Method and device for solving weak link of converter valve, storage medium and computer equipment

Abstract

The present application relates to the technical field of electrical equipment fault prediction, and discloses a method and device for solving weak links of a converter valve, a storage medium and computer equipment, the method comprising: establishing a substructure model corresponding to a system of a to-be-judged converter valve, and respectively performing interpolation optimization to obtain multiple reduced-order substructure models; then distributing the reduced-order substructure models to multiple processors for parallel processing, calculating and analyzing the results by using the isogeometric analysis method, and solving the analysis results by using a multi-objective evolutionary algorithm; optimizing and iterating the solution of each finite element model by using a Pareto transfer strategy to obtain an optimal parameter set; constructing a nonlinear equation according to the operation data of the to-be-judged converter valve; solving the nonlinear equation by using the optimal parameter set; and identifying temperature abnormal points of the converter valve to determine weak links. The above method improves the speed and accuracy of identifying weak links of the converter valve, accurately locates the temperature abnormal area, and provides clear fault diagnosis information for maintenance personnel.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical equipment fault prediction, and in particular to a thyristor valve weak link solving method and device, a storage medium and computer equipment. BACKGROUND

[0002] The thyristor valve is a key equipment in high-voltage direct current engineering, and has high current-carrying capacity, and is particularly suitable for application scenarios of large-capacity and long-distance power transmission. Considering that the structure of the thyristor valve is very complex, and long-term operation in a complex electromagnetic multi-physical field interwoven harsh working environment will accelerate the device aging process and increase the failure rate, the thyristor valve operation has a relatively weak reliability problem.

[0003] The weak link of the thyristor valve mainly refers to the temperature high abnormal heating point. At present, the operation and maintenance process of the thyristor valve mainly relies on manual inspection, robot inspection and regular shutdown maintenance, and the existing thyristor valve monitoring technology and valve hall monitoring method are limited to the volume structure of the sensor and the mutual layering of the valve tower, and cannot accurately perceive the working state of the key components in the valve tower, and there are problems such as slow speed and low precision in discriminating the weak link of the thyristor valve. SUMMARY

[0004] Therefore, the present application provides a thyristor valve weak link solving method and device, a storage medium and computer equipment, which mainly aims to solve the technical problems of slow speed and low precision in discriminating the weak link of the thyristor valve in the prior art.

[0005] According to a first aspect of the present application, a thyristor valve weak link solving method is provided, which comprises:

[0006] Identifying a plurality of subsystems of a to-be-discriminated thyristor valve, establishing a corresponding substructure model for each of the subsystems, and respectively performing interpolation optimization on a plurality of the substructure models to obtain a plurality of reduced-order substructure models;

[0007] Parallel processing of the plurality of reduced-order substructure models in a plurality of processors, constructing a finite element model, and calculating an analysis result of each of the finite element models by using an isogeometric analysis method;

[0008] Determining the complexity of a plurality of the finite element models, and solving the analysis result of the finite element model by using a multi-objective evolutionary algorithm to obtain a solution of each of the finite element models, and optimizing and iterating the solution of each of the finite element models by using a Pareto transfer strategy to obtain an optimal parameter set;

[0009] According to the operation data of the to-be-discriminated thyristor valve, a nonlinear equation is constructed, the optimal parameter set is used to solve the nonlinear equation, and a temperature abnormal point of the to-be-discriminated thyristor valve is identified to determine the weak link of the to-be-discriminated thyristor valve.

[0010] Optionally, the interpolation optimization is performed on the plurality of substructure models respectively to obtain a plurality of reduced-order substructure models, comprising:

[0011] Collecting experimental data of the to-be-judged thyristor, and preprocessing the experimental data;

[0012] Determining a target interpolation method based on data characteristics and data requirements of the experimental data;

[0013] Establishing an interpolation model according to the target interpolation method, and performing interpolation optimization on the substructure models by using the interpolation model to obtain an initial reduced-order substructure model;

[0014] Constructing an auxiliary variable, and adjusting model parameters of the initial reduced-order substructure model based on the auxiliary variable to optimize the initial reduced-order substructure model and obtain a reduced-order substructure model.

[0015] Optionally, the interpolation optimization is performed on the plurality of substructure models respectively to obtain a plurality of reduced-order substructure models, comprising:

[0016] Obtaining an electron current equation and a hole current equation of the to-be-judged thyristor under one-dimensional conditions;

[0017] Constructing an auxiliary variable, and obtaining the electron current equation and the hole current equation under different injection conditions based on the electric field distribution of the power chip in the to-be-judged thyristor, and then solving to obtain the electron current and the hole current at discrete points in the electron current equation and the hole current equation under one-dimensional conditions;

[0018] Determining a reduced-order substructure model according to the electron current equation and the hole current equation under one-dimensional conditions and the electron current and the hole current at discrete points.

[0019] Optionally, the plurality of reduced-order substructure models are distributed to a plurality of processors for parallel processing, comprising:

[0020] Determining the complexity of each reduced-order substructure model based on the grid subdivision fineness and the calculation complexity;

[0021] Determining the calculation units of a plurality of processors, matching the calculation units of the processors with the complexity of the reduced-order substructure models, and determining the processors corresponding to each reduced-order substructure model;

[0022] Parallel processing of the plurality of reduced-order substructure models by using the plurality of processors respectively.

[0023] Optionally, the finite element model is constructed and the analysis results of each finite element model are calculated by using the isogeometric analysis method, comprising:

[0024] constructing a finite element model of the reduced order substructure model according to a transient thermal structural weak coupling problem;

[0025] analyzing the finite element model based on initial design parameters and initial constraint conditions by using an isogeometric analysis method to obtain initial analysis results;

[0026] optimizing the finite element model based on the initial analysis results and re-estimating the optimized finite element model by using a reanalysis algorithm to obtain analysis results corresponding to the finite element model.

[0027] Optionally, the complexity of the plurality of finite element models is determined, and the analysis results of the finite element models are solved by using a multi-objective evolutionary algorithm to obtain a solution of each finite element model, including:

[0028] The complexity of the finite element model is determined by using a complexity dynamic adjustment method, wherein the finite element model includes a low-complexity model and a high-complexity model;

[0029] According to a multi-objective optimization problem, related objective functions and various constraint conditions are established, the analysis results of the finite element model are solved by using a multi-objective evolutionary algorithm based on minimizing the related objective functions to obtain a solution of each finite element model.

[0030] Optionally, the solution of each finite element model is iteratively optimized by using a Pareto transfer strategy to obtain an optimal parameter set, including:

[0031] The solution of the low-complexity model is obtained, and the solution of the low-complexity model is taken as an initial search population of the high-complexity model;

[0032] The search population of the high-complexity model during iteration is determined, a local search function is obtained, the local search function is updated based on the parameters of the high-complexity model and the search population to obtain an optimal parameter set.

[0033] According to a second aspect of the present application, a device for solving weak links of a thyristor is provided, and the device comprises:

[0034] A data acquisition module is configured to acquire simulation data samples of a multi-physical field of the thyristor under all working conditions by using a preset multi-physical field simulation model of the thyristor.

[0035] A data expansion module is configured to construct a least square generative adversarial network, train the least square generative adversarial network based on the simulation data samples to obtain a multi-physical field parameter sample, and combine the simulation data samples and the multi-physical field parameter sample to obtain an expanded data sample.

[0036] A model evaluation module is configured to preprocess the extended data samples, train and evaluate a preset evaluation model by using the preprocessed extended data samples, and generate a target evaluation model.

[0037] A result output module is configured to establish a fault prediction model based on the extended data samples and the target evaluation model, collect multi-physical field data of the converter valve, process the multi-physical field data of the converter valve by using the fault prediction model, and obtain a converter valve fault prediction result.

[0038] According to a third aspect of the present application, a storage medium is provided, which stores a computer program. The program is executed by a processor to implement the method for solving weak links of a converter valve.

[0039] According to a fourth aspect of the present application, a computer device is provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the method for solving weak links of a converter valve.

[0040] The application provides a method and device for solving weak links of a converter valve, a storage medium and computer equipment. Firstly, a plurality of subsystems of a to-be-judged converter valve are identified, a corresponding substructure model of each subsystem is established, interpolation optimization is respectively performed on the plurality of substructure models, a plurality of reduced-order substructure models are obtained, the plurality of reduced-order substructure models are distributed to a plurality of processors for parallel processing, a finite element model is constructed, and the analysis result of each finite element model is calculated by using an isogeometric analysis method. Then, the complexity of the plurality of finite element models is determined, the analysis result of the finite element model is solved by using a multi-objective evolutionary algorithm, the solution of each finite element model is obtained, the solution of each finite element model is optimized and iterated by using a Pareto transfer strategy, an optimal parameter set is obtained, a nonlinear equation is constructed according to the operation data of the to-be-judged converter valve, the nonlinear equation is solved by using the optimal parameter set, a temperature abnormal point of the to-be-judged converter valve is identified, and the weak link of the to-be-judged converter valve is determined. The above method is based on the establishment of a substructure model of a subsystem of a converter valve, and the model is reduced by interpolation optimization, so that the calculation complexity is greatly reduced, and the simulation is more efficient. The plurality of reduced-order substructure models are processed in parallel by using a plurality of processors, so that the entire simulation process can be significantly accelerated, and the simulation efficiency is improved. The finite element model is constructed, and the isogeometric analysis method is combined, so that the calculation efficiency can be further improved while the high precision is maintained, and the complex physical field change inside the converter valve can be more accurately captured. The multi-objective evolutionary algorithm is adopted, which is suitable for solving problems with multiple conflicting objectives. The Pareto transfer strategy is used to optimize and iterate the model solution, so that the accuracy of predicting the temperature abnormal point can be improved. The nonlinear equation is constructed according to the actual operation data, and the parameter set after optimization is used for solving, so that the working state of the converter valve can be more realistically reflected. The above method improves the speed and accuracy of identifying the weak link of the converter valve, accurately locates the temperature abnormal area, and provides clear fault diagnosis information for maintenance personnel.

[0041] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0042] The drawings described herein are used to provide a further understanding of the present application, and form a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0043] Figure 1 A flowchart of a method for solving weak links of a converter valve provided by an embodiment of the present application is shown;

[0044] Figure 2A thyristor converter valve internal power chip electric field distribution map in a weak link solving method of a converter valve provided by the embodiment of the application is shown.

[0045] Figure 3 A reanalysis algorithm-based extended finite element method block diagram in a weak link solving method of a converter valve provided by the embodiment of the application is shown.

[0046] Figure 4 A sub-geometric fast reanalysis algorithm flow chart in a weak link solving method of a converter valve provided by the embodiment of the application is shown.

[0047] Figure 5 A thyristor converter valve key device junction temperature distribution cloud chart in a weak link solving method of a converter valve provided by the embodiment of the application is shown.

[0048] Figure 6 A principle logic flow chart of a weak link solving method of a converter valve provided by the embodiment of the application is shown.

[0049] Figure 7 A structure schematic diagram of a weak link solving device of a converter valve provided by the embodiment of the application is shown.

[0050] Figure 8 A device structure schematic diagram of a computer equipment provided by the embodiment of the application is shown. DETAILED DESCRIPTION

[0051] Exemplary embodiments of the present application will be described herein below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it is understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thoroughly and completely understood, and so that the scope of the present application will be completely conveyed to those skilled in the art.

[0052] The embodiment of the present application provides a weak link solving method of a converter valve, as shown in the figure, the method comprises the following steps: Figure 1

[0053] 101、Identify a plurality of subsystems of a converter valve to be judged, establish a corresponding substructure model of each subsystem, and respectively interpolate and optimize a plurality of substructure models to obtain a plurality of reduced-order substructure models.

[0054] ​Taking the KPE4500-72 thyristor converter valve as an example, the entire converter valve needs to be decomposed first to identify the key components or subsystems that constitute the converter valve. The subsystems include but are not limited to thyristors, heat sinks, capacitors, resistors, etc. Each subsystem has its own specific functions and physical characteristics. For each identified subsystem, a mathematical or physical model that can reflect its main characteristics is established to accurately capture the dynamic behavior of the subsystem and how it interacts with other subsystems. For example, for thyristors, its conduction characteristics and thermal characteristics may need to be considered; for heat sinks, its heat dissipation efficiency, etc., needs to be considered. Then, interpolation technology or other numerical methods are applied to each substructure model for optimization. The purpose is to reduce the degrees of freedom in the model without sacrificing too much accuracy. The resulting reduced-order substructure model can be calculated and analyzed more quickly while maintaining sufficient accuracy to predict temperature anomalies and other potential problems.

[0055] In the embodiment of the present application, the reduced-order substructure model reduces the amount of calculation, making the simulation process more efficient and easier to understand and maintain. Each substructure model is relatively independent, and new components can be easily added or existing components can be modified as needed, thereby enhancing the flexibility of the system. Through detailed modeling of each subsystem, the working conditions inside the converter valve can be analyzed more carefully, so that potential problem areas can be discovered earlier, and the weak links of temperature anomalies can be identified more quickly and accurately. After the substructure model is established, it is necessary to use existing high-precision model calculation results or test data to further improve these models. Through model reduction technology, it is possible to use existing data sets for interpolation to obtain a more accurate model.

[0056] 102. Allocate multiple reduced-order substructure models to multiple processors for parallel processing, construct finite element models, and calculate the analysis results of each finite element model using the isogeometric analysis method.

[0057] The parallel computation is realized by assigning different reduced substructure models to multiple processors or computing cores, each processor being responsible for processing one or several substructure models; the finite element model is a numerical simulation technique that discretizes a continuous system into a series of small elements, and then solves partial differential equations or other mathematical expressions on these elements to obtain an approximate solution of the entire system; the isogeometric analysis method is an advanced finite element analysis method that uses the same basis functions in computer-aided design for geometric description and approximation of field variables, and can provide higher accuracy and better smoothness compared with the traditional finite element method, especially when dealing with complex shapes and high-order continuity fields, the isogeometric analysis method solves each finite element model to obtain the response of each substructure under a specific working condition, such as stress distribution, temperature field, electromagnetic field, etc., and the analysis results can be used to evaluate the working state of each substructure and identify potential problem areas.

[0058] In the embodiments of the present application, the distributed computing method can make full use of modern multi-core CPU or cluster computing resources, significantly speed up the entire simulation process, and greatly shorten the overall computing time through parallel processing; the isogeometric analysis method provides more accurate results than the traditional finite element method, which helps to better predict the thermal behavior and other physical phenomena inside the converter valve, thereby improving the safety and reliability of the device operation; the fast and accurate computing capability makes the technical solution suitable for real-time monitoring systems to timely discover abnormal conditions of the converter valve.

[0059] 103. Determine the complexity of the plurality of finite element models, and solve the analysis results of the finite element models using a multi-objective evolutionary algorithm to obtain a solution of each finite element model, and optimize and iterate the solution of each finite element model using a Pareto transfer strategy to obtain an optimal parameter set.

[0060] Wherein, the complexity of each finite element model is evaluated, including factors such as mesh density, element type, material properties, etc., and the complexity is usually proportional to the computational cost, that is, the more complex the model, the more computational resources and time are required; the multi-objective evolutionary algorithm is a global search algorithm that can find the optimal compromise solution among multiple conflicting objectives, and using the multi-objective evolutionary algorithm to process the multi-objective optimization problem, which is specifically applied in the present application, can be used to find the best parameter set to minimize the computational cost while ensuring a certain accuracy; the fast reanalysis technique uses the information in the initial analysis results, and when the structure is slightly modified, the fast reanalysis technique can provide a new accurate estimate at a lower cost, avoiding the need to perform a complete finite element analysis from scratch after each modification, thereby greatly saving computing time; the variable complexity method combines the advantages of low complexity models and high complexity models, and in the iterative process, a low complexity model is first used to quickly explore the possible solution space, and then a high complexity model is switched to near the optimal solution to ensure the accuracy of the final result; pre-training can be based on historical data or existing knowledge to help the algorithm converge to a good solution faster, thereby speeding up the entire optimization process.

[0061] In the embodiments of the present application, by combining the multi-objective evolutionary algorithm and the variable complexity method, the overall computational cost can be significantly reduced while ensuring high accuracy, speeding up the optimization process, allowing the complexity of the model to be adjusted according to different needs, and being suitable for both rapid evaluation in the preliminary design stage and the needs of detailed design in the later stage; the fast reanalysis technique reduces unnecessary repeated calculations and improves computational efficiency; the Pareto transfer strategy helps to improve the quality of the initial solution, and the multi-objective evolutionary algorithm ensures continuous improvement of the quality of the solution during the entire optimization process until the best state is reached; due to the overall improvement in computing speed, real-time monitoring and dynamic adjustment can be better supported, and rapid response to changes can be achieved.

[0062] 104. Constructing a nonlinear equation according to the operation data of the to-be-judged thyristor, solving the nonlinear equation by using the optimal parameter set, identifying the temperature abnormal point of the to-be-judged thyristor to determine the weak link of the to-be-judged thyristor.

[0063] Wherein, since the working state of the converter valve is affected by multiple factors, and there is often a complex nonlinear relationship between these factors, it is usually necessary to construct a nonlinear equation describing the internal physical phenomena of the converter valve according to the actual operation data of the converter valve to accurately reflect this complexity; the optimal parameter set obtained by the multi-objective evolutionary algorithm and other optimization techniques is substituted into the nonlinear equation, so that the equation can more accurately reflect the actual situation, and a numerical method is applied to solve the nonlinear equation to obtain the specific working state of each subsystem, through the analysis of the solving result, the area whose temperature exceeds the normal range, i.e. the temperature abnormal point, is identified, wherein, the temperature abnormality is usually an early sign of overheating or local failure of equipment, based on the location and nature of the temperature abnormal point, combined with the structural characteristics and working principle, it is judged which part is the weak link of the converter valve, and the weak link may be caused by design defects, material fatigue, cooling system failure, etc.

[0064] In the embodiments of the present application, by constructing a nonlinear equation based on actual operation data and using the optimized parameter set, the real working state of the converter valve can be more accurately simulated, thereby improving the accuracy of fault diagnosis, and potential problem areas can be quickly identified to provide clear guidance for maintenance personnel, which helps to take action faster, reduce downtime and maintenance costs; early detection of temperature abnormal points helps to take preventive measures to avoid serious failures caused by overheating and prolong the service life of equipment; accurate positioning of weak links can be targeted for maintenance and repair, avoiding unnecessary large-scale inspection, thereby reducing the overall maintenance cost.

[0065] The application provides a method and device for identifying weak links of a converter valve, a storage medium and a computer device. Firstly, a plurality of subsystems of the converter valve to be identified are recognized, a corresponding substructure model of each subsystem is established, interpolation optimization is respectively performed on the plurality of substructure models, a plurality of reduced-order substructure models are obtained, the plurality of reduced-order substructure models are then distributed to a plurality of processors for parallel processing, a finite element model is constructed, and the analysis result of each finite element model is calculated by using the equivalent geometric analysis method. Then, the complexity of the plurality of finite element models is determined, the analysis result of the finite element model is solved by using a multi-objective evolutionary algorithm, the solution of each finite element model is obtained, the solution of each finite element model is optimized and iterated by using a Pareto transfer strategy, an optimal parameter set is obtained, finally, a nonlinear equation is constructed according to the operation data of the converter valve to be identified, the nonlinear equation is solved by using the optimal parameter set, the temperature abnormal point of the converter valve to be identified is recognized, and the weak link of the converter valve to be identified is determined. The above method is based on the establishment of a substructure model of a subsystem of a converter valve, and the model is reduced by interpolation optimization, which greatly reduces the calculation complexity and makes the simulation more efficient. The plurality of reduced-order substructure models are processed in parallel by using a plurality of processors, which can significantly speed up the entire simulation process and improve the simulation efficiency. The finite element model is constructed and combined with the equivalent geometric analysis method, which can further improve the solving efficiency while maintaining high precision and more accurately capture the complex physical field changes inside the converter valve. The multi-objective evolutionary algorithm is adopted, which is suitable for solving problems with multiple conflicting objectives, and the Pareto transfer strategy is used to optimize and iterate the model solution, which can improve the accuracy of predicting the temperature abnormal point. The nonlinear equation is constructed by using the actual operation data, and the parameter set after optimization is used for solving, which can more realistically reflect the working state of the converter valve. The above method improves the speed and accuracy of identifying the weak link of the converter valve, accurately locates the temperature abnormal area, and provides clear fault diagnosis information for maintenance personnel.

[0066] Specifically, in the above embodiment, interpolation optimization is respectively performed on the plurality of substructure models to obtain the plurality of reduced-order substructure models, including first collecting experimental data of the converter valve to be identified, and preprocessing the experimental data; then determining a target interpolation method based on the data characteristics and data requirements of the experimental data; then establishing an interpolation model according to the target interpolation method, and performing interpolation optimization on the substructure model by using the interpolation model to obtain an initial reduced-order substructure model; finally, constructing an auxiliary variable, adjusting the model parameters of the initial reduced-order substructure model based on the auxiliary variable, and optimizing the initial reduced-order substructure model to obtain the reduced-order substructure model.

[0067] In the embodiment, data including temperature distribution, current, voltage, mechanical stress and other key parameters are collected from actual operation or specially designed experiments of the converter valve to be identified, and the quality and quantity of the experimental data directly affect the accuracy of subsequent analysis; the collected data are cleaned to remove noise and outliers, and may also include data standardization, normalization and other processing to ensure the consistency and comparability of the data, and then according to the characteristics of the experimental data and the analysis requirements, a suitable interpolation method is selected to ensure sufficient accuracy while simplifying the model, and based on the selected interpolation method, a mathematical model is established to describe the relationship between the data points, and the original substructure model is interpolated and optimized by using the interpolation model to reduce the degrees of freedom of the model while trying to retain its key characteristics, and then a simplified, low-dimensional substructure model, i.e. an initial reduced-order substructure model, is generated, which can accurately reflect the behavior of the original substructure to a certain extent; and the introduction of auxiliary variables can help adjust and optimize the model parameters, and the parameters of the initial reduced-order substructure model are adjusted by using the auxiliary variables to further improve the accuracy and applicability of the model, and through an iterative optimization process, the model is continuously improved until a satisfactory performance index is reached, and the optimized reduced-order substructure model can more accurately represent the dynamic behavior of the actual system and has higher computational efficiency.

[0068] Further, an auxiliary variable is constructed, and the model parameters of the initial reduced-order substructure model are adjusted based on the auxiliary variable to optimize the initial reduced-order substructure model and obtain a reduced-order substructure model, including first obtaining electron current equations and hole current equations of the converter valve to be identified under one-dimensional conditions, then constructing an auxiliary variable, obtaining electron current equations and hole current equations under different injection conditions based on the electric field distribution of the power chip in the converter valve to be identified, and then solving to obtain electron current and hole current at discrete points in the electron current equation and the hole current equation under one-dimensional conditions, and finally determining the reduced-order substructure model according to the electron current equation and the hole current equation under one-dimensional conditions and the electron current and the hole current at the discrete points.

[0069] In the embodiment, the electron current and hole current equations under one-dimensional conditions are

[0070]

[0071] where n and p are the electron and hole concentrations in the base region, respectively, n = Δp + N B , p = Δp + n si 2 / N B ; n si is the intrinsic carrier concentration of silicon; μ n and μ p are the electron and hole mobilities, respectively; E is the electric field strength;

[0072] The electric field distribution of the power chip in the thyristor converter valve is shown in formula i Figure 2 ce = i p + i n and Δn = Δp, the electric field intensity E in the current equation can be eliminated to obtain the shown electron current and hole current equations considering different injection conditions;

[0073]

[0074] The present application proposes a fast auxiliary variable K, which can effectively accelerate the solving speed:

[0075]

[0076] Based on the continuity equation of the base region charge, the electron current and hole current at the discrete points are

[0077]

[0078] In the formula, 0≤i≤N P -1; N P is the number of discrete units; n i = Δp i +N B ; p i = Δp+n si 2 / N B ; the method provided by the present application has the advantages of high efficiency and accuracy when processing large-scale nonlinear solving.

[0079] Specifically, in the above embodiment, the plurality of reduced-order substructure models are distributed to the plurality of processors for parallel processing, including first determining the complexity of each reduced-order substructure model based on the grid subdivision fineness and the calculation complexity, then determining the calculation units of the plurality of processors, matching the calculation units of the processors with the complexity of the reduced-order substructure models, determining the processor corresponding to each reduced-order substructure model, and finally using the plurality of processors to respectively parallel process the plurality of reduced-order substructure models.

[0080] In the present embodiment, the calculation tasks of the plurality of reduced-order substructure models are distributed to the plurality of processors for parallel processing, the calculation unit size is distributed according to the complexity of each part of the model, independent finite element analysis is performed on each substructure, and the processors are parallel processed, which can effectively improve the calculation efficiency; each processor can independently parallel process one reduced-order substructure model, which can improve the calculation speed, thereby accelerating the analysis process, for complex valve segment structures, the processor allocation unit number can be increased in the place where the grid subdivision is fine, and the analysis can be simplified in the unimportant place.

[0081] ​Specifically, it can be expressed as:

[0082]

[0083] where G i represents the i-th processor is independently calculated; G = UF, G represents the global stiffness matrix, U is the unknown displacement vector, and F is the external force vector; G = G1+G2+...+G n The global stiffness matrix G is decomposed into the sum of the local stiffness matrices of the plurality of reduced substructure models.

[0084] Parallel processing by multiple processors makes matrix calculation faster, and each processor is responsible for an independent substructure, without waiting for the results of other processors. Parallel computing effectively reduces the overall calculation time in large-scale finite element analysis and greatly improves the efficiency of identifying weak links of the converter valve.

[0085] Specifically, in the above embodiment, the finite element model is constructed and the analysis result of each finite element model is calculated by using the isogeometric analysis method, including first constructing the finite element model of the reduced substructure model according to the transient thermal structure weak coupling problem, then analyzing the finite element model by using the isogeometric analysis method based on the initial design parameters and the initial constraint conditions to obtain the initial analysis result, finally optimizing the finite element model based on the initial analysis result, and re-estimating the optimized finite element model by using the reanalysis algorithm to obtain the analysis result corresponding to the finite element model.

[0086] In this embodiment, the isogeometric fast reanalysis technology is used to solve the fast calculation of the multi-field finite element model. The reanalysis algorithm uses the initial analysis information to quickly estimate the modified structure, and can obtain more accurate analysis results with less calculation cost. The initial analysis information is used to quickly estimate the modified structure, and the reanalysis algorithm block diagram is as follows: Figure 3As shown, compared with traditional finite element analysis, the reanalysis algorithm can obtain more accurate analysis results with less computational cost. CA refers to detailed computational analysis of the current design scheme to evaluate its performance. IFU refers to factors considering nonlinear effects. IC refers to initial conditions set at the beginning of analysis. MGR is a method for improving computational accuracy by gradually refining the grid. In the optimization process, a target function needs to be defined as a standard for measuring the pros and cons of the design scheme. A neural network model is usually used to construct the target function, which can capture the nonlinear relationship in the complex design space. Design variables refer to parameters that can be changed and affect the performance of the final product. Variables are parameters that are expected to be optimized. Constraint functions specify some restrictions that the design must meet, such as size restrictions and strength requirements. The global optimization toolbox is a collection of various optimization algorithms aimed at finding the optimal solution, including PSO, a heuristic search algorithm that simulates bird foraging behavior; EGO, an optimization method based on proxy models; DOE, a statistical method for planning experiments to maximize information acquisition; and information flow direction. Starting from design variables and constraint functions, a series of analysis and optimization steps are performed to finally form the optimized target function.

[0087] Figure 4 The flowchart of the isogeometric fast reanalysis algorithm is introduced, and a fast optimization model for transient thermal-structural weak coupling problems and the corresponding closed-loop optimization design framework are proposed. In practical engineering, the mathematical model of the structural optimization problem can be represented as:

[0088] findx={x1,x2,...,x k ,...,x n}

[0089] minf(x)=[f1(x),f2(x),…,f m (x)] T

[0090]

[0091] Wherein, x is a finite element sub-model block, f(x) is an optimization objective function, g i (x) is the i-th equality constraint, h i (x) is the i-th inequality constraint, x k,min and x k,max are the minimum and maximum values of the finite element sub-model block, respectively.

[0092] Specifically, in the above embodiment, the complexity of a plurality of finite element models is determined, and the analysis results of the finite element models are solved by using a multi-objective evolutionary algorithm to obtain a solution of each finite element model, including first determining the complexity of the finite element model by using a complexity dynamic adjustment method, wherein the finite element model includes a low complexity model and a high complexity model, then establishing a related objective function and a plurality of constraint conditions according to a multi-objective optimization problem, based on minimizing the related objective function, the analysis results of the finite element model are solved by using a multi-objective evolutionary algorithm to obtain a solution of each finite element model.

[0093] In this embodiment, the multi-objective evolutionary algorithm is combined with the variable complexity method, according to the characteristics of low computational cost of low complexity model and high precision of high complexity model, while ensuring the calculation precision of high complexity model, the calculation cost is reduced, and the optimization efficiency is improved. It is proposed that the combination of multi-objective evolutionary algorithm and variable complexity method can fully utilize the characteristics of low computational cost of low complexity model and high precision of high complexity model, while ensuring the calculation precision of high complexity model, by reducing the calculation cost, we can improve the solving efficiency. For quickly identifying the weak link of the converter valve, the minimum CMOP (Constrained Multi-Objective Optimization Problem, Constrained Multi-Objective Optimization Problem) is innovatively proposed, which can be expressed as follows:

[0094]

[0095] Wherein x∈S is a w-dimensional decision vector in the search space S, f(x) represents an r-dimensional objective function, s h (x) represents the h-th constraint, e is the number of inequality constraints, t h (x) represents the h-th equality constraint, (q-e) is the number of equality constraints.

[0096] Further, the solution of each finite element model is optimized by using a Pareto transfer strategy to obtain an optimal parameter set, including first obtaining the solution of the low complexity model, taking the solution of the low complexity model as the initial search population of the high complexity model, then determining the search population of the high complexity model during iteration, obtaining a local search function, updating the local search function based on the parameters of the high complexity model and the search population, and obtaining the optimal parameter set.

[0097] In the embodiment, the selected low complexity model solution is used as the initial search population of the high complexity model by using a PT (Pareto Transfer) strategy, so that the high complexity model can replace the low complexity model to perform local search, reduce the optimization times of the high complexity model, and realize fast solving of the weak link of the thyristor valve. The selected low complexity model solution can be used as the initial search population of the high complexity model by using the PT strategy to reduce the optimization times of the high complexity model, so that the discrimination efficiency can be improved, and the precision can be further improved.

[0098] wherein the low complexity model solution is X low ,

[0099]

[0100] wherein, is the initial search population of the high complexity model, and the low complexity model solution is migrated to the high complexity model by using the PT strategy. The parameters of the high complexity model are gradually adjusted to approximate the optimal solution, and it is assumed that the parameters of the high complexity model are θ high ;

[0101] Local search of the high complexity model:

[0102]

[0103] The high complexity model searches the population at the t+1th iteration, and the local search function is updated by using the parameters of the high complexity model and the current search population, so that the solving time is reduced.

[0104] X opt = Optimize(X high )

[0105] wherein X opt represents the optimal parameter set obtained by the complete solving process, and the calculation efficiency and the precision are effectively improved. By this step, the different complexity models are used to accelerate the discrimination of the weak link of the thyristor valve.

[0106] Finally, a complete data-state pre-discrimination nonlinear equation fast iterative solving method is constructed to quickly and accurately find the temperature abnormal point, so that the weak link of the thyristor valve is discriminated, Figure 5 is the cloud map of the junction temperature distribution of the key device of the thyristor valve, the fast iterative solving method is used to quickly discriminate the temperature abnormal point of the thyristor valve in the finite element simulation analysis, and it is proved by the test that the method provided in the application can accelerate the solving speed and improve the discrimination efficiency of the weak link of the thyristor valve.

[0107] As shown in the above, the principle and logic flow of the weak link solving method of the thyristor valve provided in the application is as shown in Figure 6 .

[0108] Further, as Figure 1 In a specific implementation of the method, the embodiment of the present application provides a device for solving weak links of a converter valve, which comprises Figure 7 As shown in the figure, the device comprises a model reduction module 201, a parallel processing module 202, an optimization iteration module 203, and a weak link solving module 204.

[0109] The model reduction module 201 is configured to identify a plurality of subsystems of a to-be-judged converter valve, establish a corresponding substructure model for each subsystem, and perform interpolation optimization on the plurality of substructure models respectively to obtain a plurality of reduced-order substructure models.

[0110] The parallel processing module 202 is configured to distribute the plurality of reduced-order substructure models to a plurality of processors for parallel processing, construct a finite element model, and calculate an analysis result of each finite element model by using an isogeometric analysis method.

[0111] The optimization iteration module 203 is configured to determine the complexity of the plurality of finite element models, solve the analysis result of the finite element model by using a multi-objective evolutionary algorithm to obtain a solution of each finite element model, and perform optimization iteration on the solution of each finite element model by using a Pareto shift strategy to obtain an optimal parameter set.

[0112] The weak link solving module 204 is configured to construct a nonlinear equation according to the operation data of the to-be-judged converter valve, solve the nonlinear equation by using the optimal parameter set, identify a temperature abnormal point of the to-be-judged converter valve to determine a weak link of the to-be-judged converter valve.

[0113] In a specific application scenario, the model reduction module 201 can be specifically configured to collect experimental data of a to-be-judged converter valve, pre-process the experimental data, determine a target interpolation method based on the data characteristics and data requirements of the experimental data, establish an interpolation model according to the target interpolation method, perform interpolation optimization on the substructure model by using the interpolation model to obtain an initial reduced-order substructure model, construct an auxiliary variable, and adjust the model parameters of the initial reduced-order substructure model based on the auxiliary variable to optimize the initial reduced-order substructure model and obtain a reduced-order substructure model.

[0114] In a specific application scenario, the model reduction module 201 can also be configured to obtain an electron current equation and a hole current equation of the to-be-judged converter valve under one-dimensional conditions, construct an auxiliary variable, obtain the electron current equation and the hole current equation under different injection conditions based on the power chip electric field distribution in the to-be-judged converter valve, and then solve the electron current and the hole current at the discrete points in the electron current equation and the hole current equation under one-dimensional conditions; and determine the reduced-order substructure model according to the electron current equation and the hole current equation under one-dimensional conditions and the electron current and the hole current at the discrete points.

[0115] In a specific application scenario, the parallel processing module 202 can be specifically configured to determine the complexity of each reduced-order substructure model based on the mesh subdivision fineness and the calculation complexity; determine the calculation units of the plurality of processors, match the calculation units of the processors with the complexity of the reduced-order substructure models, and determine the processors corresponding to each reduced-order substructure model; and perform parallel processing on the plurality of reduced-order substructure models by using the plurality of processors.

[0116] In a specific application scenario, the parallel processing module 202 can also be configured to construct a finite element model of the reduced-order substructure model according to a transient thermal structure weak coupling problem; perform analysis on the finite element model by using an isogeometric analysis method based on initial design parameters and initial constraint conditions to obtain an initial analysis result; optimize the finite element model based on the initial analysis result, and re-estimate the optimized finite element model by using a reanalysis algorithm to obtain an analysis result corresponding to the finite element model.

[0117] In a specific application scenario, the optimization iteration module 203 can be specifically configured to determine the complexity of the finite element model by using a complexity dynamic adjustment method, wherein the finite element model includes a low-complexity model and a high-complexity model; establish a related objective function and a plurality of constraint conditions according to a multi-objective optimization problem, solve the analysis result of the finite element model by using a multi-objective evolutionary algorithm based on minimizing the related objective function, and obtain a solution of each finite element model.

[0118] In a specific application scenario, the optimization iteration module 203 can also be configured to obtain the solution of the low-complexity model, take the solution of the low-complexity model as an initial search population of the high-complexity model, determine a search population during iteration of the high-complexity model, obtain a local search function, update the local search function based on the parameters of the high-complexity model and the search population, and obtain an optimal parameter set.

[0119] It should be noted that other corresponding descriptions of the various functional units involved in the device for solving weak links of a thyristor provided in this embodiment can be referred to the corresponding descriptions in the Figure 1 , which will not be described here in detail.

[0120] Based on the method as shown in Figure 1 , correspondingly, the present embodiment also provides a storage medium having a computer program stored thereon, and the program is executed by a processor to implement the above-mentioned method for solving weak links of a thyristor.

[0121] Based on such understanding, the technical scheme of the present application can be embodied in the form of a software product. The to-be-identified software product can be stored in a nonvolatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the various implementation scenarios of the converter valve weak link solving method of the present application.

[0122] Based on the method as shown in Figure 1 and the converter valve weak link solving device embodiment as shown in Figure 7 in order to achieve the above-mentioned purpose, as shown in Figure 8 the present embodiment also provides an entity device for solving the weak link of the converter valve. The device includes a communication bus, a processor, a memory, and a communication interface, and can also include an input / output interface and a display device. The communication between the various functional units can be completed through the bus. The memory stores a computer program, and the processor is used to execute the program stored in the memory and execute the converter valve weak link solving method in the above-mentioned embodiments.

[0123] Optionally, the entity device can also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface can include a display screen (Display), an input unit such as a keyboard (Keyboard), etc. The optional user interface can also include a USB interface, a card reader interface, etc. The network interface can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0124] Those skilled in the art can understand that the structure of the entity device for solving the weak link of the converter valve provided by the present embodiment does not constitute a limitation on the entity device, and can include more or fewer components, or combine certain components, or different component arrangements.

[0125] The storage medium can also include an operating system and a network communication module. The operating system is a program for managing the hardware and to-be-identified software resources of the above-mentioned entity device, supporting the running of information processing programs and other to-be-identified software and / or programs. The network communication module is used to realize the communication between the components in the storage medium, and the communication with other hardware and software in the information processing entity device.

[0126] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software and necessary general hardware platforms, or by hardware. Through the application of the technical solutions of the present application, first, a plurality of subsystems of the to-be-judged thyristor are identified, a substructure model corresponding to each subsystem is established, and interpolation optimization is performed on the plurality of substructure models respectively to obtain a plurality of reduced-order substructure models. Then, the plurality of reduced-order substructure models are distributed to a plurality of processors for parallel processing, a finite element model is constructed, and the analysis result of each finite element model is calculated by using the isogeometric analysis method. After that, the complexity of the plurality of finite element models is determined, and the analysis result of the finite element model is solved by using a multi-objective evolutionary algorithm to obtain the solution of each finite element model. The solution of each finite element model is optimized and iterated by using a Pareto transfer strategy to obtain an optimal parameter set. Finally, a nonlinear equation is constructed according to the operation data of the to-be-judged thyristor, the nonlinear equation is solved by using the optimal parameter set, and the temperature abnormal point of the to-be-judged thyristor is identified to determine the weak link of the to-be-judged thyristor. The above method establishes a substructure model based on the subsystem of the thyristor, and reduces the order of the model by interpolation optimization, which greatly reduces the computational complexity and makes the simulation more efficient. The plurality of reduced-order substructure models are processed in parallel by using a plurality of processors, which can significantly speed up the entire simulation process and improve the simulation efficiency. The finite element model is constructed and combined with the isogeometric analysis method, which can further improve the solving efficiency while maintaining high precision, and more accurately capture the complex physical field changes inside the thyristor. The multi-objective evolutionary algorithm is adopted, which is suitable for solving problems with multiple conflicting objectives. The Pareto transfer strategy is used to optimize and iterate the model solution, which can improve the accuracy of predicting the temperature abnormal point. The actual operation data is used to construct a nonlinear equation, and the parameter set after optimization is used for solving, which can more realistically reflect the working state of the thyristor. The above method improves the speed and accuracy of identifying the weak link of the thyristor, accurately locates the temperature abnormal area, and provides clear fault diagnosis information for maintenance personnel.

[0127] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred implementation scenario, and the modules or flows in the drawings are not necessarily essential for implementing the present application. Those skilled in the art can understand that the modules in the devices in the implementation scenario can be distributed in the devices in the implementation scenario according to the description of the implementation scenario, or can be correspondingly changed and located in one or more devices different from the implementation scenario. The modules of the above implementation scenario can be combined as one module, or can be further split into a plurality of sub-modules.

[0128] The above application numbers are only for description, and do not represent the advantages and disadvantages of the implementation scenario. The above disclosure is only a few specific implementation scenarios of the present application, but the present application is not limited thereto, and any changes that can be thought of by those skilled in the art should fall within the protection scope of the present application.

Claims

1. A method for solving weak links in converter valves, characterized in that: include: Identifying multiple subsystems of the converter valve to be identified, establishing a substructure model corresponding to each of the subsystems, and performing interpolation optimization on the multiple substructure models to obtain multiple reduced-order substructure models; Distributing the plurality of reduced-order substructure models to a plurality of processors for parallel processing, constructing a finite element model and calculating an analysis result of each finite element model using an isogeometric analysis method; Determining the complexity of the plurality of finite element models, solving the analysis results of the finite element models using a multi-objective evolutionary algorithm to obtain a solution for each of the finite element models, and iterating and optimizing the solution of each of the finite element models using a Pareto transfer strategy to obtain an optimal parameter set; Constructing a nonlinear equation based on the operating data of the converter valve to be identified, solving the nonlinear equation using the optimal parameter set, identifying temperature anomalies of the converter valve to be identified, and determining the weak links of the converter valve to be identified; The determining of the complexity of the plurality of finite element models, solving the analysis results of the finite element models using a multi-objective evolutionary algorithm to obtain a solution for each of the finite element models, and iterating and optimizing the solution of each of the finite element models using a Pareto transfer strategy to obtain an optimal parameter set includes: Determining the complexity of the finite element model using a complexity dynamic adjustment method, wherein the finite element model includes a low-complexity model and a high-complexity model; Establishing relevant objective functions and multiple constraints according to the multi-objective optimization problem, solving the analysis results of the finite element model using a multi-objective evolutionary algorithm based on minimizing the relevant objective functions, and obtaining a solution for each of the finite element models; Obtaining a solution to the low-complexity model, and using the solution to the low-complexity model as an initial search population for the high-complexity model; A search population during iteration of the high-complexity model is determined, a local search function is obtained, and the local search function is updated based on the parameters of the high-complexity model and the search population to obtain an optimal parameter set.

2. The method according to claim 1, characterized in that The interpolation optimization is performed on the plurality of substructure models respectively to obtain a plurality of reduced-order substructure models, including: Collecting experimental data of the converter valve to be identified and preprocessing the experimental data; determining a target interpolation method based on data characteristics and data requirements of the experimental data; An interpolation model is established according to the target interpolation method, and the substructure model is interpolated and optimized using the interpolation model to obtain an initial reduced-order substructure model; Auxiliary variables are constructed, and model parameters of the initial reduced-order substructure model are adjusted based on the auxiliary variables to optimize the initial reduced-order substructure model to obtain a reduced-order substructure model.

3. The method according to claim 2, characterized in that The constructing of auxiliary variables and adjusting the model parameters of the initial reduced-order substructure model based on the auxiliary variables to optimize the initial reduced-order substructure model to obtain the reduced-order substructure model includes: Obtaining the electron current equation and hole current equation of the converter valve to be identified under one-dimensional conditions; Construct auxiliary variables and obtain the electron current equation and hole current equation under different injection conditions based on the electric field distribution of the power chip in the converter valve to be identified. Then, solve the electron current and hole current equations under one-dimensional conditions at discrete points. The reduced-order substructure model is determined according to the electron current equation and the hole current equation under the one-dimensional condition and the electron current and the hole current at the discrete points.

4. The method according to claim 1, wherein The step of distributing the plurality of reduced-order substructure models to a plurality of processors for parallel processing comprises: Determining the complexity of each of the reduced-order substructure models based on the meshing level and computational complexity; determining computing units of a plurality of processors, matching the computing units of the processors with the complexity of the reduced-order substructure model, and determining a processor corresponding to each of the reduced-order substructure models; A plurality of the processors are used to process the plurality of reduced-order substructure models in parallel.

5. The method according to claim 1, wherein The constructing of the finite element model and calculating the analysis results of each finite element model using the isogeometric analysis method includes: Constructing a finite element model of the reduced-order substructure model according to a transient thermal-structural weak coupling problem; Based on the initial design parameters and initial constraints, the finite element model is analyzed using an isogeometric analysis method to obtain initial analysis results; The finite element model is optimized based on the initial analysis result, and the optimized finite element model is re-estimated using a reanalysis algorithm to obtain an analysis result corresponding to the finite element model.

6. A device for solving weak links in converter valves, characterized in that: The device comprises: A model reduction module is used to identify multiple subsystems of the converter valve to be identified, establish a substructure model corresponding to each of the subsystems, and perform interpolation optimization on the multiple substructure models to obtain multiple reduced-order substructure models; A parallel processing module, configured to distribute the plurality of reduced-order substructure models to a plurality of processors for parallel processing, construct a finite element model, and calculate analysis results of each finite element model using an isogeometric analysis method; an optimization iteration module, configured to determine the complexity of the plurality of finite element models, solve the analysis results of the finite element models using a multi-objective evolutionary algorithm to obtain a solution for each of the finite element models, and iterate and optimize the solution of each finite element model using a Pareto transfer strategy to obtain an optimal parameter set; a weak link solving module, configured to construct a nonlinear equation based on the operating data of the converter valve to be identified, solve the nonlinear equation using the optimal parameter set, identify the temperature abnormality point of the converter valve to be identified, and determine the weak link of the converter valve to be identified; The parallel processing module is further configured to determine the complexity of the finite element model using a complexity dynamic adjustment method, wherein the finite element model includes a low complexity model and a high complexity model; Establishing relevant objective functions and multiple constraints according to the multi-objective optimization problem, solving the analysis results of the finite element model using a multi-objective evolutionary algorithm based on minimizing the relevant objective functions, and obtaining a solution for each of the finite element models; Obtaining a solution to the low-complexity model, and using the solution to the low-complexity model as an initial search population for the high-complexity model; A search population during iteration of the high-complexity model is determined, a local search function is obtained, and the local search function is updated based on the parameters of the high-complexity model and the search population to obtain an optimal parameter set.

7. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images