A method and apparatus for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media

Abstract

This invention provides a method and apparatus for optimizing the breaking performance of GIS high-voltage switches based on a mixed gas medium, relating to the field of high-voltage electrical equipment technology. The apparatus includes: a simulation optimization platform, a parameterized management module, a proxy model engine, an intelligent optimization solver, a scheme database, and an experimental data interface. By integrating and co-optimizing the mixed gas formulation, switch operating parameters, and key structural components, it breaks through the limitations of traditional isolated optimization methods, finding the globally optimal solution at the system level. Through a digital twin + proxy model approach, it transforms numerous time-consuming physical experiments into efficient computer simulation and optimization, significantly shortening the R&D cycle and reducing R&D costs. The optimized digital twin model can be used for product condition monitoring, fault diagnosis, and predictive maintenance, supporting the full lifecycle management of GIS equipment.

Description

Technical Field

[0001] This invention relates to the field of high-voltage electrical equipment technology, specifically to a method and apparatus for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media. Background Technology

[0002] Gas-insulated metal-enclosed switchgear (GIS) is a core protection and control device in modern power systems, and its reliability is directly related to the safe and stable operation of the power grid. Currently, GIS equipment generally uses sulfur hexafluoride (SF6) gas as the insulation and arc-extinguishing medium, mainly due to the excellent insulation strength and superior arc-extinguishing performance of SF6 gas. However, SF6 is an extremely potent greenhouse gas, with a global warming potential 23,500 times that of carbon dioxide, and an atmospheric lifetime of up to 3,200 years. With the deepening implementation of international environmental regulations such as the Kyoto Protocol and the Paris Agreement, reducing and ultimately replacing SF6 has become an urgent and irreversible technological trend in the field of power equipment manufacturing.

[0003] Currently, the research and development of mixed gas GIS switches mainly relies on two traditional models: one is the experience-based trial-and-error method, which involves repeatedly preparing prototypes and conducting costly synthetic circuit experiments to screen parameters. This method has a long development cycle, high cost, and cannot exhaust the huge design space. The other is simulation-aided design based on a single physical field, which optimizes only the structure or airflow locally and fails to comprehensively consider the full coupling effect of arc plasma, electromagnetic field, compressible airflow field, and material properties at the system level. More importantly, existing methods lack a systematic framework that can integrate environmental indicators with electrical breaking performance for intelligent and coordinated optimization. Therefore, those skilled in the art provide a method and device for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media to solve the problems mentioned in the background. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a method and apparatus for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media, solving the problem that existing methods lack an integrated and intelligent collaborative optimization mechanism that can combine environmental indicators with electrical breaking performance.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention provides the following technical solution: a device for optimizing the breaking performance of GIS high-voltage switches based on a mixed gas medium, comprising:

[0008] Simulation optimization platform, a high-performance computing workstation or server integrating CAD modeling software, multiphysics coupling simulation software and data analysis modules;

[0009] The parameterized management module is used to define and manage optimizable variables and provides a human-computer interaction interface for users to set optimization goals and constraints.

[0010] The proxy model engine has a built-in library of various machine learning algorithms for automatically training, evaluating, and calling proxy models. It can complete the process semi-automatically or fully automatically, from receiving simulation data, feature engineering, model selection and training, hyperparameter tuning to model performance evaluation.

[0011] The intelligent optimization solver has a built-in multi-objective optimization algorithm to drive the proxy model engine to perform optimization searches;

[0012] The solution database is used to store all simulation data, surrogate models, optimization history, and final optimization solutions;

[0013] The test data interface is used to receive test data from the physical test bench and feed the data back to the database for model correction and verification.

[0014] Preferably, the simulation optimization platform provides a graphical interface or script interface, allowing users to import or create switch 3D models, and to parametrically mark and associate key dimensions and shapes, automatically generating high-quality meshes suitable for simulation software, mesh generation strategies, and presets.

[0015] Preferably, the intelligent optimization solver includes an algorithm library and an optimization process controller. The algorithm library has built-in NSGA-II, MOEA / D and particle swarm optimization multi-objective optimization algorithms. The optimization process controller configures the optimization objective, constraints, variable range and algorithm parameters, and drives the entire optimization loop, calling the surrogate model for evaluation, executing the evolutionary algorithm and generating a new population.

[0016] An optimization method for the breaking performance of GIS high-voltage switches based on mixed gas media includes the following steps:

[0017] S1. Construct a digital twin model, establish a refined three-dimensional model of the target GIS switch, and integrate a multi-physics field coupling simulation module based on the physical parameters of the mixed gas. The multi-physics field includes, but is not limited to, electromagnetic field, flow field, temperature field and arc plasma dynamics model, forming a "digital twin" of the switch.

[0018] S2. Set optimization objectives and constraints, define the objective function for breaking performance optimization, the objective function includes key performance indicators, including but not limited to minimum arc duration, maximum dielectric recovery rate, minimum reignition probability, minimum operating overvoltage, and environmental indicators; at the same time, set constraints, including gas ratio range, total gas pressure range, mechanical operating parameter range, and switch size limit;

[0019] S3. Parametric design and data acquisition: Define the optimizable variables parametrically, including: the type and component ratio of the mixed gas, the charging pressure, the geometric parameters of the nozzle structure, the volume of the compressor cylinder and the characteristics of the operating piston, and the opening speed curve. Using the digital twin model in step S1, simulate the opening process for different combinations of variables to form a database containing input variables and output performance indicators.

[0020] S4. Proxy model training: Based on the database, a high-performance proxy model is trained using machine learning algorithms. The proxy model can quickly predict the breaking performance index under any given combination of variables at a speed much higher than that of multi-physics simulation.

[0021] S5. Multi-objective collaborative optimization: Using intelligent optimization algorithms and surrogate models as evaluators, multi-objective optimization search is performed on the objective function under constraints to obtain one or more sets of Pareto optimal solutions. Each solution represents an optimization scheme that achieves the best balance between breaking performance and environmental friendliness.

[0022] S6. Scheme verification and physical testing: Select recommended schemes from the Pareto optimal solution set, conduct high-fidelity simulation verification through digital twin models, and guide the manufacturing and testing of physical prototypes. Use test data feedback to correct the digital twin model and surrogate model.

[0023] Preferably, the construction of the digital twin model is based on the CAD drawings of the target GIS switch, using parametric modeling tools, including but not limited to Python-based script-driven models, Solidworks, or ANSYS Spaceclaim, to create a 3D model. The model needs to be detailed down to the arc-extinguishing chamber, nozzle, compressor cylinder, piston, moving / stationary contacts, and grid plates, and ensure that all dimensions to be optimized are parametrically defined. The temperature field and arc plasma dynamics model needs to establish a local thermodynamic equilibrium or non-equilibrium arc model of the mixed gas. The input to this model is a database of physical property parameters of a mixed gas in a specific proportion, including but not limited to the relationship between electrical conductivity, enthalpy, thermal conductivity, and radiation coefficient as a function of temperature and pressure. By solving the mass, momentum, and energy conservation equations and Maxwell's equations, the ignition, contraction, movement, cooling, and extinction processes of the arc are dynamically simulated.

[0024] Preferably, the multi-objective collaborative optimization process includes the following steps:

[0025] S1. Initialization: Randomly generate an initial population in the variable space;

[0026] S2. Evaluation: Using the trained agent model, quickly calculate the objective function value for each individual in the population;

[0027] S3. Sorting and Selection: The population is sorted according to the Pareto dominance relationship. Individuals that are not superior to any other individual in all objectives are formed into the first Pareto front. The algorithm prioritizes individuals with higher front rankings.

[0028] S4. Evolution: Crossover and mutation of selected individuals to produce offspring populations;

[0029] S5. Iteration: Repeat the evaluation to the evolution step until the preset number of iterations or convergence criteria are reached; finally, a set of Pareto optimal solutions is obtained, which represent the set of best compromise solutions within the existing design space that cannot further improve any objective without harming other objectives.

[0030] (III) Beneficial Effects

[0031] This invention provides a method and apparatus for optimizing the breaking performance of GIS high-voltage switches based on a mixed gas medium. It has the following beneficial effects:

[0032] 1. In this invention, by integrating and co-optimizing the mixed gas formulation, switching operation parameters and key structural components, the limitations of traditional isolated optimization are broken. The global optimal solution can be found at the system level. Through the technical path of digital twin + proxy model, a large number of time-consuming physical experiments are transformed into efficient computer simulation and optimization, which significantly shortens the R&D cycle and reduces R&D costs.

[0033] 2. In this invention, by directly incorporating environmental indicators such as GWP into the optimization objective function, the optimization scheme is ensured to meet the most stringent environmental regulations while improving or maintaining the switching performance.

[0034] 3. The methods and devices used in this invention can be customized and optimized for different voltage levels, including but not limited to 72.5kV, 145kV, 252kV and 550kV, different types of switches and different application scenarios, and have strong versatility.

[0035] 4. In this invention, the optimized digital twin model can be used for product status monitoring, fault diagnosis and predictive maintenance, supporting the full lifecycle management of GIS equipment. Attached Figure Description

[0036] Figure 1 This is a schematic diagram illustrating the specific structure of the optimization device in this invention;

[0037] Figure 2 This is a schematic diagram illustrating the specific process of the optimization method in this invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1:

[0040] like Figure 1-2 As shown, this embodiment of the invention provides a device for optimizing the breaking performance of GIS high-voltage switches based on a mixed gas medium, comprising:

[0041] Simulation optimization platform, a high-performance computing workstation or server integrating CAD modeling software, multiphysics coupling simulation software and data analysis modules;

[0042] The parameterized management module is used to define and manage optimizable variables and provides a human-computer interaction interface for users to set optimization goals and constraints.

[0043] The proxy model engine has a built-in library of various machine learning algorithms for automatically training, evaluating, and calling proxy models. It can complete the process semi-automatically or fully automatically, from receiving simulation data, feature engineering, model selection and training, hyperparameter tuning to model performance evaluation.

[0044] The intelligent optimization solver has a built-in multi-objective optimization algorithm to drive the proxy model engine to perform optimization searches;

[0045] The solution database is used to store all simulation data, surrogate models, optimization history, and final optimization solutions;

[0046] The test data interface is used to receive test data from the physical test bench and feed the data back to the database for model correction and verification.

[0047] The simulation optimization platform provides a graphical interface or script interface, allowing users to import or create switch 3D models, and parametrically mark and associate key dimensions and shapes, automatically generating high-quality meshes suitable for simulation software, mesh generation strategies, and presets.

[0048] The intelligent optimization solver includes an algorithm library and an optimization process controller. The algorithm library has built-in NSGA-II, MOEA / D and particle swarm optimization multi-objective optimization algorithms. The optimization process controller configures the optimization objective, constraints, variable range and algorithm parameters, and drives the entire optimization loop, calling the surrogate model for evaluation, executing the evolutionary algorithm and generating a new population.

[0049] An optimization method for the breaking performance of GIS high-voltage switches based on mixed gas media includes the following steps:

[0050] S1. Construct a digital twin model, establish a refined three-dimensional model of the target GIS switch, and integrate a multi-physics field coupling simulation module based on the physical parameters of the mixed gas. The multi-physics field includes, but is not limited to, electromagnetic field, flow field, temperature field and arc plasma dynamics model, forming a "digital twin" of the switch.

[0051] S2. Set optimization objectives and constraints, define the objective function for breaking performance optimization, and the objective function includes key performance indicators, including but not limited to minimum arc duration, maximum dielectric recovery rate, minimum reignition probability, minimum operating overvoltage, and environmental indicators; at the same time, set constraints, including gas ratio range, total gas pressure range, mechanical operating parameter range, and switch size limit;

[0052] S3. Parametric design and data acquisition: Define the optimizable variables parametrically, including: the type and component ratio of the mixed gas, the charging pressure, the geometric parameters of the nozzle structure, the volume of the compressor cylinder and the characteristics of the operating piston, and the opening speed curve. Using the digital twin model in step S1, simulate the opening process for different combinations of variables to form a database containing input variables and output performance indicators.

[0053] S4. Proxy model training: Based on the database, a high-performance proxy model is trained using machine learning algorithms. The proxy model can quickly predict the breaking performance index under any given combination of variables at a speed much higher than that of multi-physics simulation.

[0054] S5. Multi-objective collaborative optimization: Using intelligent optimization algorithms and surrogate models as evaluators, multi-objective optimization search is performed on the objective function under constraints to obtain one or more sets of Pareto optimal solutions. Each solution represents an optimization scheme that achieves the best balance between breaking performance and environmental friendliness.

[0055] S6. Scheme verification and physical testing: Select recommended schemes from the Pareto optimal solution set, conduct high-fidelity simulation verification through digital twin models, and guide the manufacturing and testing of physical prototypes. Use test data feedback to correct the digital twin model and surrogate model.

[0056] The digital twin model is constructed based on the CAD drawings of the target GIS switch. Parametric modeling tools are used, including but not limited to Python-based script-driven models, Solidworks, or ANSYS Spaceclaim, to create a 3D model. The model needs to be detailed down to the arc-extinguishing chamber, nozzle, compressor cylinder, piston, moving / stationary contacts, and grid plates, and ensure that all dimensions to be optimized are parametrically defined. The temperature field and arc plasma dynamics model need to establish a local thermodynamic equilibrium or non-equilibrium arc model of the mixed gas. The input to this model is a database of physical property parameters of a mixed gas with a specific ratio, including but not limited to the relationship between electrical conductivity, enthalpy, thermal conductivity, and radiation coefficient as a function of temperature and pressure. By solving the mass, momentum, and energy conservation equations and Maxwell's equations, the ignition, contraction, movement, cooling, and extinction processes of the arc are dynamically simulated.

[0057] The operational process of multi-objective collaborative optimization includes the following steps:

[0058] S1. Initialization: Randomly generate an initial population in the variable space;

[0059] S2. Evaluation: Using the trained agent model, quickly calculate the objective function value for each individual in the population;

[0060] S3. Sorting and Selection: The population is sorted according to the Pareto dominance relationship. Individuals that are not superior to any other individual in all objectives are formed into the first Pareto front. The algorithm prioritizes individuals with higher front rankings.

[0061] S4. Evolution: Crossover and mutation of selected individuals to produce offspring populations;

[0062] S5. Iteration: Repeat the evaluation to the evolution step until the preset number of iterations or convergence criteria are reached; finally, a set of Pareto optimal solutions is obtained, which represent the set of best compromise solutions within the existing design space that cannot further improve any objective without harming other objectives.

[0063] Example 2: An optimization was performed using a 145kV GIS circuit breaker containing a mixture of sulfur hexafluoride and nitrogen.

[0064] Step S1: Construct a digital twin model; use 3D modeling software to establish an accurate geometric model of the circuit breaker, focusing on refining key components such as the arc-extinguishing chamber, nozzle, air cylinder, and moving contact; import the model into multiphysics simulation software, set material properties and boundary conditions; introduce a temperature-dependent physical property parameter library of SF6 / N2 mixed gas, establish a transient simulation model coupled with electromagnetic-fluid-thermal-arc, and simulate the short-circuit current breaking process;

[0065] Step S2: Set optimization objectives and constraints; The optimization objectives are: minimize arc energy when interrupting a 40kA short-circuit current; maximize the dielectric recovery strength within 10ms after interruption; minimize the GWP value of the mixed gas; The constraints are: SF6 volume fraction between 10% and 30%, total gas pressure between 0.5 and 0.7MPa, nozzle throat diameter variation range ±15%, and average opening speed of 8-12m / s;

[0066] Step S3: Parametric design and data acquisition; parameterize the nozzle profile using B-spline curve control points, and parameterize the cylinder volume and brake opening speed curves piecewise linearly; generate 500 different variable combination samples using experimental design methods; perform batch simulations using the twin model from step S101, collect the arc voltage and current waveforms, air pressure changes, electric field distribution, etc. corresponding to each sample, calculate the objective function value, and form a database;

[0067] Step S4: Proxy model training; 80% of the data is extracted from the database as the training set, and the Proxy model is trained using the Gaussian process regression algorithm; the remaining 20% ​​of the data is used for validation, and the prediction determination coefficients R² of the model for the three objective functions are all greater than 0.95, which meets the accuracy requirements;

[0068] Step S5: Multi-objective collaborative optimization; a non-dominated sorting genetic algorithm with an elitist strategy is used as an intelligent optimization solver; the trained surrogate model is used as a fast evaluator to run the optimization algorithm; after thousands of iterations, the Pareto optimal solution set is obtained;

[0069] Step S6: Scheme Verification and Testing; Engineers select a "balance point" scheme from the Pareto front, for example: 18% SF6, gas pressure 0.62MPa, specifically optimized nozzle profile, and optimized tripping speed curve; High-fidelity simulation is performed on this scheme, and the results verify that its breaking performance is better than the original pure SF6 design, and the GWP value is reduced by about 80%. Based on this, a prototype is manufactured, and a synthetic test is conducted at the high-voltage switch test station. The test results are in good agreement with the simulation predictions, verifying the effectiveness of this method.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A device for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media, characterized in that: include: Simulation optimization platform, a high-performance computing workstation or server integrating CAD modeling software, multiphysics coupling simulation software and data analysis modules; The parameterized management module is used to define and manage optimizable variables and provides a human-computer interaction interface for users to set optimization goals and constraints. The proxy model engine has a built-in library of various machine learning algorithms for automatically training, evaluating, and calling proxy models. It can complete the process semi-automatically or fully automatically, from receiving simulation data, feature engineering, model selection and training, hyperparameter tuning to model performance evaluation. The intelligent optimization solver has a built-in multi-objective optimization algorithm to drive the proxy model engine to perform optimization searches; The solution database is used to store all simulation data, surrogate models, optimization history, and final optimization solutions; The test data interface is used to receive test data from the physical test bench and feed the data back to the database for model correction and verification.

2. The GIS high-voltage switch breaking performance optimization device based on mixed gas medium according to claim 1, characterized in that: The simulation optimization platform provides a graphical interface or script interface, allowing users to import or create switch 3D models, and to parametrically mark and associate key dimensions and shapes, automatically generating high-quality meshes suitable for simulation software, mesh generation strategies, and presets.

3. The GIS high-voltage switchgear breaking performance optimization device based on mixed gas medium according to claim 1, characterized in that: The intelligent optimization solver includes an algorithm library and an optimization process controller. The algorithm library has built-in NSGA-II, MOEA / D and particle swarm optimization multi-objective optimization algorithms. The optimization process controller configures the optimization objective, constraints, variable range and algorithm parameters, and drives the entire optimization loop, calling the surrogate model for evaluation, executing the evolutionary algorithm and generating a new population.

4. A method for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media, characterized in that: Includes the following steps: S1. Construct a digital twin model, establish a refined three-dimensional model of the target GIS switch, and integrate a multi-physics field coupling simulation module based on the physical parameters of the mixed gas. The multi-physics field includes, but is not limited to, electromagnetic field, flow field, temperature field and arc plasma dynamics model, forming a "digital twin" of the switch. S2. Set optimization objectives and constraints, define the objective function for breaking performance optimization, the objective function includes key performance indicators, including but not limited to minimum arc duration, maximum dielectric recovery rate, minimum reignition probability, minimum operating overvoltage, and environmental indicators; at the same time, set constraints, including gas ratio range, total gas pressure range, mechanical operating parameter range, and switch size limit; S3. Parametric design and data acquisition: Define the optimizable variables parametrically, including: the type and component ratio of the mixed gas, the charging pressure, the geometric parameters of the nozzle structure, the volume of the compressor cylinder and the characteristics of the operating piston, and the opening speed curve. Using the digital twin model in step S1, simulate the opening process for different combinations of variables to form a database containing input variables and output performance indicators. S4. Proxy model training: Based on the database, a high-performance proxy model is trained using machine learning algorithms. The proxy model can quickly predict the breaking performance index under any given combination of variables at a speed much higher than that of multi-physics simulation. S5. Multi-objective collaborative optimization: Using intelligent optimization algorithms and surrogate models as evaluators, multi-objective optimization search is performed on the objective function under constraints to obtain one or more sets of Pareto optimal solutions. Each solution represents an optimization scheme that achieves the best balance between breaking performance and environmental friendliness. S6. Scheme verification and physical testing: Select recommended schemes from the Pareto optimal solution set, conduct high-fidelity simulation verification through digital twin models, and guide the manufacturing and testing of physical prototypes. Use test data feedback to correct the digital twin model and surrogate model.

5. The method for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media according to claim 4, characterized in that: The construction of the digital twin model is based on the CAD drawings of the target GIS switch. Parametric modeling tools are used, including but not limited to Python-based script-driven models, Solidworks, or ANSYS Spaceclaim, to create a 3D model. The model needs to be detailed down to the arc-extinguishing chamber, nozzle, compressor cylinder, piston, moving / stationary contacts, and grid plates, and ensure that all dimensions to be optimized are parameterized. The temperature field and arc plasma dynamics model needs to establish a local thermodynamic equilibrium or non-equilibrium arc model of the mixed gas. The input to this model is a database of physical property parameters of a mixed gas with a specific ratio, including but not limited to the relationship between electrical conductivity, enthalpy, thermal conductivity, and radiation coefficient as a function of temperature and pressure. By solving the mass, momentum, and energy conservation equations and Maxwell's equations, the ignition, contraction, movement, cooling, and extinction processes of the arc are dynamically simulated.

6. The method for optimizing the breaking performance of GIS high-voltage switches based on mixed gas media according to claim 4, characterized in that: The multi-objective collaborative optimization process includes the following steps: S1. Initialization: Randomly generate an initial population in the variable space; S2. Evaluation: Using the trained agent model, quickly calculate the objective function value for each individual in the population; S3. Sorting and Selection: The population is sorted according to the Pareto dominance relationship. Individuals that are not superior to any other individual in all objectives are formed into the first Pareto front. The algorithm prioritizes individuals with higher front rankings. S4. Evolution: Crossover and mutation of selected individuals to produce offspring populations; S5. Iteration: Repeat the evaluation to the evolution step until the preset number of iterations or convergence criteria are reached; finally, a set of Pareto optimal solutions is obtained, which represent the set of best compromise solutions within the existing design space that cannot further improve any objective without harming other objectives.

Images