Electric power system electromagnetic transient simulation system containing multiple electric power electronic switches

Abstract

The invention relates to the technical field of electromagnetic transient simulation, in particular to an electromagnetic transient simulation system of a power system containing multiple electronic power switches, which is characterized in that numbering and topological serial number analysis are performed on multi-switch node, resistance, inductance and capacitance parameters, and a connection mapping matrix is formed by adopting impedance conversion and admittance construction modes; a sequence structure is formed by coding control variables through a genetic algorithm, a variable sequence is constructed through switching time points and topological connection, error and duration combined evaluation is executed, an optimal combination is screened, time matrix construction is carried out after energy storage voltage and filter current are collected, and a principal component vector and a covariance projection coefficient are extracted; meanwhile, an input matrix is constructed, harmonic information is fused in a unified time step, a residual network structure is introduced to carry out time-step-by-step iterative updating processing on an initial weight, dynamic adjustment is completed in combination with residual offset calculation and local error compression in each step, and a continuous and complete reconstructed waveform is output.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic transient simulation, and in particular to an electromagnetic transient simulation system for a power system containing multiple power electronic switches. Background Art

[0002] The field of electromagnetic transient simulation technology aims to model and simulate the rapid electromagnetic processes in power systems caused by power electronic switches, faults, control behaviors, etc. through numerical calculation methods, so as to accurately characterize the dynamic changes of voltage, current, and electromagnetic energy on the time scale of microseconds to milliseconds, evaluate the stability and reliability of the system under transient disturbances, analyze the propagation path of electromagnetic interference, verify the effectiveness of protection strategies, and assist in the design and optimization of power electronic devices and control systems.

[0003] The purpose of the electromagnetic transient simulation system for power systems containing multiple power electronic switches is to accurately model and simulate the operation process of these switching devices in the system, capture the transient changes in current, voltage, and electromagnetic energy caused by changes in device driving signals, device conduction and shutdown, network topology transformation, etc., evaluate the system's voltage sag, current surge, electromagnetic interference, and the adaptability of the system protection strategy, and thus provide reliable support for system design, operation evaluation, and fault analysis.

[0004] In the process of constructing electromagnetic transient simulation, existing technologies mainly simulate the response of power electronic switches through fixed modeling and static parameter settings. The processing logic does not include an effective evolution path and feedback iteration mechanism, resulting in the system relying heavily on manual settings in the control parameter matching process and lacking an optimization process, making it difficult to complete multi-index trade-offs. Most current simulations use high sampling frequency and long-cycle calculation modes, which dramatically increase the computational load but still cannot guarantee the validity of data extraction. The dimensionality reduction stage only performs simplified processing based on the number of truncated principal components, ignoring the contribution of each vector to the overall information volume, resulting in insufficient information retention. In the reconstruction process, simple interpolation and splicing methods are often used to merge the feature segments, resulting in waveform breakage and boundary discontinuity problems, which form a systematic constraint on simulation accuracy, scheduling efficiency and parameter traceability. Summary of the Invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an electromagnetic transient simulation system for a power system containing multiple power electronic switches.

[0006] In order to achieve the above-mentioned object, the present invention adopts the following technical solution: a power system electromagnetic transient simulation system containing multiple power electronic switches comprises: Topology analysis module: Based on the parameters of each switch node, inductance, capacitance, and resistance in multiple power electronic switch devices, it extracts the switch number and topology sequence to perform impedance calculation and admittance conversion, normalizes the matrix elements, and maps the node connection structure to generate the network topology matrix; Parameter evolution module: Based on the network topology matrix, a genetic algorithm is used in combination with the timing of switching devices to establish a variable sequence by extracting switching time points and connection information. The simulation runs and records errors and durations, sorts and determines retention and mutation, and generates an optimal combination of control parameters. Reduced-order mapping module: Based on the preferred combination of control parameters, collects the energy storage unit voltage and filter current sequence, constructs a time matrix and performs singular value calculation and eigenvector selection, calculates the projection coefficient and determines the residual adjustment dimension to obtain a reduced-order low-dimensional expression; State reconstruction module: Based on the reduced low-dimensional expression, a residual network is used to construct an input matrix by combining the switch state and harmonics, the time step and coefficient are matched to construct a combined sequence, the offset difference is calculated and the weight is adjusted and updated iteratively to obtain a reconstructed waveform sequence table; Process scheduling module: Based on the reconstructed waveform sequence table, combined with the capacitor charging and discharging and resonance stages, the task unit is defined, the task duration and waiting ratio are counted, the parallel conditions and triggering sequence are determined, and a scheduling trigger structure is established.

[0007] As a further solution of the present invention, the topology parsing module includes: Node parameter mapping submodule: Based on the parameters of each switch node, inductor, capacitor, and resistor in the multi-power electronic switch device, node information matching is used to extract the switch number and calibrate the topological sequence number. The complex impedance value array is constructed by electrical parameter conversion. The real and imaginary parts of the admittance are output by numerical conversion. The admittance elements are scaled to form a vector set, generating a normalized admittance data set. Connection structure construction submodule: Based on the normalized admittance data set, index mapping is used to process the correspondence between node numbers and vectors, matrix assembly is used to construct the node pair connection matrix, element fusion is used to present the two-dimensional mapping diagram of admittance and structure, and the multi-node link information set is summarized and output to obtain the network topology matrix.

[0008] As a further solution of the present invention, the parameter evolution module includes: Switching sequence generation submodule: Based on the network topology matrix, the switching time points of the switching devices and the corresponding connection node numbers are extracted. A genetic algorithm is used to construct a node index mapping table in combination with the topological row and column labels. A timing axis alignment operation is performed to correct the time points of each channel. A connection information filtering operation is performed to remove invalid node states. The effective switching segments in the entire time domain are sorted. The variable state list is spliced ​​in time sequence to construct a complete sequence structure and fill in missing fields to obtain a variable sequence set. Error screening submodule: Based on the variable sequence set, it performs circuit model simulation calculation to output waveform groups, calculates the error curve between the target waveform and the actual response through residual extraction operation, generates sequence quality rankings by sorting the error integrated area, performs runtime statistics and establishes a secondary indicator matrix, combines error rankings and runtime rankings through weight allocation, extracts top-ranked sequences and adds retention marks, and simultaneously generates a variation mark matrix to obtain the optimal parameter index; Optimal combination generation submodule: Based on the preferred parameter index, the retained state and the corresponding variable number index are extracted, a mapping matching operation is performed to associate the original control parameter data, a structurally complete control vector set is formed through a combined reconstruction operation, the mutation labeling matrix is ​​superimposed to generate multiple groups of structural perturbation candidates, different control vector sets and corresponding mutation marker labels are summarized, a two-dimensional control parameter set array is constructed, and the optimal combination of control parameters is obtained.

[0009] As a further solution of the present invention, the genetic algorithm first uses the variable sequence set obtained after time axis alignment and invalid node screening as the initial population, each sequence corresponds to a chromosome, and then uses the weighted sum of the simulation error integral area and the running time as the fitness function to evaluate the fitness of each chromosome in the population. Then, the selection operator selects the parent from the high-fitness chromosomes, and reorganizes the time point and node number mapping table of the parent sequence through a single-point or multi-point crossover operator to obtain the offspring chromosomes. Subsequently, the mutation operator is used to slightly perturb the switching time or connection information of some offspring chromosomes. At the same time, the parents and offspring are sorted according to fitness, and the top N optimal chromosomes are retained as the new generation population. The iteration is repeated until the convergence conditions are met and the preset algebraic upper limit is reached, and the optimal control parameter combination scheme is output.

[0010] As a further solution of the present invention, the genetic algorithm is according to the formula: ; in: Indicates the candidate node index sequence in The updated permutation result in the genetic iteration, Indicates the candidate node index sequence in the first The current permutation state in iterations, represents the global adjustment coefficient, Represents the weight of the fitness difference component in the index update, represents the weight of the channel switching time consistency component, represents the weight of the node topological connectivity score, represents the weight of the frequency of historical elite selection, Indicates in The individual fitness value with the minimum number of switch state conflicts and the best connectivity performance in the iteration, Indicates the Individuals in the The current fitness value in iterations, Indicates the The mean square error of the individual channel switching time point series, Indicates the The score of the proportion of connected edges in the node connection matrix constructed by the individual, Indicates the The frequency ratio of an individual being selected as an elite individual in successive genetic iterations.

[0011] As a further solution of the present invention, the reduced-order mapping module includes: Characteristic matrix generation submodule: Based on the preferred combination of the control parameters, the voltage and current acquisition method is used to read the energy storage unit voltage and filter current values, a time index is used to construct a two-dimensional sequence matrix, a column vector energy decomposition method is used to divide the energy distribution, and a peak contribution is used to screen and identify the main vector set to generate a principal element characteristic matrix; Low-dimensional mapping derivation submodule: Based on the principal component feature matrix, matrix mapping is used to calculate the output coefficient list, residual variance signal accumulation is used to determine the projection dimension, combined screening is used to refine the minimum dimension vector array, and time rearrangement is used to reconstruct the low-dimensional feature sequence to obtain a reduced-order low-dimensional expression.

[0012] As a further solution of the present invention, the state reconstruction module includes: Input matrix construction submodule: Based on the reduced low-dimensional expression, the switching device state data and harmonic interference channel signals corresponding to each stage are collected, and the data of different frequency bands are uniformly resampled using the equal time step interpolation method. An index table is constructed to synchronously annotate time labels and variable names. A three-dimensional matrix structure is formed through node number mapping and harmonic frequency band splitting. A tensor expansion operation is performed to compress it into a two-dimensional input structure to obtain the input matrix; Weight iteration submodule: Based on the input matrix, a residual network is used, an initial weight vector is introduced as a reference value of the superposition coefficient, a residual calculation operation is performed to evaluate the degree of waveform offset under the current combination, the alignment index is updated using the maximum offset position, the difference in weights of adjacent time steps is corrected through the local error compression method, an iterative update is performed and the result of each step is retained in the record sequence, and the iteration is repeated until the offset change is lower than the threshold, thereby obtaining a weight iteration sequence; Waveform output submodule: Based on the weighted iterative sequence, feature segments are extracted from the input matrix according to the time dimension and weighted superposition is performed to generate initial waveform segments, continuous frame splicing operations are performed to restore the complete waveform sequence, the consistency of critical area overlapping segments is verified through the sliding window mode, boundary transition correction and length normalization processing are completed, a stable and traceable waveform output table is generated, and a reconstructed waveform sequence table is obtained.

[0013] As a further solution of the present invention, the residual network first inputs a sample set and introduces an initialized weight vector as the initial parameter of each connection layer in the residual network. Then, it is expanded according to the time series structure. Each layer of neural units performs a combination of linear transformation and activation function operations to generate a predicted output sequence, and performs difference calculation with the target waveform to obtain a residual sequence. Then, the mean square error is used as the loss function to quantify the residual size, the error information is backpropagated, and the network weights are updated. In each round of iteration, the offset degree of the current output waveform and the target reference waveform is compared, the offset point is extracted and the time alignment index is adjusted, and the local weight difference between the previous and next time steps is compressed and adjusted to reduce oscillation. The iterative process continues until the residual change is lower than the set threshold, and the converged weight iterative sequence is output as a stable parameter set.

[0014] As a further solution of the present invention, the residual network is according to the formula: ; in: Indicates in The weight vector generated in the iteration is Indicates in The current weight vector in iteration , represents the learning rate of the residual network, represents the sampled power system input matrix, Represents the input matrix The transpose operation, represents the desired target waveform vector, represents the weight coefficient of the residual smoothing term, Indicates the The residual vector of the iteration, represents the weight coefficient of the regularization term, represents the regularization gradient applied to the current weight vector, represents the penalty coefficient of the residual variance term, Indicates the The variance of the residual vector in the iteration, represents the weight coefficient of the historical momentum term, Indicates the The momentum term accumulated over iterations.

[0015] As a further solution of the present invention, the process scheduling module includes: Task unit division submodule: Based on the reconstructed waveform sequence table, the continuous amplitude rising segment detection method is adopted, the continuous amplitude falling segment identification method is adopted in the superimposed discharge segment extraction operation, and the zero crossing interval extraction method is used in combination with the resonance segment extraction operation to complete the division. The superimposed equal interval sampling operation is used to extract the peak and valley point series, and the start and end time of each segment is matched to form a duration statistics table to obtain the task timing distribution; Trigger structure construction submodule: Based on the task timing distribution, extract the parallel segment numbers, superimpose the trigger sequence generation operation to construct a time series number group, perform the dependent path sorting operation to establish a sequence chain according to the adjacent task numbers, construct a trigger priority grouping table to generate a priority index according to the waiting ratio, and obtain the scheduling trigger structure.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, multiple switch nodes, resistors, inductors, and capacitors are numbered and topologically sequenced, impedance conversion and admittance construction are used to form a connection mapping matrix, and a genetic algorithm is applied to encode control variables to form a sequence structure. The variable sequence is constructed by switching time points and topological connections, and a joint evaluation of error and duration is performed to screen the optimal combination, so that the control parameter evolution process has the ability to automatically screen and optimize.

[0017] In the present invention, after collecting the energy storage voltage and filter current, a time matrix is ​​constructed and the principal component vector and covariance projection coefficient are extracted. The projection error residual is dynamically dimensioned to obtain a compressed and effective low-dimensional sequence. At the same time, an input matrix is ​​constructed and harmonic information is fused within a unified time step. A residual network structure is introduced to iteratively update the initial weights time by time. In each step, dynamic adjustment is completed by combining residual offset calculation and local error compression to obtain a weight sequence that is more in line with the actual response characteristics, and a continuous and complete reconstructed waveform is output. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a system flow chart of the present invention. DETAILED DESCRIPTION

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

[0020] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0021] Example: See Figure 1 The present invention provides a technical solution: an electromagnetic transient simulation system for a power system containing multiple power electronic switches includes: Topology analysis module: Based on the parameters of each switch node, inductance, capacitance, and resistance in multiple power electronic switch devices, it extracts the switch number and topology sequence to perform impedance calculation and admittance conversion, normalizes the matrix elements, and maps the node connection structure to generate the network topology matrix; Parameter evolution module: Based on the network topology matrix, using genetic algorithms and combining the timing of switching devices, it establishes a variable sequence by extracting switching time points and connection information. The simulation runs and records errors and durations, sorts and determines retention and mutation, and generates the optimal combination of control parameters. Reduced-order mapping module: Based on the optimal combination of control parameters, it collects the energy storage unit voltage and filter current series, constructs a time matrix, calculates singular values ​​and selects eigenvectors, calculates projection coefficients and determines the residual adjustment dimension to obtain a reduced-order low-dimensional expression; State reconstruction module: Based on the reduced-order low-dimensional expression, the residual network is used to combine the switch state and harmonics to construct the input matrix, match the time step and coefficient to construct the combined sequence, calculate the offset difference, adjust the weights and iteratively update them to obtain the reconstructed waveform sequence table; Process scheduling module: Based on the reconstructed waveform sequence table, combined with the capacitor charging and discharging and resonance stages, the task unit is defined, the task duration and waiting ratio are counted, the parallel conditions and trigger sequence are determined, and the scheduling trigger structure is established.

[0022] The topology parsing module includes: Node parameter mapping submodule: Based on the parameters of each switch node, inductor, capacitor, and resistor in the multi-power electronic switch device, node information matching is used to extract the switch number and calibrate the topological sequence number. The complex impedance value array is constructed by electrical parameter conversion. The real and imaginary parts of the admittance are output by numerical conversion. The admittance elements are scaled to form a vector set, generating a normalized admittance data set. Connection structure construction submodule: Based on the normalized admittance data set, index mapping is used to process the correspondence between node numbers and vectors, matrix assembly is used to construct the node pair connection matrix, element fusion is used to present the two-dimensional mapping diagram of admittance and structure, and the multi-node link information set is summarized and output to obtain the network topology matrix; Node parameter mapping submodule: Based on multiple switch node parameters, inductance parameters, capacitance parameters, and resistance parameters in multiple power electronic switch devices, a node information matching method is used to perform a one-to-one matching operation on the node ID and the node parameter mapping table. During the hash retrieval process, a key-value pair structure is set to store all node parameters. The node ID is used as the key value to retrieve and extract the corresponding switch number. During the execution process, a traversal operation is performed on each node and the switch number append write command is called to form a number set. When performing the topology number calibration operation, a number mapping table is constructed to use the switch number as the retrieval input. The relationship between the number and the topology number is set against the static configuration table. The topology number is generated in index order and written into the topology number vector. The electrical parameter conversion method is used to construct Create a complex impedance value array, set a unified operating frequency for the resistance, inductance, and capacitance values ​​as the benchmark for impedance calculation, with the resistance directly input as the real part, and the inductive and capacitive reactances converted by frequency and component values ​​as the imaginary parts. The complex impedance is encapsulated in a dual-channel data structure as real and imaginary part pairs and written into the array. The real and imaginary parts of the admittance are output using a numerical conversion method. During execution, the reciprocal conversion operation is performed using the complex impedance value as the input parameter. The result output uses floating-point decomposition to extract the real and imaginary parts and record them as the real and imaginary admittance arrays, respectively. The admittance elements are scaled to form a vector set. All real and imaginary elements are multiplied by the scaling factor to generate the scaled vector set and encapsulate it as a list of two-element tuples to generate a normalized admittance data set. Connection structure construction submodule: Based on the normalized admittance data set, the index mapping method is used to process the correspondence between node numbers and vectors. When performing the number and vector binding operation, the positions are mapped one by one with the node number list in the order of the vectors. Each group of node numbers is used as the key, and the corresponding admittance vector is written as the value into the mapping dictionary. The node pair connection matrix is ​​constructed using the matrix assembly method. All node pair combinations are traversed. For each group of node pairs, the number information is extracted and the corresponding admittance vector value is obtained from the mapping dictionary. The assignment operation is performed to write the value into the corresponding position of the two-dimensional matrix. After the matrix construction is completed, the element fusion method is used to present the two-dimensional mapping diagram of admittance and structure. The real and imaginary parts of the admittance vector are used as the two-dimensional image coordinate input values ​​respectively. The corresponding coordinate lattice is constructed with the node pair number. The image rendering window size and resolution are set. All valid admittance points are plotted in graphical form and a two-dimensional distribution map is generated. The multi-node link information set is summarized and output. The information of all node pairs involved in the construction is numbered and encapsulated, and written into the triple form structure set in the order of the numbers to obtain the network topology matrix.

[0023] The parameter evolution module includes: Switching sequence generation submodule: Based on the network topology matrix, it extracts the switching time points of switching devices and the corresponding connection node numbers. It uses a genetic algorithm and combines topological row and column labels to build a node index mapping table. It then performs a timing axis alignment operation to correct the time points of each channel. It also performs a connection information filtering operation to remove invalid node states, organizes the effective switching segments in the entire time domain, and splices the variable state list in time order to construct a complete sequence structure and fill in missing fields to obtain a variable sequence set. Error screening submodule: Based on the variable sequence set, it performs circuit model simulation to calculate the output waveform group, calculates the error curve between the target waveform and the actual response through residual extraction, generates sequence quality rankings based on the error integral area, performs runtime statistics and establishes a secondary indicator matrix, combines the error ranking and runtime ranking through weight allocation, extracts the top-ranked sequences and adds retention marks, and simultaneously generates a mutation mark matrix to obtain the optimal parameter index; The optimal combination generation submodule extracts the retained state and the corresponding variable number index based on the optimal parameter index, performs a mapping and matching operation to associate the original control parameter data, forms a complete control vector set through a combined reconstruction operation, superimposes the mutation label matrix to generate multiple sets of structural perturbation candidates, summarizes different control vector sets and corresponding mutation marker labels, constructs a two-dimensional control parameter set array, and obtains the optimal combination of control parameters; Switching sequence generation submodule: Based on the network topology matrix, the switching time points of the switching devices and the corresponding connection node numbers are extracted. A genetic algorithm is used, the population size is set to 100 chromosomes, the length of each chromosome is set to 20 bits, the encoding method is binary encoding, and each chromosome is randomly generated as 0 or 1 when the population is initialized. The fitness function uses the sum of squared errors as the calculation basis. The control variables of each chromosome are traversed item by item and the error is calculated in the simulation platform to generate the fitness value. The roulette method is used for selection operation. The fitness array is normalized and weighted sampling is performed to generate the new generation chromosome index. When performing a single-point crossover operation, the crossover probability is set to 0.8. The crossover point index is randomly generated in each pair of parents and the gene segment after the crossover point is exchanged. The mutation operation sets the mutation rate to 0.02 , traverse all sites in each chromosome bit by bit and flip the bit with a probability of 0.02 to form a new set of mutant chromosomes, build a node index mapping table based on the topological row and column labels, establish a mapping dictionary by one-to-one correspondence between the node number and the row and column position number in the topological matrix, and perform a time axis alignment operation to correct the time points of each channel, uniformly align all time channels to the minimum unit and fill in the missing time point state to form a consistent time step list, perform a connection information screening operation to remove invalid node states, set an invalid node list and traverse the state sequence by number to remove the corresponding records, organize the full time domain effective switching segments, sort the time points, and splice the corresponding state sets to generate a unified state sequence, fill in the missing fields through the field definition template to generate a complete state sequence structure, and generate a variable sequence set; The error screening submodule performs circuit model simulation based on the variable sequence set to calculate the output waveform group. The simulation model input parameters include the control vector, power supply conditions, a sampling time step of 1e-6 seconds, and a total simulation time of 0.01 seconds. An independent simulation process is performed for each variable sequence, and the waveform output array on the corresponding time axis is collected. The difference between the target waveform and the actual simulation response is calculated through residual extraction. At each time point, the current target value is subtracted from the simulation value to obtain an error signal list. The error absolute value is numerically integrated over all time periods to form an error integral area array. The error integral values ​​are sorted from small to large to generate an error level list. A runtime statistics operation is performed to calculate the total duration of the state change time points in each sequence and generate a runtime list. After establishing a secondary indicator matrix, the error level weight is set to 0.7 and the runtime weight is set to 0.3. A weighted summation of the error level value and the runtime level value is performed to generate a total score value. The top-ranked sequence indexes are extracted by score sorting and the retained states are marked. A retained marker array is constructed and a mutation marker matrix is ​​simultaneously generated based on the index to obtain the preferred parameter index. Optimal combination generation submodule: Based on the optimal parameter index, the retained state and the corresponding variable number index are extracted, the control vector corresponding to the variable number in the original control parameter set is extracted, a mapping matching operation is performed to perform a one-to-one search between the variable index and the record in the control parameter table and the corresponding control value is extracted, a combination reconstruction operation is performed to reassemble the extracted control vector list in a unified order to generate a control vector set, the mutation mark matrix is ​​superimposed to generate multiple groups of structural perturbation candidates, the parameter perturbation operation is performed on the position with a value of 1 in the mark matrix, and a new parameter group is constructed based on the corresponding parameter in the positive and negative directions and together with the original parameters, a candidate vector group is formed, all control vectors and mutation flag labels are summarized, and organized into a two-dimensional array structure in sequence to construct an optimal combination of control parameters.

[0024] The genetic algorithm first uses the variable sequence set obtained after time axis alignment and invalid node screening as the initial population, with each sequence corresponding to a chromosome. The fitness function is then used to evaluate the fitness of each chromosome in the population, using the weighted sum of the simulation error integral area and the running time as the fitness function. The selection operator then selects the parent from the chromosomes with high fitness, and reorganizes the time point and node number mapping table of the parent sequence through a single-point or multi-point crossover operator to obtain the offspring chromosomes. The mutation operator is then used to slightly perturb the switching time or connection information of some offspring chromosomes. The parents and offspring are sorted by fitness, and the top N optimal chromosomes are retained as the new generation population. The iteration is repeated until the convergence condition is met and the preset algebraic upper limit is reached, and the optimal control parameter combination scheme is output. Genetic algorithm, according to the formula: ; in: Indicates the candidate node index sequence in the first The updated permutation result in the genetic iteration, Indicates the candidate node index sequence in the first The current permutation state in iterations, represents the global adjustment coefficient, Represents the weight of the fitness difference component in the index update, represents the weight of the channel switching time consistency component, represents the weight of the node topological connectivity score, represents the weight of the frequency of historical elite selection, Indicates in The individual fitness value with the minimum number of switch state conflicts and the best connectivity performance in the iteration, Indicates the Individuals in the The current fitness value in iterations, Indicates the The mean square error of the individual channel switching time point series, Indicates the The score of the proportion of connected edges in the node connection matrix constructed by the individual, Indicates the The frequency ratio of an individual being selected as an elite individual in successive genetic iterations; Execution process: First, extract the key event timestamps recorded by each sample in multiple processing channels from multiple monitoring nodes to build the initial node index sequence , and then calculate the fitness value of the current sequence under indicators such as task process integrity, sampling path rationality, and recording time consistency , and select the individual with the best fitness from all candidates to obtain Then, the time balance of each node on the event flow axis is evaluated, and the time consistency parameter is generated by the standard deviation of the key time points of each channel. , the actual connectivity of the computing nodes appearing in the sample processing path constitutes an indicator , that is, whether the node pair has actually established a physical or logical transmission path in the task, and at the same time, the frequency ratio of the node sequence being selected as a high-quality sample in the evolution process is counted , forming the historical preferred weight, and multiplying it by the normalized weight 、 、 、 , comprehensively expresses the performance of a candidate sample processing sequence in terms of traceability, stability, structural rationality and time continuity, and combines the weighted results with Multiply and add to the current node index sequence , and finally generate the updated sequence , achieving evolutionary optimization of sample processing paths and high-precision construction of traceable sequences.

[0025] The reduced-order mapping module includes: The feature matrix generation submodule: Based on the optimal combination of control parameters, the voltage and current acquisition method is used to read the energy storage unit voltage and filter current values, a time index is used to construct a two-dimensional sequence matrix, a column vector energy decomposition method is used to divide the energy distribution, and peak contribution screening is used to identify the main vector set to generate the principal element feature matrix; Low-dimensional mapping derivation submodule: Based on the principal element feature matrix, matrix mapping is used to calculate the output coefficient list, residual variance signal accumulation is used to determine the projection dimension, combined screening is used to refine the minimum dimensional vector array, and time rearrangement is used to reconstruct the low-dimensional feature sequence to obtain a reduced-order low-dimensional expression; Feature matrix generation submodule: Based on the optimal combination of control parameters, the fast Fourier transform algorithm is used, the sampling rate is set to 1000 Hz, the sampling window length is 256 samples, and the overlap is 128 samples. The energy storage unit voltage and filter current values ​​are read and stored in the timing buffer. The sampling timestamp is used to construct a two-dimensional sequence matrix. The vector is extracted by column and the fast Fourier transform operation is performed. The amplitude and phase of each frequency component are calculated. The peak detection algorithm is used to set the threshold to the maximum amplitude multiplied by 0.8 to filter the main frequency index. The corresponding spectrum vector is extracted based on the filtered index and forms a main frequency vector set to generate the main element feature matrix. Low-dimensional mapping derivation submodule: Based on the principal element feature matrix, the principal component analysis algorithm is adopted, and the maximum number of principal components is set to 5. The feature matrix is ​​first row-centered, the covariance matrix of the centralized matrix is ​​calculated, and the eigenvalue and eigenvector solution operations are performed to obtain the eigenvalue list and eigenvector matrix. The eigenvalues ​​are arranged in descending order to calculate the cumulative variance contribution rate. The cumulative variance threshold is set to 0.90 to screen the minimum number of principal components and extract the corresponding eigenvectors to generate a projection matrix. The projection matrix is ​​used to perform matrix multiplication on the centralized matrix and rearrange it according to the original time series order to generate a low-dimensional feature sequence to obtain a reduced-order low-dimensional expression.

[0026] The state reconstruction module includes: Input matrix construction submodule: Based on the reduced-order low-dimensional representation, it collects the switching device status data and harmonic interference channel signals corresponding to each stage, uniformly resamples the data of different frequency bands using equal time step interpolation, constructs an index table to synchronously annotate time labels and variable names, forms a three-dimensional matrix structure through node number mapping and harmonic frequency band splitting, performs tensor expansion operation to compress it into a two-dimensional input structure, and obtains the input matrix; Weight iteration submodule: Based on the input matrix, a residual network is used, and the initial weight vector is introduced as the reference value of the superposition coefficient. The residual calculation operation is performed to evaluate the degree of waveform offset under the current combination. The alignment index is updated using the maximum offset position. The difference in weights of adjacent time steps is corrected through the local error compression method. The iterative update is performed and the results of each step are retained in the record sequence. The iteration is repeated until the offset change is lower than the threshold, and the weight iteration sequence is obtained. Waveform output submodule: Based on the weighted iterative sequence, feature segments are extracted from the input matrix according to the time dimension and weighted superposition is performed to generate the initial waveform segment. Continuous frame splicing operations are performed to restore the complete waveform sequence. The consistency of overlapping segments in the critical region is verified through the sliding window mode. Boundary transition correction and length normalization are completed to generate a stable and traceable waveform output table and obtain the reconstructed waveform sequence table. Input matrix construction submodule: Based on the reduced-order low-dimensional expression, the state data of the switching devices in each stage and the harmonic interference channel signal are collected, and the linear amplitude interpolation method is used to uniformly resample the data of different frequency bands. The unified time step is set to 1 millisecond, and the original sampling time series is converted to a standard time axis. The time steps in each channel data are aligned in channel order and the corresponding amplitude points are inserted. The interpolation method adopts equidistant interpolation. The amplitude at the target time point is obtained by performing linear segmented interpolation on the amplitudes of adjacent time points. The index table is constructed to synchronously mark the time label and variable name. The time label is in milliseconds from the start time in sequence. Record incrementally, and use the numbering sequence to match the corresponding channel data in a unified manner for variable names. Use the node number mapping method to map the channel number to the topological node index to which it belongs. Use the node number to distinguish the frequency band channels. Group each segment of signal data by the set frequency band according to the frequency division. Classify the signal values ​​of the same frequency band into a unified frequency channel set. Use the frequency band grouping structure to form a three-dimensional matrix structure. The three-dimensional matrix dimension is set to the number of frequency bands multiplied by the time step multiplied by the number of channels. Perform the tensor expansion operation and compress it into a two-dimensional input structure in a channel-first manner. The dimension is set to the total number of channels multiplied by the time step to obtain the input matrix. Weight iteration submodule: Based on the input matrix, a residual network structure is used to perform multi-layer step-by-step correction operations. The initial weight vector is introduced as the reference value of the superposition coefficient. The length of the initial weight vector is consistent with the time step of the input matrix, and all values ​​are set to 1. The residual calculation operation is performed to evaluate the degree of waveform offset under the current combination. The fitted waveform calculated at each time step is numerically subtracted from the original waveform. The results are arranged into an offset sequence according to the time dimension. The time alignment index is updated using the position of the maximum offset in the offset sequence. The offset sequence is traversed through a sliding window with a window length of 5 time steps. The mean error in each window is calculated, and a compression adjustment operation is performed on the weights between adjacent window segments with large error differences. The adjustment method is to compare the average offset difference between the previous and next two time periods, introduce an adjustment amount in the opposite direction into the weight vector for fine-tuning, and set the adjustment step size to 0.01. After each round of iteration, the current weight vector is recorded and saved in the record sequence. The iteration is repeated until the change amplitude of the offset sequence in two consecutive rounds of iteration is lower than the set threshold of 0.001, and the weight iteration sequence is obtained. Waveform output submodule: Based on the weighted iterative sequence, feature segments are extracted from the input matrix according to the time dimension and weighted superposition is performed to generate the initial waveform segment. Each input signal segment is matched one-to-one with the weight of the corresponding time point according to a unified time step. The product of the corresponding values ​​is used as the weighted amplitude. All time segments are arranged in the original time order, and the continuous frame splicing operation is performed by sliding splicing. The sliding window length is set to 50 time steps and the step size is 25. The linear interpolation fusion operation is performed on the overlapping segments between each two consecutive frames. The fusion method is to linearly superimpose the tail of the front segment and the head of the back segment in an equal weight ratio to generate a smooth transition segment of the splicing area. All spliced ​​waveform segments are sequentially merged into a complete waveform sequence. The waveform sequence length is normalized. By setting the target length to 1024 time steps and redistributing the original waveform sequence data points using equal time intervals, a unified time length reconstruction is completed to generate a reconstructed waveform sequence table. Residual network, first input sample set, introduce initialized weight vector as the initial parameter of each connection layer in residual network, then expand according to time series structure, each layer of neural units performs linear transformation and activation function combination operation, generates predicted output sequence, and calculates difference with target waveform to obtain residual sequence, then uses mean square error as loss function to quantize residual size, backpropagates error information, updates network weight, compares the offset degree of current output waveform with target reference waveform in each iteration, extracts offset point and adjusts time alignment index, and compresses and adjusts the local weight difference of previous and next time steps to reduce oscillation, the iterative process continues until the residual change is lower than the set threshold, and outputs the converged weight iteration sequence as stable parameter set; Residual network, according to the formula: ; in: Indicates in The weight vector generated in the iteration is Indicates in The current weight vector in iteration , represents the learning rate of the residual network, represents the sampled power system input matrix, Represents the input matrix The transpose operation, represents the desired target waveform vector, represents the weight coefficient of the residual smoothing term, Indicates the The residual vector of the iteration, represents the weight coefficient of the regularization term, represents the regularization gradient applied to the current weight vector, represents the penalty coefficient of the residual variance term, Indicates the The variance of the residual vector in the iteration, represents the weight coefficient of the historical momentum term, Indicates the The momentum term accumulated in the iterations; Execution process: First, obtain the input matrix from multiple detection channels , each column represents the response waveform of the radiation sample in a specific process node or measurement link, extracting the expected output waveform As a reference standard, it reflects the ideal response trajectory of the target sample in the whole process and initializes the weight vector , and calculate the current residual , substitute the residual into the formula to get the gradient update direction , combined with the weighted adjustment of the residual term itself Used to alleviate the impact of local fluctuations and then introduce regularization gradient Reflects the constraints on the controllability and physical rationality of the weights of each link in the traceability system, and multiplies it by the coefficient Suppress overfitting and then introduce As the first The round residual variance reflects the error discreteness of the system in the time dimension. After adjustment, the overall stability of the sequence fitting is improved, and finally the momentum term is combined Reflects the cumulative trend of the weight evolution direction in the previous rounds, through Weighted balance of convergence speed and correction step size Combine the above items with learning rate Multiply and add to the current weight , get the next round of updated weights , repeat and record each round of results until the residual fluctuation converges to the system set threshold, generating a complete weight iteration sequence for achieving traceable reconstruction of each signal node during the radiation experiment.

[0027] The process scheduling module includes: Task unit division submodule: Based on the reconstructed waveform sequence table, the continuous amplitude rising segment detection method is adopted. The continuous amplitude falling segment identification method is adopted in the superimposed discharge segment extraction operation. The zero-crossing interval extraction method is used in combination with the resonance segment extraction operation to complete the division. The superimposed equal-interval sampling operation extracts the peak and valley point series. The start and end times of each segment are matched to form a duration statistics table to obtain the task timing distribution; The trigger structure construction submodule extracts the parallel segment numbers based on the task timing distribution, superimposes the trigger sequence generation operation to construct a time series number group, performs the dependency path sorting operation to establish a sequence chain according to the adjacent task numbers, constructs a trigger priority grouping table, generates a priority index based on the waiting ratio, and obtains the scheduling trigger structure. Based on the reconstructed waveform sequence table, the amplitude continuous rising segment detection method is adopted, the detection window length is set to 5 time steps, and a traversal comparison operation is performed on each point. If the current amplitude is greater than the previous amplitude value, the index is added to the rising sequence cache. Otherwise, it is determined whether the cache length is greater than or equal to 5, and the cache start and end indexes that meet the conditions are recorded as rising segments. After the cache is cleared, the traversal is continued. The amplitude continuous falling segment recognition method is adopted for the superimposed discharge segment extraction operation. The falling detection window length is set to 5 time steps. Amplitude comparison is performed on each point. If the current amplitude is less than the previous amplitude value, the index is appended to the falling sequence cache. Otherwise, when the cache length is greater than or equal to 5, the cache start and end are recorded. After the index record is a descending segment, the cache is cleared. The zero-crossing interval extraction method is used in combination with the resonance segment extraction operation to traverse the complete waveform sequence table. For each adjacent sample pair, if the signs of the previous and next samples are opposite, the phase crossing time point index is recorded and the resonance segment is defined according to the interval of the two crossing points. The peak and valley point list is extracted by superimposing the equal-interval sampling operation and setting the sampling interval to 10 time steps. The maximum and minimum amplitude point indexes are taken out from each ascending and descending segment at a fixed step size to form a peak and valley index list. The start and end time of each segment are matched to merge the peak and valley point indexes of the ascending, descending and resonance segments, and the duration is calculated with the start index and end index and written into the duration statistics table to obtain the task timing distribution. Based on the task timing distribution, the parallel segment numbers are extracted. A topological sorting algorithm is used to initialize the in-degree mapping of all task segment nodes and build an adjacency list. The dependency relationship is traversed for each task segment number. If the end time of the predecessor task segment is less than the start time of the successor task segment, a directed edge is added to the adjacency list and the in-degree of the successor task segment is increased by one. A trigger sequence generation operation is constructed. Task segment numbers with in-degree zero are filtered in the in-degree mapping and added to the queue. They are then dequeued and written into the time series number group. When writing, the in-degree of all successor nodes in the adjacency list is reduced by one. If the in-degree is zero after decrement, the nodes are added to the queue and continue processing. A dependency path sorting operation is performed. A sequence chain is established according to the queue dequeue order. Each dequeue node and its dequeue time are recorded in sequence as a dependency path table. A trigger priority grouping table is constructed to generate a priority index based on the waiting ratio. The waiting ratio threshold is set to 0.5. All waiting times are divided by the longest waiting time to calculate the normalized value. If the normalized value is greater than the threshold, it is marked as high priority, otherwise it is marked as low priority. The task segment number and priority index are mapped and written into the priority grouping table to obtain the scheduling trigger structure.

[0028] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. An electromagnetic transient simulation system for a power system containing multiple power electronic switches, characterized in that: The system comprises: Topology analysis module: Based on the parameters of each switch node, inductance, capacitance, and resistance in multiple power electronic switch devices, it extracts the switch number and topology sequence to perform impedance calculation and admittance conversion, normalizes the matrix elements, and maps the node connection structure to generate the network topology matrix; Parameter evolution module: Based on the network topology matrix, a genetic algorithm is used in combination with the timing of switching devices to establish a variable sequence by extracting switching time points and connection information. The simulation runs and records errors and durations, sorts and determines retention and mutation, and generates an optimal combination of control parameters. Reduced-order mapping module: Based on the preferred combination of control parameters, collects the energy storage unit voltage and filter current sequence, constructs a time matrix and performs singular value calculation and eigenvector selection, calculates the projection coefficient and determines the residual adjustment dimension to obtain a reduced-order low-dimensional expression; State reconstruction module: Based on the reduced low-dimensional expression, a residual network is used to construct an input matrix by combining the switch state and harmonics, the time step and coefficient are matched to construct a combined sequence, the offset difference is calculated and the weight is adjusted and updated iteratively to obtain a reconstructed waveform sequence table; Process scheduling module: Based on the reconstructed waveform sequence table, combined with the capacitor charging and discharging and resonance stages, the task unit is defined, the task duration and waiting ratio are counted, the parallel conditions and triggering sequence are determined, and a scheduling trigger structure is established.

2. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 1, characterized in that: The topology analysis module includes: Node parameter mapping submodule: Based on the parameters of each switch node, inductor, capacitor, and resistor in the multi-power electronic switch device, node information matching is used to extract the switch number and calibrate the topological sequence number. The complex impedance value array is constructed by electrical parameter conversion. The real and imaginary parts of the admittance are output by numerical conversion. The admittance elements are scaled to form a vector set, generating a normalized admittance data set. Connection structure construction submodule: Based on the normalized admittance data set, index mapping is used to process the correspondence between node numbers and vectors, matrix assembly is used to construct the node pair connection matrix, element fusion is used to present the two-dimensional mapping diagram of admittance and structure, and the multi-node link information set is summarized and output to obtain the network topology matrix.

3. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 1, characterized in that: The parameter evolution module includes: Switching sequence generation submodule: Based on the network topology matrix, the switching time points of the switching devices and the corresponding connection node numbers are extracted. A genetic algorithm is used to construct a node index mapping table in combination with the topological row and column labels. A timing axis alignment operation is performed to correct the time points of each channel. A connection information filtering operation is performed to remove invalid node states. The effective switching segments in the entire time domain are sorted. The variable state list is spliced ​​in time sequence to construct a complete sequence structure and fill in missing fields to obtain a variable sequence set. Error screening submodule: Based on the variable sequence set, it performs circuit model simulation calculation to output waveform groups, calculates the error curve between the target waveform and the actual response through residual extraction operation, generates sequence quality rankings by sorting the error integrated area, performs runtime statistics and establishes a secondary indicator matrix, combines error rankings and runtime rankings through weight allocation, extracts top-ranked sequences and adds retention marks, and simultaneously generates a variation mark matrix to obtain the optimal parameter index; Optimal combination generation submodule: Based on the preferred parameter index, the retained state and the corresponding variable number index are extracted, a mapping matching operation is performed to associate the original control parameter data, a structurally complete control vector set is formed through a combined reconstruction operation, the mutation labeling matrix is ​​superimposed to generate multiple groups of structural perturbation candidates, different control vector sets and corresponding mutation marker labels are summarized, a two-dimensional control parameter set array is constructed, and the optimal combination of control parameters is obtained.

4. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 3, characterized in that: The genetic algorithm first uses a variable sequence set obtained after time axis alignment and invalid node screening as the initial population, with each sequence corresponding to a chromosome. Then, the fitness of each chromosome in the population is evaluated using the weighted sum of the simulation error integral area and the running time as the fitness function. Then, a selection operator selects the parent from the high-fitness chromosomes, and reorganizes the time point and node number mapping table of the parent sequence through a single-point or multi-point crossover operator to obtain the offspring chromosomes. Subsequently, a mutation operator is used to slightly perturb the switching time or connection information of some offspring chromosomes. At the same time, the parents and offspring are sorted according to fitness, and the top N optimal chromosomes are retained as the new generation population. Iteration is repeated until the convergence condition is met and the preset generation upper limit is reached, and the optimal control parameter combination scheme is output.

5. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 3, characterized in that: The genetic algorithm is based on the formula: ； in: Indicates the candidate node index sequence in the first The updated permutation result in the genetic iteration, Indicates the candidate node index sequence in the first The current permutation state in iterations, represents the global adjustment coefficient, Represents the weight of the fitness difference component in the index update, represents the weight of the channel switching time consistency component, represents the weight of the node topological connectivity score, represents the weight of the frequency of historical elite selection, Indicates in The individual fitness value with the minimum number of switch state conflicts and the best connectivity performance in the iteration, Indicates the Individuals in the The current fitness value in iterations, Indicates the The mean square error of the individual channel switching time point series, Indicates the The score of the proportion of connected edges in the node connection matrix constructed by the individual, Indicates the The frequency ratio of an individual being selected as an elite individual in successive genetic iterations.

6. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 1, characterized in that: The reduced-order mapping module includes: Characteristic matrix generation submodule: Based on the preferred combination of the control parameters, the voltage and current acquisition method is used to read the energy storage unit voltage and filter current values, a time index is used to construct a two-dimensional sequence matrix, a column vector energy decomposition method is used to divide the energy distribution, and a peak contribution is used to screen and identify the main vector set to generate a principal element characteristic matrix; Low-dimensional mapping derivation submodule: Based on the principal component feature matrix, matrix mapping is used to calculate the output coefficient list, residual variance signal accumulation is used to determine the projection dimension, combined screening is used to refine the minimum dimension vector array, and time rearrangement is used to reconstruct the low-dimensional feature sequence to obtain a reduced-order low-dimensional expression.

7. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 1, characterized in that: The state reconstruction module includes: Input matrix construction submodule: Based on the reduced low-dimensional expression, the switching device state data and harmonic interference channel signals corresponding to each stage are collected, and the data of different frequency bands are uniformly resampled using the equal time step interpolation method. An index table is constructed to synchronously annotate time labels and variable names. A three-dimensional matrix structure is formed through node number mapping and harmonic frequency band splitting. A tensor expansion operation is performed to compress it into a two-dimensional input structure to obtain the input matrix; Weight iteration submodule: Based on the input matrix, a residual network is used, an initial weight vector is introduced as a reference value of the superposition coefficient, a residual calculation operation is performed to evaluate the degree of waveform offset under the current combination, the alignment index is updated using the maximum offset position, the difference in weights of adjacent time steps is corrected through the local error compression method, an iterative update is performed and the result of each step is retained in the record sequence, and the iteration is repeated until the offset change is lower than the threshold, thereby obtaining a weight iteration sequence; Waveform output submodule: Based on the weighted iterative sequence, feature segments are extracted from the input matrix according to the time dimension and weighted superposition is performed to generate initial waveform segments, continuous frame splicing operations are performed to restore the complete waveform sequence, the consistency of critical area overlapping segments is verified through the sliding window mode, boundary transition correction and length normalization processing are completed, a stable and traceable waveform output table is generated, and a reconstructed waveform sequence table is obtained.

8. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 7, characterized in that: The residual network first inputs a sample set and introduces an initialized weight vector as the initial parameter of each connection layer in the residual network. Then, it is expanded according to the time series structure. Each layer of neural units performs a combination of linear transformation and activation function operations to generate a predicted output sequence, and calculates the difference with the target waveform to obtain a residual sequence. Then, the mean square error is used as a loss function to quantize the residual size, the error information is back-propagated, and the network weights are updated. In each round of iteration, the offset degree of the current output waveform and the target reference waveform is compared, the offset point is extracted and the time alignment index is adjusted, and the local weight difference between the previous and next time steps is compressed and adjusted to reduce oscillation. The iterative process continues until the residual change is lower than the set threshold, and the converged weight iterative sequence is output as a stable parameter set.

9. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 7, characterized in that: The residual network, according to the formula: ； in: Indicates in The weight vector generated in the iteration is Indicates in The current weight vector in iteration , represents the learning rate of the residual network, represents the sampled power system input matrix, Represents the input matrix The transpose operation, represents the desired target waveform vector, represents the weight coefficient of the residual smoothing term, Indicates the The residual vector of the iteration, represents the weight coefficient of the regularization term, represents the regularization gradient applied to the current weight vector, represents the penalty coefficient of the residual variance term, Indicates the The variance of the residual vector in the iteration, represents the weight coefficient of the historical momentum term, Indicates the The momentum term accumulated over iterations.

10. The power system electromagnetic transient simulation system containing multiple power electronic switches according to claim 1, characterized in that: The process scheduling module includes: Task unit division submodule: Based on the reconstructed waveform sequence table, the continuous amplitude rising segment detection method is adopted, the continuous amplitude falling segment identification method is adopted in the superimposed discharge segment extraction operation, and the zero crossing interval extraction method is used in combination with the resonance segment extraction operation to complete the division. The superimposed equal interval sampling operation is used to extract the peak and valley point series, and the start and end time of each segment is matched to form a duration statistics table to obtain the task timing distribution; Trigger structure construction submodule: Based on the task timing distribution, extract the parallel segment numbers, superimpose the trigger sequence generation operation to construct a time series number group, perform the dependent path sorting operation to establish a sequence chain according to the adjacent task numbers, construct a trigger priority grouping table to generate a priority index according to the waiting ratio, and obtain the scheduling trigger structure.

Images