed by the high-load device. The computing tasks usually include different types such as electromagnetic transient calculation, power flow calculation, stability analysis and protection action simulation. The task decomposition adopts a functional modularization method, which decomposes large computing tasks into multiple independent computing modules according to functional boundaries. Each module has a clear input-output interface and independent computing logic. The splitting process also considers the dependency relationship between tasks, identifies which sub-tasks have data dependency or time sequence dependency, and ensures that the split task units maintain relative independence and necessary correlation. The splitting algorithm preferentially selects computing-intensive tasks with less dependency for separation, which has the least impact on the original device and the lowest integration difficulty on the new device. The redistributable computing task unit is a computing module with standardized interface and clear functional definition formed after decomposition and splitting. These modules contain complete information such as task computing logic, data interface, resource demand and execution time.

[0089] The matching and allocation operation of the redistributable computing task unit and the remaining computing capacity of the low-load device is the implementation link of load redistribution. The matching and allocation operation needs to consider the matching degree of task demand and device capacity to ensure the rationality and feasibility of task allocation. The matching and allocation operation first evaluates the remaining computing capacity of each low-load device. The remaining computing capacity is obtained by subtracting the current computing load from the maximum computing capacity of the device, including resource indicators such as remaining CPU processing capacity, available memory space and idle network bandwidth. The matching degree of the task and the device considers multiple matching factors, including computing resource matching degree, technical compatibility matching degree and communication efficiency matching degree. The computing resource matching degree is calculated by comparing the resource demand of the task and the remaining resources of the device. When the resource demand is less than the remaining resources, the matching degree is higher. The technical compatibility matching degree evaluates the degree of fit between the type of the task and the expertise of the device. The same type of computing task is more efficient on similar devices. The communication efficiency matching degree considers the data exchange cost after task transfer and the original device. Devices with short physical distance and good network connection quality have higher communication efficiency matching degree. The allocation operation uses the Hungarian algorithm to solve the optimal matching scheme of the task and the device. This algorithm finds the best matching combination under multiple constraints to maximize the overall matching degree. The final device load redistribution strategy specifies the target device, transfer time, interface protocol and monitoring scheme of each redistributable computing task unit in detail.

[0090] The above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalent features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for joint operation and information data interaction processing of a multi-simulation device, characterized in that, The method comprises: S1 step, collecting the boundary node voltage amplitude, phase angle and current data of multiple simulation devices, calculating the electrical distance weight factor to determine the coupling strength, and dividing the boundary nodes into strong coupling boundary nodes, medium coupling boundary nodes and weak coupling boundary nodes; S2 step, respectively predicting the boundary node electrical state change trend in the future time step based on the historical electrical quantity data of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes; S3 step, solving the boundary data interaction strategy optimization solution set through the predictive boundary state estimation of the boundary node electrical state change trend, comprising: performing deviation calculation processing on the boundary node electrical state change trend and the real-time voltage amplitude of each simulation device to obtain a boundary node voltage amplitude deviation matrix; performing synchronization error accumulation operation processing on the phase angle difference between devices based on the boundary node electrical state change trend to obtain a phase angle synchronization error accumulation value; performing variance analysis processing on the calculation load of each simulation device according to the boundary node electrical state change trend to obtain a load imbalance measurement value between devices; performing Pareto frontier analysis processing on the boundary node voltage amplitude deviation matrix, the phase angle synchronization error accumulation value and the load imbalance measurement value between devices through weighted coefficient calculation to obtain the boundary data interaction strategy optimization solution set; S4 step, assigning a time stamp to the boundary data in the optimization solution set, and dynamically adjusting the transmission priority through the electrical distance weight factor to generate a boundary data transmission queue; S5 step, each simulation device performs data exchange operation according to the boundary data transmission queue, and monitors the data synchronization accuracy and system response performance indicators.

2. The joint operation and information data interaction processing method for multi-simulation devices according to claim 1, characterized in that, The S1 step comprises: Performing electrical quantity data acquisition processing on the boundary nodes of each simulation device to obtain the original data set of the boundary node voltage amplitude, phase angle and current; Calculating the equivalent impedance and physical distance parameters between each boundary node based on the original data set to obtain the electrical distance weight factor; Comparing and analyzing the electrical distance weight factor with the preset coupling strength threshold to obtain the coupling strength level of each boundary node; Classifying and marking the boundary nodes according to the coupling strength level to obtain the classification results of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes.

3. The method for joint operation and information data interaction processing of multi-simulation devices according to claim 1, characterized in that, The S2 step comprises: Performing time series construction processing on the historical electrical quantity data of the strong coupling boundary nodes, the medium coupling boundary nodes and the weak coupling boundary nodes to obtain the voltage amplitude time series, the phase angle time series and the current time series of each type of boundary node; Establishing an improved characteristic line equation parameter matrix based on the voltage amplitude time series, the phase angle time series and the current time series to obtain the boundary node electrical state transition law parameters; Performing electromagnetic transient wave propagation analysis processing on the boundary node electrical state transition law parameters through boundary condition correction term calculation to obtain the voltage amplitude prediction value, the phase angle prediction value and the current prediction value of each boundary node in the future time step; A boundary node state change vector is constructed according to the voltage amplitude prediction value, the phase angle prediction value and the current prediction value, and a boundary node electrical state change trend is obtained.

4. The method for joint operation and information data interaction processing of multi-simulation devices according to claim 1, characterized in that, The S4 step comprises: nanosecond-level timestamp assignment processing is performed on each boundary data in the boundary data interaction strategy optimization solution set, to obtain timestamp-labeled boundary data packets; importance weight coefficients of each boundary data are calculated based on the electrical distance weight factor, and priority operation processing is performed in combination with data transmission urgency, to obtain boundary data dynamic priority values; transmission scheduling analysis processing is performed on the boundary data dynamic priority values and data packet size parameters, to obtain a boundary data transmission sequencing strategy; queue construction processing is performed on the timestamp-labeled boundary data packets according to the boundary data transmission sequencing strategy, to obtain a boundary data transmission queue.

5. The joint operation and information data interaction processing method for multi-simulation devices according to claim 4, characterized in that, The queue construction processing on the timestamp-labeled boundary data packets according to the boundary data transmission sequencing strategy to obtain the boundary data transmission queue comprises: a hierarchical cache architecture is established based on the boundary data transmission sequencing strategy, and cache level assignment processing is performed on the timestamp-labeled boundary data packets, to obtain cache boundary data packets and regular cache boundary data packets; clock deviation correction operation processing is performed on the timestamps of the cache boundary data packets and the regular cache boundary data packets, to obtain corrected timestamp boundary data packets; sorting algorithm processing is performed on the corrected timestamp boundary data packets in timestamp order, to obtain a time-series arrangement boundary data packet sequence; a queue overflow protection mechanism is established according to the time-series arrangement boundary data packet sequence, and intelligent discard strategy configuration processing is performed, to obtain the boundary data transmission queue.

6. The joint operation and information data interaction processing method for multi-simulation devices according to claim 1, characterized in that, The S5 step comprises: a distributed coordinated control architecture is established according to the boundary data transmission queue, and running state monitoring processing is performed on each simulation device, to obtain device computation load rate, communication load rate and boundary data synchronization error parameters; load balancing analysis operation processing is performed based on the device computation load rate, the communication load rate and the boundary data synchronization error parameters, to obtain an inter-device load redistribution strategy; the inter-device load redistribution strategy is matched with the boundary data transmission queue, to obtain data exchange execution instructions of each simulation device; data synchronization accuracy and system response time monitoring processing is performed on each simulation device according to the data exchange execution instructions, to obtain data synchronization accuracy and system response performance indicators.

7. The joint operation and information data interaction processing method for multi-simulation devices according to claim 6, characterized in that, The load balancing analysis operation processing based on the device computation load rate, the communication load rate and the boundary data synchronization error parameters to obtain the inter-device load redistribution strategy comprises: standardized normalization processing is performed on the device computation load rate, the communication load rate and the boundary data synchronization error parameters, to obtain a device load comprehensive evaluation index; load difference degrees between each simulation device are calculated based on the device load comprehensive evaluation index, and threshold comparison analysis processing is performed, to obtain a load imbalance device identification result; computation tasks of a high-load device are decomposed and split according to the load imbalance device identification result, to obtain a redistributable computation task unit; The redistributable computing task units are allocated to the remaining computing capacity of the low-load devices to obtain an inter-device load redistribution strategy.

Images