WO2024016623A1 - Ssa-svm-based gis fault mode recognition method - Google Patents

Ssa-svm-based gis fault mode recognition method Download PDF

Info

Publication number
WO2024016623A1
WO2024016623A1 PCT/CN2023/073989 CN2023073989W WO2024016623A1 WO 2024016623 A1 WO2024016623 A1 WO 2024016623A1 CN 2023073989 W CN2023073989 W CN 2023073989W WO 2024016623 A1 WO2024016623 A1 WO 2024016623A1
Authority
WO
WIPO (PCT)
Prior art keywords
svm
ssa
gis
gas
support vector
Prior art date
Application number
PCT/CN2023/073989
Other languages
French (fr)
Chinese (zh)
Inventor
张英
王为
黄杰
姚望
徐龙舞
王明伟
余鹏程
刘喆
冯楚杰
赵世钦
潘云
Original Assignee
贵州电网有限责任公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 贵州电网有限责任公司 filed Critical 贵州电网有限责任公司
Publication of WO2024016623A1 publication Critical patent/WO2024016623A1/en

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02B—BOARDS, SUBSTATIONS OR SWITCHING ARRANGEMENTS FOR THE SUPPLY OR DISTRIBUTION OF ELECTRIC POWER
    • H02B13/00—Arrangement of switchgear in which switches are enclosed in, or structurally associated with, a casing, e.g. cubicle
    • H02B13/02—Arrangement of switchgear in which switches are enclosed in, or structurally associated with, a casing, e.g. cubicle with metal casing
    • H02B13/035—Gas-insulated switchgear
    • H02B13/065—Means for detecting or reacting to mechanical or electrical defects

Definitions

  • the present invention relates to the technical field of gas insulated substations, and in particular to a GIS fault mode identification method based on SSA-SVM.
  • GIS Gas-insulated substations
  • These defects mainly include metal particles left in the air chamber caused by poor manufacturing quality, metal burrs or powder caused by mechanical friction or vibration during operation, and tiny air gaps caused by fractures of conductors and supporting insulators.
  • Such defects can distort the inherent electric field of the GIS, causing partial discharges in some cases and, in severe cases, sparks or arcing. These faults have a greater impact and serious consequences. If the above defects in GIS are not solved in time, insulation faults will occur during the operation of the equipment. The safe and reliable operation of GIS and the entire power system itself will be threatened to varying degrees. .
  • GIS fault types can be summarized as the following fault types, including high-voltage conductor protrusions (N category), free conductive particles (P category), insulator metal contamination (M category) or insulator outer air gap insulation defects (GE category) )wait.
  • the technical problem solved by the present invention is that the existing fault identification technology has the problems of low fault diagnosis rate and poor stability.
  • a GIS fault mode identification method based on SSA-SVM including:
  • a support vector machine fault diagnosis model optimized by the Sparrow search algorithm is established, and the data is input into the diagnosis model to obtain the fault results.
  • the real model of the gas insulated substation is a closed container, and the diffusion of characteristic gas component molecules can only be achieved through molecular thermodynamic movement. deal with.
  • the decomposition of SF 6 gas includes:
  • the initial dissociation process of SF 6 gas can be roughly divided into three stages: (1) Inside the GIS cavity, SF 6 molecules are subject to electron collisions to form metastable molecular groups (SF 6 )* with poor stability. In a very short period of time, it will be converted into negative ions such as SF 6 - or SF 5 - ; (2) The electric field strength and gas molecular density will affect the form of decomposition products of SF 6 gas. When the ratio of electric field strength to gas molecular density is low, SF 6 - will react with SF 6 to generate SF 5 - .
  • SF 6 - or SF 5 - will react with SF 6 to generate SF 5 or SF 4 , etc.
  • SF 5 - will react with SF 6 to produce a variety of low-fluorine sulfides, including: SF 4 , SF 3 , SF 2 , SF, etc.; (3)
  • Various low-fluorine sulfides formed during the dissociation process of SF 6 gas will also react with each other to produce a more stable product S 2 F 10 , but when the ambient temperature is higher than 200°C, S 2 F 10 will It is completely decomposed, so it is easier to generate this product under discharge conditions with lower energy such as corona discharge, spark discharge, and partial discharge.
  • H 2 O and O 2 will dissociate due to factors such as electron collision to produce highly active particles such as O and OH.
  • Low fluorine sulfide will combine with OH and O to generate HF, SOF 4 and SO 2 Products such as F 2 , S 2 OF 10 , SOF 2 and SO 2 .
  • SOF 2 is mainly produced by the reaction of SF 4 and H 2 O or OH
  • SO 2 F 2 is mainly produced by the reaction of SF 2 and O 2
  • SO 2 is mainly produced by the reaction of SOF 2 and H 2 O
  • S 2 OF 10 is mainly produced by the reaction of SF 5 is produced by the reaction with O 2
  • SOF 4 is mainly produced by the reaction of SF 5 with O or OH. Since SF 5 O has poor stability, most of it will decompose into SOF 4 and a small part will react with SF 5 to produce S 2 OF 10 .
  • the effective gas production rate R RMS refers to the average volume fraction of a certain gas component per hour, and the gas production rate The rate equation is expressed as:
  • R RMS is the absolute gas production rate
  • c i,1 is the content of component i measured for the first time
  • c i,2 is the content of component i measured for the second time
  • ⁇ t is the time interval between two measurements .
  • the real model is imported into fluent for internal fluid simulation to obtain original data, including:
  • characteristic gases such as SO 2 F 2 , SOF 2 , SO 2 , CO 2 and CF 4
  • the Sparrow optimization algorithm includes:
  • n is the number of sparrows
  • d is the dimension of the variable to be optimized
  • the fitness value of all sparrows is expressed as:
  • the discoverer’s individual position update formula is:
  • the joiner’s location update formula is:
  • the individual fitness value of this iteration is compared with the current best fitness value.
  • the location is updated.
  • the formula is expressed as:
  • f g is the current global best fitness value
  • f w is the current global worst fitness value
  • ⁇ is the smallest constant.
  • Support Vector Machine is a machine learning algorithm based on statistics. It has the advantage of small training volume. For small clusters with few samples, it can achieve short training time and good classification effect.
  • the core concept is to map the input space to a high-dimensional space through a kernel function, where linear classification can be done using methods. By constructing an optimal classification hyperplane in high-dimensional space and performing classification, the generalization ability and reliability of interval classification between data categories are maximized.
  • the hyperplane function is expressed as:
  • mapping function is the mapping function
  • ⁇ is the normal vector of the optimal classification hyperplane
  • b is the classification threshold
  • the hyperplane function is expressed as:
  • C is the penalty parameter.
  • the hyperplane optimization problem is reduced to a quadratic planning problem, expressed as:
  • the optimal hyperplane function is expressed as:
  • SV is the support vector
  • the kernel function is expressed as:
  • a support vector machine fault diagnosis model optimized by the Sparrow search algorithm including:
  • Preprocess the original data and import the data into the algorithm select the training set and test set, and import the data according to the research situation; initialize the parameters of the sparrow search algorithm and support vector machine; update the position of the sparrow; calculate the position of each population of sparrows to obtain The current new position; calculate the fitness value and save the best value; save the corresponding support vector machine parameter combination according to the best fitness value; import the best parameter combination and use the support vector machine for calculation, determine the termination condition, and obtain the GIS failure mode code.
  • the GIS fault mode identification method based on SSA-SVM obtained by the present invention obtains original data of characteristic gas components under different fault defects by establishing a real GIS model, and imports the data into SVM faults optimized based on the SSA algorithm.
  • the diagnostic model can accurately obtain the fault mode of GIS. Compared with traditional methods, it has the advantages of faster calculation and higher recognition rate.
  • Figure 1 is an SSA-SVM flow chart of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 2 is a real GIS model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 3 is a SF 6 gas decomposition path diagram of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 4 is an actual and predicted classification diagram of the SSA-SVM diagnostic model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 5 is a function fitness diagram of each optimization algorithm of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 6 is a comparison algorithm WOA-SVM diagnostic model actual and predicted classification diagram of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention
  • Figure 7 is a comparison diagram of the actual and predicted classification of the PSO-SVM diagnostic model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention.
  • references herein to "one embodiment” or “an embodiment” refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. "In one embodiment” appearing in different places in this specification does not all refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.
  • connection should be understood in a broad sense.
  • it can be a fixed connection, a detachable connection, or an integrated connection; it can also be a mechanical connection, an electrical connection, or a direct connection.
  • a connection can also be indirectly connected through an intermediary, or it can be an internal connection between two components.
  • the specific meanings of the above terms in the present invention can be understood on a case-by-case basis.
  • an embodiment of the present invention provides a GIS fault mode identification method based on SSA-SVM, including:
  • this embodiment uses the size of GIS under 110KV to build a real model in ANSYS software.
  • the real GIS model is shown in Figure 2, with a length of 9538.31mm, a widest diameter of 2188.3mm, and a narrowest diameter of 499.74mm.
  • the gas monitoring point is located in the middle of the upper end of the equipment. Since the device is a closed container, the diffusion of characteristic gas component molecules can only be handled through molecular thermodynamic motion.
  • the initial dissociation process of SF 6 gas can be roughly divided into three stages: (1) Inside the GIS cavity, SF 6 molecules are subject to electron collisions to form metastable molecular groups (SF 6 )*, will be converted into negative ions such as SF 6 - or SF 5 - in a very short period of time; (2) The electric field strength and gas molecular density will affect the form of decomposition products of SF 6 gas. When the electric field strength and gas molecular density are When the ratio is low, SF 6 - will react with SF 6 to generate SF 5 - .
  • SF 6 - or SF 5 - will react with SF 6 to generate SF 5 Or low-fluorine sulfides such as SF 4.
  • SF 5 - will react with SF 6 to produce a variety of low-fluorine sulfides, including: SF 4 , SF 3 , SF 2 , SF, etc.; (3)
  • Various low-fluorine sulfides formed during the dissociation process of SF 6 gas will also react with each other to produce a more stable product S 2 F 10 , but when the ambient temperature is higher than 200°C, S 2 F 10 will decompose completely, so it is easier to generate this product under lower energy discharge conditions such as corona discharge, spark discharge, and partial discharge.
  • H 2 O and O 2 will dissociate due to factors such as electron collision to produce highly active particles such as O and OH.
  • Low fluorine sulfide will combine with OH and O to generate HF and SOF. 4.
  • SOF 2 is mainly produced by the reaction of SF 4 and H 2 O or OH
  • SO 2 F 2 is mainly produced by the reaction of SF 2 and O 2
  • SO 2 is mainly produced by the reaction of SOF 2 and H 2 O
  • S 2 OF 10 is mainly produced by the reaction of SF 5 is produced by the reaction with O 2
  • SOF 4 is mainly produced by the reaction of SF 5 with O or OH. Since SF 5 O has poor stability, most of it will decompose into SOF 4 and a small part will react with SF 5 to produce S 2 OF 10 .
  • the effective gas production rate R RMS refers to the average volume fraction of a certain gas component per hour.
  • the gas production rate equation is expressed as:
  • R RMS is the absolute gas production rate
  • c i,1 is the content of component i measured for the first time
  • c i,2 is the content of component i measured for the second time
  • ⁇ t is the time interval between two measurements .
  • This embodiment uses data in the literature to obtain the gas production rate of each gas-producing component under each fault condition. Based on the above conditions, the gas production rate equation is obtained by fitting. The gas production rate equation for type N faults is shown in Table 1.
  • the real model is imported into Fluent for GIS internal fluid simulation, and the original data is obtained, including:
  • characteristic gases such as SO 2 F 2 , SOF 2 , SO 2 , CO 2 and CF 4
  • the five characteristic gas components are all less than 2300 when calculating the Raylaw coefficient.
  • the GIS chamber can be roughly seen as a pipe shape, so its diffusion mode is set to laminar flow.
  • the original data in this article is generally taken as four hundred sets, and each of the four defect states corresponds to one hundred sets of data.
  • seventy sets of data in the four defect status databases are used as training samples, and the remaining thirty sets of test data are used as test sample data.
  • Table 2 the local data after normalization is given.
  • the Sparrow Optimization Algorithm includes:
  • n is the number of sparrows
  • d is the dimension of the variable to be optimized
  • the discoverer’s individual position update formula is:
  • the joiner’s location update formula is:
  • f g is the current global best fitness value
  • f w is the current global worst fitness value
  • ⁇ is the smallest constant.
  • support vector machine is a machine learning algorithm based on statistics, which has the advantage of smaller training volume. For small clusters with a small number of samples, the training time can be short and the classification effect is good.
  • the core concept is to map the input space to a high-dimensional space through a kernel function, where linear classification can be done using methods. By constructing an optimal classification hyperplane in high-dimensional space and performing classification, the generalization ability and reliability of interval classification between data categories are maximized.
  • the hyperplane function is expressed as:
  • mapping function is the mapping function
  • ⁇ is the normal vector of the optimal classification hyperplane
  • b is the classification threshold
  • the hyperplane function is expressed as:
  • C is the penalty parameter.
  • C is an artificially set hyperparameter. Generally, C>0 is mainly used to balance the relationship between the two. When C is relatively large, the penalty for misclassified sample data will be relatively large, and vice versa. Therefore, minimizing the objective function includes minimizing the sum of distances from the hyperplane S to the nearest sample. It is also necessary to minimize the number of misclassified data samples.
  • the optimal hyperplane function is expressed as:
  • SV is the support vector
  • the kernel function is expressed as:
  • the penalty factor C and the kernel function parameter ⁇ must be given in the SVM model, because the performance of the classification support vector machine is closely related to C and ⁇ . If parameters are given artificially, improper selection may easily lead to poor classification results. Therefore, the factor C and kernel function parameter ⁇ must be optimized. The optimized parameters can then be used to build support vector machine modeling and classify samples. In this way, the SVM model has higher classification accuracy.
  • the classification accuracy of the SVM fault model is the current fitting value of each sparrow, and the maximum real-time updated fitness value. If the sparrow's current fitness value is greater than the saved fitness value, the original fitness value has been updated; otherwise, the original fitness value remains unchanged. And save the best value;
  • sparrow position corresponds to the penalty factor C and kernel function parameter ⁇ in SVM.
  • an embodiment of the present invention provides a GIS fault mode identification method based on SSA-SVM.
  • scientific demonstration is carried out by comparing with traditional algorithms.
  • this embodiment Based on the three-concentration ratio method, this embodiment adds the ratios between several other main decomposition characteristic gas components as characteristic quantities for judging faults on the basis of the three characteristic component ratios. Part of the data is shown in Table 3. Show.
  • test samples In order to test the accuracy of the proposed pattern recognition method on partial discharge patterns, seventy test samples and thirty training samples were selected from the sample set of each defect type and identified. The test results are shown in Figure 4.
  • This embodiment uses the Whale Optimization Algorithm (WOA) and the Particle Swarm Optimization Algorithm (PSO) to optimize the parameters of SVM and compare the optimization effects with SSA.
  • Figure 5 shows the fitness curves under each optimization algorithm. It can be seen that the whale algorithm performs global search in the early stage and local search in the later stage, and it is easy to enter the local optimal situation; while the particle swarm algorithm has a weak local search function. The difference is that it cannot quickly find the local optimal solution, nor can it guarantee that the global optimal solution can be quickly searched.
  • Sparrow optimization algorithm is added After incorporating the investigation and early warning mechanism, the time required for search is significantly reduced and the classification effect of samples is improved. It can be judged from this that SSA-SVM has certain significance in establishing a diagnostic model for GIS fault mode recognition.

Abstract

An SSA-SVM-based gas insulated substation (GIS) fault mode recognition method, comprising: establishing a real model of a GIS, and importing a gas production rate equation of each decomposition product of SF6 under a fault, so as to acquire original data; optimizing a kernel function parameter and a penalty coefficient of a support vector machine on the basis of a sparrow search algorithm; and establishing a support vector machine fault diagnosis model, which is optimized on the basis of the sparrow search algorithm, and inputting the data into the diagnosis model to obtain a fault result. Compared with traditional methods, the method has the advantages of a high calculation speed and a higher recognition rate.

Description

一种基于SSA-SVM的GIS故障模式识别方法A GIS fault mode identification method based on SSA-SVM 技术领域Technical field
本发明涉及气体绝缘变电站的技术领域，尤其涉及一种基于SSA-SVM的GIS故障模式识别方法。The present invention relates to the technical field of gas insulated substations, and in particular to a GIS fault mode identification method based on SSA-SVM.
背景技术Background technique
气体绝缘变电站(GIS)具备占用面积较小、架设结构简单、不受外部影响等优势，在高压输变电系统中获得了较为广泛的运用。然而，GIS的设计、生产、运输、装配、操作、维修等各环节不可避免地都会产生许多内部绝缘缺陷，从而危害GIS的安全运行。这些缺陷主要包括制造质量差导致的留在气室的金属颗粒、运行中机械摩擦或振动导致的金属毛刺或粉末，以及导体和支撑绝缘子断裂导致的微小气隙等。此类缺陷会扭曲GIS的固有电场，从而在某些情况下导致局部放电，并在严重情况下导致火花或电弧。这些故障造成的影响较大，后果严重，如果不及时解决GIS中出现的上述缺陷，则设备在运行过程中会发生绝缘故障，GIS的安全可靠运行以及整个电力系统本身都将受到不同程度的威胁。Gas-insulated substations (GIS) have the advantages of small occupation area, simple erection structure, and are not affected by external influences, and have been widely used in high-voltage power transmission and transformation systems. However, the design, production, transportation, assembly, operation, maintenance and other aspects of GIS will inevitably produce many internal insulation defects, thus endangering the safe operation of GIS. These defects mainly include metal particles left in the air chamber caused by poor manufacturing quality, metal burrs or powder caused by mechanical friction or vibration during operation, and tiny air gaps caused by fractures of conductors and supporting insulators. Such defects can distort the inherent electric field of the GIS, causing partial discharges in some cases and, in severe cases, sparks or arcing. These faults have a greater impact and serious consequences. If the above defects in GIS are not solved in time, insulation faults will occur during the operation of the equipment. The safe and reliable operation of GIS and the entire power system itself will be threatened to varying degrees. .
经研究发现，在故障条件下，绝缘气体SF6将分解并与设备中的杂质(H2O、O2、有机物等)反应，生成各种气体成分。分解成分的性质可以为检修提供重要信息，因为成分的类型、含量、不同成分之间的比值与故障的类型和程度密切相关。因此，研究人员正在努力研究SF6的分解现象、机理以及基于上述相关性的绝缘故障诊断。有关文献中指出，GIS故障类型可总结为以下故障类型，包括高压导体突出物(N类)、自由导电微粒(P类)、绝缘子金属污染(M类)或绝缘子外气隙绝缘缺陷(GE类)等。Research has found that under fault conditions, the insulating gas SF 6 will decompose and react with impurities (H 2 O, O 2 , organic matter, etc.) in the equipment to generate various gas components. The properties of decomposed components can provide important information for maintenance, because the type, content, and ratio between different components are closely related to the type and degree of failure. Therefore, researchers are working hard to study the decomposition phenomenon and mechanism of SF 6 as well as insulation fault diagnosis based on the above correlation. Relevant literature points out that GIS fault types can be summarized as the following fault types, including high-voltage conductor protrusions (N category), free conductive particles (P category), insulator metal contamination (M category) or insulator outer air gap insulation defects (GE category) )wait.
目前针对于故障类型识别已经提出了较多的智能优化方法，包括神经网络、概率神经网络和极限学习机等。上述传统方法应用于GIS设备故障识别，存在故障诊断率偏低以及稳定性差等问题，本发明的提出可以很好的解决这一系列问题。At present, many intelligent optimization methods have been proposed for fault type identification, including neural networks, probabilistic neural networks and extreme learning machines. The above-mentioned traditional method is applied to GIS equipment fault identification, but there are problems such as low fault diagnosis rate and poor stability. The proposal of the present invention can well solve this series of problems.
发明内容Contents of the invention
本部分的目的在于概述本发明的实施例的一些方面以及简要介绍一些较佳实施例。在本部分以及本申请的说明书摘要和发明名称中可能会做些简化或省略以避免使本部分、说明书摘要和发明名称的目的模糊，而这种简化或省略 不能用于限制本发明的范围。The purpose of this section is to outline some aspects of embodiments of the invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, the abstract and the title of the invention to avoid obscuring the purpose of this section, the abstract and the title of the invention, and such simplifications or omissions It is not intended to limit the scope of the invention.
鉴于上述现有存在的问题，提出了本发明。In view of the above-mentioned existing problems, the present invention is proposed.
因此，本发明解决的技术问题是：现有的故障识别技术存在故障诊断率低以及稳定性差的问题。Therefore, the technical problem solved by the present invention is that the existing fault identification technology has the problems of low fault diagnosis rate and poor stability.
为解决上述技术问题，本发明提供如下技术方案：一种基于SSA-SVM的GIS故障模式识别方法，包括：In order to solve the above technical problems, the present invention provides the following technical solution: a GIS fault mode identification method based on SSA-SVM, including:
建立气体绝缘变电站真实模型，导入故障下SF6各个分解产物的产气速率方程获取原始数据；Establish a real model of a gas-insulated substation, and introduce the gas production rate equation of each decomposition product of SF6 under fault to obtain original data;
基于麻雀搜索算法优化支持向量机的核函数参数和惩罚系数；Optimize the kernel function parameters and penalty coefficients of the support vector machine based on the Sparrow search algorithm;
建立麻雀搜索算法优化下的支持向量机故障诊断模型，将数据输入诊断模型得到故障结果。A support vector machine fault diagnosis model optimized by the Sparrow search algorithm is established, and the data is input into the diagnosis model to obtain the fault results.
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：所述气体绝缘变电站真实模型是一个封闭容器，特征气体组分分子的扩散只能通过分子热力学运动来处理。As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the real model of the gas insulated substation is a closed container, and the diffusion of characteristic gas component molecules can only be achieved through molecular thermodynamic movement. deal with.
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：所述SF6气体的分解，包括：As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the decomposition of SF 6 gas includes:
SF6气体的初步解离过程大致可分为3个阶段：(1)在GIS腔体内部，SF6分子受到电子碰撞，形成稳定性较差的亚稳态分子团(SF6)*，在极短的时间内又会转化为SF6 -或者SF5 -等负离子；(2)电场强度气体分子密度均会影响SF6气体的分解产物形式，当电场强度与气体分子密度的比值较低时，SF6 -会与SF6发生反应，生成SF5 -，随着电场强度与气体分子密度比值的增大，SF6 -或SF5 -会与SF6发生反应，产生SF5或SF4等低氟硫化物，当电场强度与气体分子密度的比值进一步增大，SF5 -会与SF6发生反应，产生多种低氟硫化物，包括：SF4、SF3、SF2、SF等；(3)在SF6气体解离过程中形成的各种低氟硫化物也会相互反应，产生更为稳定的产物S2F10，但当环境温度高于200℃时S2F10又会彻底分解，因此在电晕放电、火花放电、局部放电等能量较低的放电条件下更易生成该产物。The initial dissociation process of SF 6 gas can be roughly divided into three stages: (1) Inside the GIS cavity, SF 6 molecules are subject to electron collisions to form metastable molecular groups (SF 6 )* with poor stability. In a very short period of time, it will be converted into negative ions such as SF 6 - or SF 5 - ; (2) The electric field strength and gas molecular density will affect the form of decomposition products of SF 6 gas. When the ratio of electric field strength to gas molecular density is low, , SF 6 - will react with SF 6 to generate SF 5 - . As the ratio of electric field intensity to gas molecule density increases, SF 6 - or SF 5 - will react with SF 6 to generate SF 5 or SF 4 , etc. For low-fluorine sulfides, when the ratio between the electric field intensity and the density of gas molecules further increases, SF 5 - will react with SF 6 to produce a variety of low-fluorine sulfides, including: SF 4 , SF 3 , SF 2 , SF, etc.; (3) Various low-fluorine sulfides formed during the dissociation process of SF 6 gas will also react with each other to produce a more stable product S 2 F 10 , but when the ambient temperature is higher than 200°C, S 2 F 10 will It is completely decomposed, so it is easier to generate this product under discharge conditions with lower energy such as corona discharge, spark discharge, and partial discharge.
SF6气体在初步解离过程中，会形成各种低氟硫化物，大部分的低氟硫化物会F原子迅速结合，结合并恢复为SF6分子，但当腔体内部含有水分和氧气时，这些低氟硫化物会与H2O、O2等发生更加复杂的化学反应，生成多种新 型产物。During the initial dissociation process of SF 6 gas, various low-fluorine sulfides will be formed. Most of the low-fluorine sulfides will be quickly combined with F atoms, combined and restored to SF 6 molecules. However, when the cavity contains moisture and oxygen, , these low-fluorine sulfides will undergo more complex chemical reactions with H 2 O, O 2, etc., generating a variety of new type product.
在故障条件下，受电子碰撞等因素影响H2O和O2会解离产生O、OH等具有高活性的粒子，低氟硫化物将与OH、O结合，生成HF、SOF4、SO2F2、S2OF10、SOF2及SO2等产物。其中SOF2主要由SF4与H2O或OH反应产生，SO2F2主要由SF2与O2反应产生，SO2主要由SOF2与H2O反应产生，S2OF10主要由SF5与O2反应产生，SOF4主要由SF5与O或OH反应产生。由于SF5O稳定性较差，其中大部分会分解成SOF4，少部分会与SF5反应产生S2OF10。Under fault conditions, H 2 O and O 2 will dissociate due to factors such as electron collision to produce highly active particles such as O and OH. Low fluorine sulfide will combine with OH and O to generate HF, SOF 4 and SO 2 Products such as F 2 , S 2 OF 10 , SOF 2 and SO 2 . Among them, SOF 2 is mainly produced by the reaction of SF 4 and H 2 O or OH, SO 2 F 2 is mainly produced by the reaction of SF 2 and O 2 , SO 2 is mainly produced by the reaction of SOF 2 and H 2 O, and S 2 OF 10 is mainly produced by the reaction of SF 5 is produced by the reaction with O 2 , and SOF 4 is mainly produced by the reaction of SF 5 with O or OH. Since SF 5 O has poor stability, most of it will decompose into SOF 4 and a small part will react with SF 5 to produce S 2 OF 10 .
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：有效产气速率RRMS是指每小时某种气体组分体积分数的平均值，所述产气速率方程表示为：
As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the effective gas production rate R RMS refers to the average volume fraction of a certain gas component per hour, and the gas production rate The rate equation is expressed as:
其中，RRMS为绝对产气速率，ci，1为第1次测得组分i的含量，ci，2为第2次测得组分i的含量，Δt为两次检测的时间间隔。Among them, R RMS is the absolute gas production rate, c i,1 is the content of component i measured for the first time, c i,2 is the content of component i measured for the second time, Δt is the time interval between two measurements .
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：将所述真实模型下导入fluent进行内部流体仿真，得到原始数据，包括：As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the real model is imported into fluent for internal fluid simulation to obtain original data, including:
将SF6以及SO2F2、SOF2、SO2、CO2、CF4这5种特征气体的密度、导热率等参数导入到fluent；在气体绝缘变电站仿真模型里定义一个区域，在此区域内特征气体将按照特征速率方程产生特征气体；在固定温度以及压力的条件下，将不同故障缺陷下的产气速率方程导入此区域，然后扩散至整个气体绝缘变电站，得到各个缺陷下特征气体组分的原始数据。Import the density, thermal conductivity and other parameters of SF 6 and five characteristic gases such as SO 2 F 2 , SOF 2 , SO 2 , CO 2 and CF 4 into fluent; define an area in the gas insulated substation simulation model, and in this area The characteristic gas inside will produce characteristic gas according to the characteristic rate equation; under the conditions of fixed temperature and pressure, the gas production rate equation under different fault defects is introduced into this area, and then diffused to the entire gas insulated substation to obtain the characteristic gas group under each defect. original data.
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：所述麻雀优化算法(SSA)，包括：As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the Sparrow optimization algorithm (SSA) includes:
在算法寻优的模拟仿真实验中，使用虚拟麻雀觅食，麻雀的位置表示为：
In the simulation experiment of algorithm optimization, a virtual sparrow is used to forage, and the position of the sparrow is expressed as:
其中，n为麻雀的数量，d为待优化变量的维数； Among them, n is the number of sparrows, and d is the dimension of the variable to be optimized;
所有麻雀的适应度值表示为：
The fitness value of all sparrows is expressed as:
在算法迭代过程中发现者的个体位置更新公式为：
During the algorithm iteration process, the discoverer’s individual position update formula is:
其中，为第i个个体迭代t次时的第j维数，α∈(0，1]为随机数，R2∈[0，1]为报警值，ST为安全阈值,Q为服从正态分布的随机数，L为一个1×d的矩阵，其中该矩阵内所有元素都是1；in, is the j-th dimension of the i-th individual when iterating t times, α∈(0, 1] is a random number, R 2 ∈ [0, 1] is the alarm value, ST is the safety threshold, and Q is a normal distribution. Random number, L is a 1×d matrix, where all elements in the matrix are 1;
加入者的位置更新公式为：
The joiner’s location update formula is:
其中，Xp为目前发现者的最佳位置，Xworst为当前全局的最差位置，A为一个1×d的矩阵，其中每个元素随机赋值为1或-1，并且A+＝AT(AAT)-1； Among them , _ (AA T ) -1 ;
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：对比本次迭代个体适应度值与当前最佳适应度值，当算法陷入局部最优时，位置更新公式表示为：
As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the individual fitness value of this iteration is compared with the current best fitness value. When the algorithm falls into a local optimum, the location is updated. The formula is expressed as:
其中，Xbest为当前全局最优位置，β为步长控制参数，服从均值为0，方差为1的正态分布的随机数，k∈[-1，1]为随机数，fi为当前麻雀的个体适应度值，fg为当前全局最佳的适应度值，fw为当前全局最差的适应度值，∈为最小的常数。当fi＞fg时，表示麻雀位于种群的边缘，当fi＝fg时，表明处于种群中间的麻雀意识到危险，需要移动到其他地方。 Among them , For the individual fitness value of the sparrow, f g is the current global best fitness value, f w is the current global worst fitness value, and ∈ is the smallest constant. When fi > f g , it means that the sparrow is located at the edge of the population. When fi = f g , it means that the sparrow in the middle of the population is aware of the danger and needs to move to other places.
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方 案，其中：支持向量机(SVM)是一种基于统计的机器学习算法，其具有训练量较小的优势，对于小型集群样本数少的情况，可以做到训练时间短，分类效果好。核心概念是在输入空间中映射到高维空间通过核函数，其中线性分类可以使用方法。通过在高维空间中构造了一个最优分类超平面并进行分类，最大限度地提高数据类别间的间隔分类的泛化能力和可信度。As a preferred method of GIS fault mode identification method based on SSA-SVM according to the present invention, Case, among which: Support Vector Machine (SVM) is a machine learning algorithm based on statistics. It has the advantage of small training volume. For small clusters with few samples, it can achieve short training time and good classification effect. The core concept is to map the input space to a high-dimensional space through a kernel function, where linear classification can be done using methods. By constructing an optimal classification hyperplane in high-dimensional space and performing classification, the generalization ability and reliability of interval classification between data categories are maximized.
所述超平面函数表示为：
The hyperplane function is expressed as:
其中，为映射函数，ω为最优分类超平面的法向量，b为分类阈值；in, is the mapping function, ω is the normal vector of the optimal classification hyperplane, and b is the classification threshold;
由于线性不可分的数据集须满足yi(ω·x+b)≥1这个条件才能使用硬间隔SVM，通过对每一个条件进行一定的松弛操作，引入松弛变量ξ，使每一个硬间隔支持向量机对于函数距离的使用满足:Since the linearly inseparable data set must satisfy the condition y i (ω·x+b)≥1 in order to use hard-margin SVM, by performing certain relaxation operations on each condition, the relaxation variable ξ is introduced to make each hard-margin support vector The machine's use of functional distance satisfies:
yi(ω·xi+b)≥1-Kξi(1≤i≤m)y i (ω·x i +b)≥1-Kξ i (1≤i≤m)
其中，ξi≥0；Among them, ξ i ≥ 0;
因为引入了另外的参数，所以需要对原目标函数和松弛变量之间的关系进行平衡，此时超平面函数表示为：
Because additional parameters are introduced, the relationship between the original objective function and the slack variable needs to be balanced. At this time, the hyperplane function is expressed as:
其中，C为惩罚参数。Among them, C is the penalty parameter.
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：将超平面的寻优问题归结为二次规划问题，表示为：
As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, the hyperplane optimization problem is reduced to a quadratic planning problem, expressed as:
s.t.yi(ω·xi+b)≥1-ξi(1≤i≤m)sty i (ω·x i +b)≥1-ξ i (1≤i≤m)
ξi≥0(1≤i≤m) ξi≥0 (1≤i≤m)
基于拉格朗日乘子算法，最优超平面函数表示为：
Based on the Lagrange multiplier algorithm, the optimal hyperplane function is expressed as:
其中，SV为支持向量；Among them, SV is the support vector;
核函数表示为：
The kernel function is expressed as:
作为本发明所述的基于SSA-SVM的GIS故障模式识别方法的一种优选方案，其中：建立麻雀搜索算法优化下的支持向量机故障诊断模型，包括：As a preferred solution of the GIS fault mode identification method based on SSA-SVM of the present invention, a support vector machine fault diagnosis model optimized by the Sparrow search algorithm is established, including:
对原始数据进行预处理并将数据导入算法；选择训练集和测试集，根据研究情况导入数据；初始化麻雀搜索算法和支持向量机的参数；更新麻雀的位置；计算每个种群麻雀的位置以获得当前的新位置；计算适应值，并保存最佳值；根据最佳适应值保存相应的支持向量机参数组合；导入最佳参数组合使用支持向量机进行计算，判断终止条件，得到GIS故障的模式代码。Preprocess the original data and import the data into the algorithm; select the training set and test set, and import the data according to the research situation; initialize the parameters of the sparrow search algorithm and support vector machine; update the position of the sparrow; calculate the position of each population of sparrows to obtain The current new position; calculate the fitness value and save the best value; save the corresponding support vector machine parameter combination according to the best fitness value; import the best parameter combination and use the support vector machine for calculation, determine the termination condition, and obtain the GIS failure mode code.
本发明的有益效果：本发明提供的基于SSA-SVM的GIS故障模式识别方法通过建立GIS真实模型得到不同故障缺陷下特征气体组分的原始数据，通过将数据导入基于SSA算法优化下的SVM故障诊断模型，可以准确得到GIS的故障模式，相比传统方法拥有计算速度快，识别率更高的优点。Beneficial effects of the present invention: The GIS fault mode identification method based on SSA-SVM provided by the present invention obtains original data of characteristic gas components under different fault defects by establishing a real GIS model, and imports the data into SVM faults optimized based on the SSA algorithm. The diagnostic model can accurately obtain the fault mode of GIS. Compared with traditional methods, it has the advantages of faster calculation and higher recognition rate.
附图说明Description of drawings
为了更清楚地说明本发明实施例的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其它的附图。其中：In order to explain the technical solutions of the embodiments of the present invention more clearly, the drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. Those of ordinary skill in the art can also obtain other drawings based on these drawings without exerting any creative effort. in:
图1为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的SSA-SVM流程图；Figure 1 is an SSA-SVM flow chart of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图2为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的GIS真实模型；Figure 2 is a real GIS model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图3为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的SF6气体分解路径图；Figure 3 is a SF 6 gas decomposition path diagram of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图4为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的SSA-SVM诊断模型实际与预测分类图；Figure 4 is an actual and predicted classification diagram of the SSA-SVM diagnostic model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图5为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的各个优化算法的函数适应度图；Figure 5 is a function fitness diagram of each optimization algorithm of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图6为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的对比算法WOA-SVM诊断模型实际与预测分类图； Figure 6 is a comparison algorithm WOA-SVM diagnostic model actual and predicted classification diagram of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention;
图7为本发明一个实施例提供的一种基于SSA-SVM的GIS故障模式识别方法的对比算法PSO-SVM诊断模型实际与预测分类图。Figure 7 is a comparison diagram of the actual and predicted classification of the PSO-SVM diagnostic model of a GIS fault mode identification method based on SSA-SVM provided by an embodiment of the present invention.
具体实施方式Detailed ways
为使本发明的上述目的、特征和优点能够更加明显易懂，下面结合说明书附图对本发明的具体实施方式做详细的说明，显然所描述的实施例是本发明的一部分实施例，而不是全部实施例。基于本发明中的实施例，本领域普通人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本发明的保护的范围。In order to make the above objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. It is obvious that the described embodiments are part of the embodiments of the present invention, not all of them. Example. Based on the embodiments of the present invention, all other embodiments obtained by ordinary people in the art without creative efforts should fall within the protection scope of the present invention.
在下面的描述中阐述了很多具体细节以便于充分理解本发明，但是本发明还可以采用其他不同于在此描述的其它方式来实施，本领域技术人员可以在不违背本发明内涵的情况下做类似推广，因此本发明不受下面公开的具体实施例的限制。Many specific details are set forth in the following description to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described here. Those skilled in the art can do so without departing from the connotation of the present invention. Similar generalizations are made, and therefore the present invention is not limited to the specific embodiments disclosed below.
其次，此处所称的“一个实施例”或“实施例”是指可包含于本发明至少一个实现方式中的特定特征、结构或特性。在本说明书中不同地方出现的“在一个实施例中”并非均指同一个实施例，也不是单独的或选择性的与其他实施例互相排斥的实施例。Second, reference herein to "one embodiment" or "an embodiment" refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. "In one embodiment" appearing in different places in this specification does not all refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.
本发明结合示意图进行详细描述，在详述本发明实施例时，为便于说明，表示器件结构的剖面图会不依一般比例作局部放大，而且所述示意图只是示例，其在此不应限制本发明保护的范围。此外，在实际制作中应包含长度、宽度及深度的三维空间尺寸。The present invention will be described in detail with reference to schematic diagrams. When describing the embodiments of the present invention in detail, for the convenience of explanation, the cross-sectional diagrams showing the device structure will be partially enlarged according to the general scale. Moreover, the schematic diagrams are only examples and shall not limit the present invention. scope of protection. In addition, the three-dimensional dimensions of length, width and depth should be included in actual production.
同时在本发明的描述中，需要说明的是，术语中的“上、下、内和外”等指示的方位或位置关系为基于附图所示的方位或位置关系，仅是为了便于描述本发明和简化描述，而不是指示或暗示所指的装置或元件必须具有特定的方位、以特定的方位构造和操作，因此不能理解为对本发明的限制。此外，术语“第一、第二或第三”仅用于描述目的，而不能理解为指示或暗示相对重要性。At the same time, in the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "upper, lower, inner and outer" are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention. The invention and simplified description are not intended to indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore are not to be construed as limitations of the invention. Furthermore, the terms "first, second or third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
本发明中除非另有明确的规定和限定，术语“安装、相连、连接”应做广义理解，例如：可以是固定连接、可拆卸连接或一体式连接；同样可以是机械连接、电连接或直接连接，也可以通过中间媒介间接相连，也可以是两个元件内部的连通。对于本领域的普通技术人员而言，可以具体情况理解上述术语在本发明中的具体含义。 Unless otherwise clearly stated and limited in the present invention, the terms "installation, connection, and connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integrated connection; it can also be a mechanical connection, an electrical connection, or a direct connection. A connection can also be indirectly connected through an intermediary, or it can be an internal connection between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood on a case-by-case basis.
实施例1Example 1
参照图1—3，为本发明的一个实施例，例提供了一种基于SSA-SVM的GIS故障模式识别方法，包括：Referring to Figures 1-3, an embodiment of the present invention provides a GIS fault mode identification method based on SSA-SVM, including:
S1：建立气体绝缘变电站真实模型，导入故障下SF6各个分解产物的产气速率方程获取原始数据；S1: Establish a real model of a gas-insulated substation, and introduce the gas production rate equation of each decomposition product of SF 6 under fault to obtain original data;
更进一步的，本实施例采用110KV下GIS的尺寸在ANSYS软件中搭建了真实模型，GIS真实模型如图2所示，长9538.31mm，直径最宽处2188.3mm，直径最窄处499.74mm。气体监测点位于设备上端的正中部位。由于设备是一个封闭容器，特征气体组分分子的扩散只能通过分子热力学运动来处理。Furthermore, this embodiment uses the size of GIS under 110KV to build a real model in ANSYS software. The real GIS model is shown in Figure 2, with a length of 9538.31mm, a widest diameter of 2188.3mm, and a narrowest diameter of 499.74mm. The gas monitoring point is located in the middle of the upper end of the equipment. Since the device is a closed container, the diffusion of characteristic gas component molecules can only be handled through molecular thermodynamic motion.
应说明的是，考虑到断路器内的气体压力高于其他气室，整个模拟过程在气体压力为0.4MPa的条件下进行。设备或其他组织的内部母线对空间中的扩散过程几乎没有影响，因此可以忽略它们。It should be noted that considering that the gas pressure in the circuit breaker is higher than that in other gas chambers, the entire simulation process is carried out under the condition of a gas pressure of 0.4MPa. Internal busbars of equipment or other organizations have little impact on the diffusion process in space, so they can be ignored.
更进一步的，SF6气体的初步解离过程大致可分为3个阶段：(1)在GIS腔体内部，SF6分子受到电子碰撞，形成稳定性较差的亚稳态分子团(SF6)*，在极短的时间内又会转化为SF6 -或者SF5 -等负离子；(2)电场强度气体分子密度均会影响SF6气体的分解产物形式，当电场强度与气体分子密度的比值较低时，SF6 -会与SF6发生反应，生成SF5 -，随着电场强度与气体分子密度比值的增大，SF6 -或SF5 -会与SF6发生反应，产生SF5或SF4等低氟硫化物，当电场强度与气体分子密度的比值进一步增大，SF5 -会与SF6发生反应，产生多种低氟硫化物，包括：SF4、SF3、SF2、SF等；(3)在SF6气体解离过程中形成的各种低氟硫化物也会相互反应，产生更为稳定的产物S2F10，但当环境温度高于200℃时S2F10又会彻底分解，因此在电晕放电、火花放电、局部放电等能量较低的放电条件下更易生成该产物。Furthermore, the initial dissociation process of SF 6 gas can be roughly divided into three stages: (1) Inside the GIS cavity, SF 6 molecules are subject to electron collisions to form metastable molecular groups (SF 6 )*, will be converted into negative ions such as SF 6 - or SF 5 - in a very short period of time; (2) The electric field strength and gas molecular density will affect the form of decomposition products of SF 6 gas. When the electric field strength and gas molecular density are When the ratio is low, SF 6 - will react with SF 6 to generate SF 5 - . As the ratio of electric field intensity to gas molecule density increases, SF 6 - or SF 5 - will react with SF 6 to generate SF 5 Or low-fluorine sulfides such as SF 4. When the ratio of electric field intensity to gas molecular density further increases, SF 5 - will react with SF 6 to produce a variety of low-fluorine sulfides, including: SF 4 , SF 3 , SF 2 , SF, etc.; (3) Various low-fluorine sulfides formed during the dissociation process of SF 6 gas will also react with each other to produce a more stable product S 2 F 10 , but when the ambient temperature is higher than 200°C, S 2 F 10 will decompose completely, so it is easier to generate this product under lower energy discharge conditions such as corona discharge, spark discharge, and partial discharge.
应说明的是，SF6气体的解离产物与放电过程中能量泄放形式存在着一定的联系。对于电弧放电和火花放电而言，放电过程能量释放强度较高，SF6分子的解离主要是由电子碰撞和局部过热造成的；而在电晕放电过程中，SF6气体的解离则主要是由于电子碰撞引发的。It should be noted that there is a certain relationship between the dissociation products of SF 6 gas and the form of energy release during the discharge process. For arc discharge and spark discharge, the energy release intensity during the discharge process is high, and the dissociation of SF 6 molecules is mainly caused by electron collision and local overheating; while during the corona discharge process, the dissociation of SF 6 gas is mainly caused by It is caused by the collision of electrons.
SF6气体在初步解离过程中，会形成各种低氟硫化物，大部分的低氟硫化物会F原子迅速结合，结合并恢复为SF6分子，但当腔体内部含有水分和氧气时，这些低氟硫化物会与H2O、O2等发生更加复杂的化学反应，如图3所示生 成多种新型产物。During the initial dissociation process of SF 6 gas, various low-fluorine sulfides will be formed. Most of the low-fluorine sulfides will be quickly combined with F atoms, combined and restored to SF 6 molecules. However, when the cavity contains moisture and oxygen, , these low-fluorine sulfides will undergo more complex chemical reactions with H 2 O, O 2, etc., as shown in Figure 3. into a variety of new products.
更进一步的，在故障条件下，受电子碰撞等因素影响H2O和O2会解离产生O、OH等具有高活性的粒子，低氟硫化物将与OH、O结合，生成HF、SOF4、SO2F2、S2OF10、SOF2及SO2等产物。其中SOF2主要由SF4与H2O或OH反应产生，SO2F2主要由SF2与O2反应产生，SO2主要由SOF2与H2O反应产生，S2OF10主要由SF5与O2反应产生，SOF4主要由SF5与O或OH反应产生。由于SF5O稳定性较差，其中大部分会分解成SOF4，少部分会与SF5反应产生S2OF10。Furthermore, under fault conditions, H 2 O and O 2 will dissociate due to factors such as electron collision to produce highly active particles such as O and OH. Low fluorine sulfide will combine with OH and O to generate HF and SOF. 4. SO 2 F 2 , S 2 OF 10 , SOF 2 and SO 2 and other products. Among them, SOF 2 is mainly produced by the reaction of SF 4 and H 2 O or OH, SO 2 F 2 is mainly produced by the reaction of SF 2 and O 2 , SO 2 is mainly produced by the reaction of SOF 2 and H 2 O, and S 2 OF 10 is mainly produced by the reaction of SF 5 is produced by the reaction with O 2 , and SOF 4 is mainly produced by the reaction of SF 5 with O or OH. Since SF 5 O has poor stability, most of it will decompose into SOF 4 and a small part will react with SF 5 to produce S 2 OF 10 .
更进一步的，有效产气速率RRMS是指每小时某种气体组分体积分数的平均值，所述产气速率方程表示为：
Furthermore, the effective gas production rate R RMS refers to the average volume fraction of a certain gas component per hour. The gas production rate equation is expressed as:
其中，RRMS为绝对产气速率，ci，1为第1次测得组分i的含量，ci，2为第2次测得组分i的含量，Δt为两次检测的时间间隔。Among them, R RMS is the absolute gas production rate, c i,1 is the content of component i measured for the first time, c i,2 is the content of component i measured for the second time, Δt is the time interval between two measurements .
本实施例采用文献中的数据得到各产气组分在各个故障条件下的产气速率，基于上述条件拟合得到产气速率方程，N类故障的产气速率方程如表1所示。This embodiment uses data in the literature to obtain the gas production rate of each gas-producing component under each fault condition. Based on the above conditions, the gas production rate equation is obtained by fitting. The gas production rate equation for type N faults is shown in Table 1.
表1各组分产气速率方程(N类)
Table 1 Gas production rate equation of each component (N type)
更进一步的，将真实模型下导入fluent进行GIS内部流体仿真，得到原始数据，包括： Furthermore, the real model is imported into Fluent for GIS internal fluid simulation, and the original data is obtained, including:
将SF6以及SO2F2、SOF2、SO2、CO2、CF4这5种特征气体的密度、导热率等参数导入到fluent；在气体绝缘变电站仿真模型里定义一个区域，在此区域内特征气体将按照特征速率方程产生特征气体；在固定温度以及压力的条件下，将不同故障缺陷下的产气速率方程导入此区域，然后扩散至整个气体绝缘变电站，得到各个缺陷下特征气体组分的原始数据。Import the density, thermal conductivity and other parameters of SF 6 and five characteristic gases such as SO 2 F 2 , SOF 2 , SO 2 , CO 2 and CF 4 into fluent; define an area in the gas insulated substation simulation model, and in this area The characteristic gas inside will produce characteristic gas according to the characteristic rate equation; under the conditions of fixed temperature and pressure, the gas production rate equation under different fault defects is introduced into this area, and then diffused to the entire gas insulated substation to obtain the characteristic gas group under each defect. original data.
应说明的是，5种特征气体组分在计算雷洛系数时均小于2300，GIS的腔室大致可以看出管道状，所以其扩散方式均设定为层流。It should be noted that the five characteristic gas components are all less than 2300 when calculating the Raylaw coefficient. The GIS chamber can be roughly seen as a pipe shape, so its diffusion mode is set to laminar flow.
考虑到采集原始数据需要的时间成本，在本文中原始数据一般取为四百组，其中的四种缺陷状态各对应于一百组数据。在故障识别系统运行过程中，将四种缺陷状态数据库中的七十组数据作为训练样本，而其余的三十组测试数据则作为试验样品数据。如表2所示，给出了经归一化处理后的局部数据。Considering the time cost required to collect original data, the original data in this article is generally taken as four hundred sets, and each of the four defect states corresponds to one hundred sets of data. During the operation of the fault identification system, seventy sets of data in the four defect status databases are used as training samples, and the remaining thirty sets of test data are used as test sample data. As shown in Table 2, the local data after normalization is given.
表2 4种故障下各个气体特征量归一化后的数据
Table 2 Normalized data of each gas characteristic quantity under four types of faults
S2：基于麻雀搜索算法优化支持向量机的核函数参数和惩罚系数；S2: Optimize the kernel function parameters and penalty coefficients of the support vector machine based on the Sparrow search algorithm;
更进一步的，麻雀优化算法(SSA)，包括：Furthermore, the Sparrow Optimization Algorithm (SSA) includes:
在算法寻优的模拟仿真实验中，使用虚拟麻雀觅食，麻雀的位置表示为：
In the simulation experiment of algorithm optimization, a virtual sparrow is used to forage, and the position of the sparrow is expressed as:
其中，n为麻雀的数量，d为待优化变量的维数；Among them, n is the number of sparrows, and d is the dimension of the variable to be optimized;
所有麻雀的适应度值表示为：
The fitness values of all sparrows are expressed as:
在算法迭代过程中发现者的个体位置更新公式为：
During the algorithm iteration process, the discoverer’s individual position update formula is:
其中，为第i个个体迭代t次时的第j维数，α∈(0，1]为随机数，R2∈[0，1]为报警值，ST为安全阈值,Q为服从正态分布的随机数，L为一个1×d的矩阵，其中该矩阵内所有元素都是1；in, is the j-th dimension of the i-th individual when iterating t times, α∈(0, 1] is a random number, R 2 ∈ [0, 1] is the alarm value, ST is the safety threshold, and Q is a normal distribution. Random number, L is a 1×d matrix, where all elements in the matrix are 1;
加入者的位置更新公式为：
The joiner’s location update formula is:
其中，Xp为目前发现者的最佳位置，Xworst为当前全局的最差位置，A为一个1×d的矩阵，其中每个元素随机赋值为1或-1，并且A+＝AT(AAT)-1； Among them , _ (AA T ) -1 ;
更进一步的，对比本次迭代个体适应度值与当前最佳适应度值，当算法陷入局部最优时，位置更新公式表示为：
Furthermore, comparing the individual fitness value of this iteration with the current best fitness value, when the algorithm falls into a local optimum, the position update formula is expressed as:
其中，Xbest为当前全局最优位置，β为步长控制参数，服从均值为0，方差为1的正态分布的随机数，k∈[-1，1]为随机数，fi为当前麻雀的个体适应度值，fg为当前全局最佳的适应度值，fw为当前全局最差的适应度值，ε为最小的常数。当fi＞fg时，表示麻雀位于种群的边缘，当fi＝fg时，表明处于种群中间的麻雀意识到危险，需要移动到其他地方。 Among them , For the individual fitness value of the sparrow, f g is the current global best fitness value, f w is the current global worst fitness value, and ε is the smallest constant. When fi > f g , it means that the sparrow is located at the edge of the population. When fi = f g , it means that the sparrow in the middle of the population is aware of the danger and needs to move to other places.
更进一步的，支持向量机(SVM)是一种基于统计的机器学习算法，其具 有训练量较小的优势，对于小型集群样本数少的情况，可以做到训练时间短，分类效果好。核心概念是在输入空间中映射到高维空间通过核函数，其中线性分类可以使用方法。通过在高维空间中构造了一个最优分类超平面并进行分类，最大限度地提高数据类别间的间隔分类的泛化能力和可信度。Furthermore, support vector machine (SVM) is a machine learning algorithm based on statistics, which has It has the advantage of smaller training volume. For small clusters with a small number of samples, the training time can be short and the classification effect is good. The core concept is to map the input space to a high-dimensional space through a kernel function, where linear classification can be done using methods. By constructing an optimal classification hyperplane in high-dimensional space and performing classification, the generalization ability and reliability of interval classification between data categories are maximized.
超平面函数表示为：
The hyperplane function is expressed as:
其中，为映射函数，ω为最优分类超平面的法向量，b为分类阈值；in, is the mapping function, ω is the normal vector of the optimal classification hyperplane, and b is the classification threshold;
由于线性不可分的数据集须满足yi(ω·x+b)≥1这个条件才能使用硬间隔SVM，通过对每一个条件进行一定的松弛操作，引入松弛变量ξ，使每一个硬间隔支持向量机对于函数距离的使用满足:Since the linearly inseparable data set must satisfy the condition y i (ω·x+b)≥1 in order to use hard-margin SVM, by performing certain relaxation operations on each condition, the relaxation variable ξ is introduced to make each hard-margin support vector The machine's use of functional distance satisfies:
yi(ω·xi+b)≥1-Kξi(1≤i≤m)y i (ω·x i +b)≥1-Kξ i (1≤i≤m)
其中，ξi≥0；Among them, ξ i ≥ 0;
因为引入了另外的参数，所以需要对原目标函数和松弛变量之间的关系进行平衡，此时超平面函数表示为：
Because additional parameters are introduced, the relationship between the original objective function and the slack variable needs to be balanced. At this time, the hyperplane function is expressed as:
其中，C为惩罚参数。Among them, C is the penalty parameter.
应说明的是，C是人为设定的超参数，一般情况下C＞0，主要是用来平衡两者之间的关系。当C比较大的时候对那些误分类的样本数据的惩罚作用就会比较大，反之则会比较小，所以最小化目标函数就包括了既要使得超平面S到最近样本的距离之和最小，又要使得误分类的数据样本的数目最小。It should be noted that C is an artificially set hyperparameter. Generally, C>0 is mainly used to balance the relationship between the two. When C is relatively large, the penalty for misclassified sample data will be relatively large, and vice versa. Therefore, minimizing the objective function includes minimizing the sum of distances from the hyperplane S to the nearest sample. It is also necessary to minimize the number of misclassified data samples.
更进一步的，将超平面的寻优问题归结为二次规划问题，表示为：
Furthermore, the hyperplane optimization problem is reduced to a quadratic programming problem, expressed as:
s.t.yi(ω·xi+b)≥1-ξi(1≤i≤m)sty i (ω·x i +b)≥1-ξ i (1≤i≤m)
ξi≥0(1≤i≤m) ξi≥0 (1≤i≤m)
基于拉格朗日乘子算法，最优超平面函数表示为：
Based on the Lagrange multiplier algorithm, the optimal hyperplane function is expressed as:
其中，SV为支持向量； Among them, SV is the support vector;
核函数表示为：
The kernel function is expressed as:
应说明的是，惩罚因子C与核函数参数σ必须在SVM模型中给出，因为分类支持向量机的性能与C和σ密切相关。如果人为给出参数，选择不当容易导致分类结果不佳。因此，必须优化因子C和核函数参数σ。然后可以使用优化的参数来建立支持向量机建模和分类样本。这样，SVM该模型才具有较高的分类精度。It should be noted that the penalty factor C and the kernel function parameter σ must be given in the SVM model, because the performance of the classification support vector machine is closely related to C and σ. If parameters are given artificially, improper selection may easily lead to poor classification results. Therefore, the factor C and kernel function parameter σ must be optimized. The optimized parameters can then be used to build support vector machine modeling and classify samples. In this way, the SVM model has higher classification accuracy.
S3：建立麻雀搜索算法优化下的支持向量机故障诊断模型,将数据输入诊断模型得到故障结果。S3: Establish a support vector machine fault diagnosis model optimized by the Sparrow search algorithm, and input the data into the diagnosis model to obtain the fault results.
更进一步的，SSA-SVM的故障诊断模型的流程如图1所示，包括：Furthermore, the process of SSA-SVM fault diagnosis model is shown in Figure 1, including:
(1)首先，对原始数据进行预处理并将数据导入算法。(1) First, preprocess the original data and import the data into the algorithm.
(2)选择训练集和测试集，根据研究情况导入数据；(2) Select the training set and test set, and import data according to the research situation;
(3)初始化SVM参数和SSA的参数，包括麻雀群大小，最大迭代次数等参数；(3) Initialize SVM parameters and SSA parameters, including sparrow group size, maximum number of iterations and other parameters;
(4)更新麻雀的位置。计算每个种群麻雀的位置以获得当前的新位置；(4) Update the position of the sparrow. Calculate the position of each population of sparrows to obtain the current new position;
(5)计算适应值。SVM故障模型的分类精度是每个麻雀的当前拟合值，以及最大实时更新适应值。如果麻雀的当前适应值大于保存的适应度值，原始适应度值已更新；否则，原始适应度值保持不变。并保存最佳值；(5) Calculate the fitness value. The classification accuracy of the SVM fault model is the current fitting value of each sparrow, and the maximum real-time updated fitness value. If the sparrow's current fitness value is greater than the saved fitness value, the original fitness value has been updated; otherwise, the original fitness value remains unchanged. And save the best value;
(6)根据最佳适应值保存相应的SVM参数组合；(6) Save the corresponding SVM parameter combination according to the best fitness value;
(7)导入最佳参数组合使用支持向量机进行计算；(7) Import the best parameter combination and use support vector machine for calculation;
(8)判断终止条件。如果终止满足条件后，输出最佳参数组合及其相应的分类结果。如果不满足条件，将迭代次数增加1并返回到第4步。(8) Determine the termination conditions. If the termination condition is met, the best parameter combination and its corresponding classification result will be output. If the condition is not met, increase the number of iterations by 1 and return to step 4.
应说明的是，麻雀位置对应的是SVM里的惩罚因子C与核函数参数σ。It should be noted that the sparrow position corresponds to the penalty factor C and kernel function parameter σ in SVM.
实施例2Example 2
参照图4—7，为本发明的一个实施例，提供了一种一种基于SSA-SVM的GIS故障模式识别方法为了验证本发明的有益效果，通过与传统算法对比进行科学论证。Referring to Figures 4-7, an embodiment of the present invention provides a GIS fault mode identification method based on SSA-SVM. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out by comparing with traditional algorithms.
基于三浓度比值法，本实施例在三种特征组分比值的基本上新增其他几种主要分解特征气体组分之间的比值作为判断故障的特征量，部分数据如表3所 示。Based on the three-concentration ratio method, this embodiment adds the ratios between several other main decomposition characteristic gas components as characteristic quantities for judging faults on the basis of the three characteristic component ratios. Part of the data is shown in Table 3. Show.
表3不同故障下特征组分比值
Table 3 Characteristic component ratios under different faults
为检验所提出的模式识别方法对局部放电模式的准确率，从每类缺陷类型的样本集选取七十个个测试样本和三十个训练样本，并加以识别，测试结果如图4所示。In order to test the accuracy of the proposed pattern recognition method on partial discharge patterns, seventy test samples and thirty training samples were selected from the sample set of each defect type and identified. The test results are shown in Figure 4.
本实施例选用了鲸鱼优化算法(WOA)、粒子群优化算法(PSO)对SVM的参数进行优化并与SSA的优化效果进行比对。如图5所示为各个优化算法下的适应度曲线，可以看出鲸鱼算法在前期进行全局搜索，在后期进行局部搜索，极易进入局部最优的情况；而粒子群算法由于局部搜索功能较差，并不能很快的找寻出局部最优解，亦不能保证快速搜索出全局最优解。麻雀优化算法在加 入了侦查预警机制之后，明显降低了搜索所需要的时间，提高了对于样本的分类效果。由此可以判断，SSA-SVM对于建立GIS故障模式识别的诊断模型具有一定的意义。This embodiment uses the Whale Optimization Algorithm (WOA) and the Particle Swarm Optimization Algorithm (PSO) to optimize the parameters of SVM and compare the optimization effects with SSA. Figure 5 shows the fitness curves under each optimization algorithm. It can be seen that the whale algorithm performs global search in the early stage and local search in the later stage, and it is easy to enter the local optimal situation; while the particle swarm algorithm has a weak local search function. The difference is that it cannot quickly find the local optimal solution, nor can it guarantee that the global optimal solution can be quickly searched. Sparrow optimization algorithm is added After incorporating the investigation and early warning mechanism, the time required for search is significantly reduced and the classification effect of samples is improved. It can be judged from this that SSA-SVM has certain significance in establishing a diagnostic model for GIS fault mode recognition.
在相同条件下分别对WOA-SVM、PSO-SVM模型进行试验，分别得到如图6和图7所示的试验结果。由图4可知，在训练数据与测试数据相同的情况下，SSA-SVM正确判断出GIS故障类型的个数为116个，其故障模式识别正确率为96.7％，而由图6和图7可知，WOA-SVM、PSO-SVM对于GIS故障模式识别的正确率分别为90.8％和92.6％。结果表明，SSA-SVM对于GIS的故障模式识别具有更高的准确率，可以更为精确的判断出GIS的故障类型。The WOA-SVM and PSO-SVM models were tested under the same conditions, and the test results were obtained as shown in Figure 6 and Figure 7 respectively. As can be seen from Figure 4, when the training data and test data are the same, the number of GIS fault types correctly determined by SSA-SVM is 116, and its fault mode identification accuracy rate is 96.7%. As can be seen from Figures 6 and 7 , the accuracy rates of WOA-SVM and PSO-SVM for GIS fault mode recognition are 90.8% and 92.6% respectively. The results show that SSA-SVM has a higher accuracy in identifying GIS fault modes and can more accurately determine the fault type of GIS.
应说明的是，以上实施例仅用以说明本发明的技术方案而非限制，尽管参照较佳实施例对本发明进行了详细说明，本领域的普通技术人员应当理解，可以对本发明的技术方案进行修改或者等同替换，而不脱离本发明技术方案的精神和范围，其均应涵盖在本发明的权利要求范围当中。 It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solution of the present invention can be carried out. Modifications or equivalent substitutions without departing from the spirit and scope of the technical solution of the present invention shall be included in the scope of the claims of the present invention.

Claims (10)

  1. 一种基于SSA-SVM的GIS故障模式识别方法，其特征在于，包括：A GIS fault mode identification method based on SSA-SVM, which is characterized by including:
    建立气体绝缘变电站真实模型，导入故障下SF6各个分解产物的产气速率方程获取原始数据；Establish a real model of a gas-insulated substation, and introduce the gas production rate equation of each decomposition product of SF 6 under fault to obtain original data;
    基于麻雀搜索算法优化支持向量机的核函数参数和惩罚系数；Optimize the kernel function parameters and penalty coefficients of the support vector machine based on the Sparrow search algorithm;
    建立麻雀搜索算法优化下的支持向量机故障诊断模型，将数据输入诊断模型得到故障结果。A support vector machine fault diagnosis model optimized by the Sparrow search algorithm is established, and the data is input into the diagnosis model to obtain the fault results.
  2. 如权利要求1所述的基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述气体绝缘变电站真实模型是一个封闭容器，特征气体组分分子的扩散只能通过分子热力学运动来处理。The GIS fault mode identification method based on SSA-SVM according to claim 1, characterized in that: the real model of the gas insulated substation is a closed container, and the diffusion of characteristic gas component molecules can only be handled through molecular thermodynamic movement.
  3. 如权利要求2所述的一种基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述SF6气体的分解，包括：A GIS fault mode identification method based on SSA-SVM as claimed in claim 2, characterized in that: the decomposition of SF 6 gas includes:
    在故障条件下，受电子碰撞等因素影响H2O和O2会解离产生O、OH等具有高活性的粒子，低氟硫化物将与OH、O结合，生成HF、SOF4、SO2F2、S2OF10、SOF2及SO2等产物。Under fault conditions, H 2 O and O 2 will dissociate due to factors such as electron collision to produce highly active particles such as O and OH. Low fluorine sulfide will combine with OH and O to generate HF, SOF 4 and SO 2 Products such as F 2 , S 2 OF 10 , SOF 2 and SO 2 .
  4. 如权利要求1所述的一种基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述产气速率方程表示为：
    A GIS fault mode identification method based on SSA-SVM as claimed in claim 1, characterized in that: the gas production rate equation is expressed as:
    其中，RRMS为绝对产气速率，ci，1为第1次测得组分i的含量，ci，2为第2次测得组分i的含量，Δt为两次检测的时间间隔。Among them, R RMS is the absolute gas production rate, c i,1 is the content of component i measured for the first time, c i,2 is the content of component i measured for the second time, Δt is the time interval between two measurements .
  5. 如权利要求4所述的基于SSA-SVM的GIS故障模式识别方法，其特征在于：将GIS真实模型进行内部流体仿真得到原始数据，所述原始数据包括：The GIS fault mode identification method based on SSA-SVM as claimed in claim 4, characterized in that: performing internal fluid simulation on a real GIS model to obtain original data, and the original data includes:
    将SF6以及SO2F2、SOF2、SO2、CO2、CF4这5种特征气体的密度、导热率等参数导入到仿真软件；在气体绝缘变电站仿真模型里定义一个区域，在此区域内特征气体将按照特征速率方程产生特征气体；在固定温度以及压力的条件下，将不同故障缺陷下的产气速率方程导入此区域，然后扩散至整个气体绝缘变电站，得到各个缺陷下特征气体组分的原始数据。Import the density, thermal conductivity and other parameters of SF 6 and five characteristic gases such as SO 2 F 2 , SOF 2 , SO 2 , CO 2 and CF 4 into the simulation software; define a region in the gas insulated substation simulation model, where The characteristic gas in the area will produce characteristic gas according to the characteristic rate equation; under the conditions of fixed temperature and pressure, the gas production rate equation under different fault defects is introduced into this area, and then diffused to the entire gas insulated substation to obtain the characteristic gas under each defect. Raw data for components.
  6. 如权利要求1所述的基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述麻雀优化算法，包括： The GIS fault mode identification method based on SSA-SVM according to claim 1, characterized in that: the Sparrow optimization algorithm includes:
    在算法寻优的模拟仿真实验中，使用虚拟麻雀觅食，麻雀的位置表示为：
    In the simulation experiment of algorithm optimization, a virtual sparrow is used to forage, and the position of the sparrow is expressed as:
    其中，n为麻雀的数量，d为待优化变量的维数；Among them, n is the number of sparrows, and d is the dimension of the variable to be optimized;
    所有麻雀的适应度值表示为：
    The fitness value of all sparrows is expressed as:
    在算法迭代过程中发现者的个体位置更新公式为：
    During the algorithm iteration process, the discoverer’s individual position update formula is:
    其中，为第i个个体迭代t次时的第j维数，α为随机数，R2为报警值，ST为安全阈值,Q为服从正态分布的随机数，L为一个1×d的矩阵，其中该矩阵内所有元素都是1；in, is the j-th dimension of the i-th individual when iterating t times, α is a random number, R 2 is the alarm value, ST is the safety threshold, Q is a random number obeying the normal distribution, L is a 1×d matrix, All elements in this matrix are 1;
    加入者的位置更新公式为：
    The joiner’s location update formula is:
    其中，Xp为目前发现者的最佳位置，Xworst为当前全局的最差位置，A为一个1×d的矩阵，其中每个元素随机赋值为1或-1，并且A+＝AT(AAT)-1； Among them , _ (AA T ) -1 ;
  7. 如如权利要求6所述的基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述麻雀优化算法，还包括：The GIS fault mode identification method based on SSA-SVM as claimed in claim 6, characterized in that: the Sparrow optimization algorithm further includes:
    对比本次迭代个体适应度值与当前最佳适应度值，当算法陷入局部最优时，位置更新公式表示为：

    其中，Xbest为当前全局最优位置，β为步长控制参数，k为随机数，fi为当前麻雀的个体适应度值，fg为当前全局最佳的适应度值，fw为当前全局最差的适应度值，ε为最小的常数。    Comparing the individual fitness value of this iteration with the current best fitness value, when the algorithm falls into a local optimum, the position update formula is expressed as:

    Among them , _ The global worst fitness value, ε is the smallest constant.
  8. 如权利要求1所述的一种基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述支持向量机，包括：A GIS fault mode identification method based on SSA-SVM as claimed in claim 1, characterized in that: the support vector machine includes:
    在高维空间中构造一个最优分类超平面并进行分类,所述超平面函数表示为：
    Construct an optimal classification hyperplane in high-dimensional space and perform classification. The hyperplane function is expressed as:
    其中，为映射函数，ω为最优分类超平面的法向量，b为分类阈值；in, is the mapping function, ω is the normal vector of the optimal classification hyperplane, and b is the classification threshold;
    引入松弛变量ξ，使每一个硬间隔支持向量机对于函数距离的使用满足:Introduce the slack variable ξ so that the use of functional distance for each hard-margin support vector machine satisfies:
    yi(ω·xi+b)≥1-Kξi(1≤i≤m)y i (ω·x i +b)≥1-Kξ i (1≤i≤m)
    其中，ξi≥0；Among them, ξ i ≥ 0;
    此时，超平面函数表示为：
    At this time, the hyperplane function is expressed as:
    其中，C为惩罚参数；Among them, C is the penalty parameter;
  9. 如权利要求1所述的一种基于SSA-SVM的GIS故障模式识别方法，其特征在于：所述支持向量机，还包括：A GIS fault mode identification method based on SSA-SVM as claimed in claim 1, characterized in that: the support vector machine further includes:
    将超平面的寻优问题归结为二次规划问题，表示为：
    The hyperplane optimization problem is reduced to a quadratic programming problem, expressed as:
    基于拉格朗日乘子算法，最优超平面函数表示为：
    Based on the Lagrange multiplier algorithm, the optimal hyperplane function is expressed as:
    其中，SV为支持向量；Among them, SV is the support vector;
    核函数表示为：
    The kernel function is expressed as:
  10. 如权利要求7或9所述的一种基于SSA-SVM的GIS故障模式识别方法，其特征在于：建立麻雀搜索算法优化下的支持向量机故障诊断模型，包括：A GIS fault mode identification method based on SSA-SVM according to claim 7 or 9, characterized in that: establishing a support vector machine fault diagnosis model optimized by the Sparrow search algorithm, including:
    对原始数据进行预处理并将数据导入算法；选择训练集和测试集，根据研究情况导入数据；初始化麻雀搜索算法和支持向量机的参数；更新麻雀的位置；计算每个种群麻雀的位置以获得当前的新位置；计算适应值，并保存最佳值；根据最佳适应值保存相应的支持向量机参数组合；导入最佳参数组合使用支持向量机进行计算，判断终止条件，得到GIS故障的模式代码。 Preprocess the original data and import the data into the algorithm; select the training set and test set, and import the data according to the research situation; initialize the parameters of the sparrow search algorithm and support vector machine; update the position of the sparrow; calculate the position of each population of sparrows to obtain The current new position; calculate the fitness value and save the best value; save the corresponding support vector machine parameter combination according to the best fitness value; import the best parameter combination and use the support vector machine for calculation, determine the termination condition, and obtain the GIS failure mode code.
PCT/CN2023/073989 2022-07-22 2023-01-31 Ssa-svm-based gis fault mode recognition method WO2024016623A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210872035.9A CN115407140A (en) 2022-07-22 2022-07-22 GIS fault mode identification method based on SSA-SVM
CN202210872035.9 2022-07-22

Publications (1)

Publication Number Publication Date
WO2024016623A1 true WO2024016623A1 (en) 2024-01-25

Family

ID=84157859

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/073989 WO2024016623A1 (en) 2022-07-22 2023-01-31 Ssa-svm-based gis fault mode recognition method

Country Status (2)

Country Link
CN (1) CN115407140A (en)
WO (1) WO2024016623A1 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115407140A (en) * 2022-07-22 2022-11-29 贵州电网有限责任公司 GIS fault mode identification method based on SSA-SVM

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102498445A (en) * 2009-09-17 2012-06-13 西门子公司 Supervised fault learning using rule-generated samples for machine condition monitoring
CN102778555A (en) * 2012-08-06 2012-11-14 国网技术学院 Method for predicting concentration of gas dissolved in transformer oil
US20190056446A1 (en) * 2017-08-21 2019-02-21 Delta-X Research Inc. Method and system for assessment of fault severity, risk exposure, and gassing status for liquid-filled high-voltage apparatus
CN109918720A (en) * 2019-01-23 2019-06-21 广西大学 Diagnosis Method of Transformer Faults based on krill group's Support Vector Machines Optimized
CN110689068A (en) * 2019-09-25 2020-01-14 贵州电网有限责任公司 Transformer fault type diagnosis method based on semi-supervised SVM
CN112733603A (en) * 2020-12-11 2021-04-30 江苏大学 Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM
CN114415004A (en) * 2021-12-14 2022-04-29 贵州电网有限责任公司 Simulation true type GIS equipment fault characteristic gas diffusion method
CN114743089A (en) * 2021-12-22 2022-07-12 宁波送变电建设有限公司 Image recognition GIS fault diagnosis device and method based on SSA-SVM
CN115407140A (en) * 2022-07-22 2022-11-29 贵州电网有限责任公司 GIS fault mode identification method based on SSA-SVM

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102498445A (en) * 2009-09-17 2012-06-13 西门子公司 Supervised fault learning using rule-generated samples for machine condition monitoring
CN102778555A (en) * 2012-08-06 2012-11-14 国网技术学院 Method for predicting concentration of gas dissolved in transformer oil
US20190056446A1 (en) * 2017-08-21 2019-02-21 Delta-X Research Inc. Method and system for assessment of fault severity, risk exposure, and gassing status for liquid-filled high-voltage apparatus
CN109918720A (en) * 2019-01-23 2019-06-21 广西大学 Diagnosis Method of Transformer Faults based on krill group's Support Vector Machines Optimized
CN110689068A (en) * 2019-09-25 2020-01-14 贵州电网有限责任公司 Transformer fault type diagnosis method based on semi-supervised SVM
CN112733603A (en) * 2020-12-11 2021-04-30 江苏大学 Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM
CN114415004A (en) * 2021-12-14 2022-04-29 贵州电网有限责任公司 Simulation true type GIS equipment fault characteristic gas diffusion method
CN114743089A (en) * 2021-12-22 2022-07-12 宁波送变电建设有限公司 Image recognition GIS fault diagnosis device and method based on SSA-SVM
CN115407140A (en) * 2022-07-22 2022-11-29 贵州电网有限责任公司 GIS fault mode identification method based on SSA-SVM

Also Published As

Publication number Publication date
CN115407140A (en) 2022-11-29

Similar Documents

Publication Publication Date Title
Zhang et al. A fault diagnosis model of power transformers based on dissolved gas analysis features selection and improved krill herd algorithm optimized support vector machine
Tang et al. Partial discharge recognition through an analysis of SF 6 decomposition products part 2: Feature extraction and decision tree-based pattern recognition
Liu et al. Moisture diagnosis of transformer oil-immersed insulation with intelligent technique and frequency-domain spectroscopy
Qiu et al. A prediction method for breakdown voltage of typical air gaps based on electric field features and support vector machine
WO2024016623A1 (en) Ssa-svm-based gis fault mode recognition method
Niu et al. Study of degradation of fuel cell stack based on the collected high-dimensional data and clustering algorithms calculations
Wu et al. Triangle fault diagnosis method for SF 6 gas-insulated equipment
Zang et al. Adaptive method for harmonic contribution assessment based on hierarchical K‐means clustering and Bayesian partial least squares regression
Wu et al. Defect recognition and condition assessment of epoxy insulators in gas insulated switchgear based on multi-information fusion
Yang et al. Fault diagnosis in gas insulated switchgear based on genetic algorithm and density-based spatial clustering of applications with noise
CN111401476B (en) Transient state safety evaluation method based on boundary region importance sampling and kernel vector machine
Sun et al. Prediction of the electrical strength and boiling temperature of the substitutes for greenhouse gas SF₆ using neural network and random forest
Zhang et al. Monitoring the metering performance of an electronic voltage transformer on-line based on cyber-physics correlation analysis
CN112380676A (en) Multi-energy system digital twin data stream modeling and compressing method
Zeng et al. Reconstructing and extracting information on SF 6 decomposition characteristic components induced by partial overthermal fault in GIE
Li et al. Partial discharge fault diagnosis based on zernike moment and improved bacterial foraging optimization algorithm
Cao et al. SF6 positive DC partial discharge decomposition components under four typical insulation defects
Yi et al. LSTM neural networks with attention mechanisms for accelerated prediction of charge density at onset condition of DC corona discharge
CN115859594A (en) Health evaluation algorithm of power transformation equipment based on hierarchical analysis
Liu et al. Research on GIS partial discharge pattern recognition based on deep residual network and transfer learning in ubiquitous power internet of things context
CN102495179B (en) Insulating gas detection system and data managing method thereof
Zhong et al. Effects of voltage forms, pressure, and adsorbent on the SF6 decomposition characteristics under corona discharge
Zhang et al. Research on transformer fault diagnosis: Based on improved firefly algorithm optimized LPboost–classification and regression tree
Zhang et al. GIS Fault Diagnosis Method Based on The Identification of SF 6 Gas Decomposition Product Fault Characteristics
De La Hoz et al. Impact assessment of clearance in corona testing for a high‐voltage substation connector set using fem

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23841701

Country of ref document: EP

Kind code of ref document: A1