WO2021103496A1 - 基于异常振动的气体绝缘组合开关设备机械故障诊断方法 - Google Patents

基于异常振动的气体绝缘组合开关设备机械故障诊断方法 Download PDF

Info

Publication number
WO2021103496A1
WO2021103496A1 PCT/CN2020/096669 CN2020096669W WO2021103496A1 WO 2021103496 A1 WO2021103496 A1 WO 2021103496A1 CN 2020096669 W CN2020096669 W CN 2020096669W WO 2021103496 A1 WO2021103496 A1 WO 2021103496A1
Authority
WO
WIPO (PCT)
Prior art keywords
vibration signal
vibration
gis
abnormal
operating mechanism
Prior art date
Application number
PCT/CN2020/096669
Other languages
English (en)
French (fr)
Inventor
屈斌
张利
陈荣
周连升
甘智勇
张弛
李国豪
何金
王坤
王梓越
王建
范巍
Original Assignee
国网天津市电力公司电力科学研究院
国网天津市电力公司
国家电网有限公司
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 国网天津市电力公司电力科学研究院, 国网天津市电力公司, 国家电网有限公司 filed Critical 国网天津市电力公司电力科学研究院
Priority to US17/117,106 priority Critical patent/US20210167584A1/en
Publication of WO2021103496A1 publication Critical patent/WO2021103496A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M13/00—Testing of machine parts
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
    • G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction

Definitions

  • the invention belongs to the technical field of intelligent substations, in particular to a method for diagnosing mechanical faults of gas-insulated switchgear based on abnormal vibration.
  • GIS still has some disadvantages.
  • the internal components of the GIS equipment are all encapsulated in a metal shell, once the GIS fails, it is difficult to determine the specific location of the fault, and because the GIS structure is complex and the internal components are difficult to disassemble, it is difficult to achieve on-site maintenance and the maintenance work is complicated. , The power outage takes a long time for maintenance. Therefore, it is necessary to monitor the operation of GIS in real time to improve the reliability of power supply.
  • the factors that usually cause abnormal vibration of GIS mechanical failure include the following aspects: 1Partial discharge inside the GIS; 2Vibration generated by operating mechanisms such as circuit breakers or isolating switches during operation; 3Abnormal vibration of stationary mechanisms such as loose and deformed GIS cylinder fasteners 4Under the excitation of electromotive force, some parts may produce abnormal vibration; 5When the vibration frequency between adjacent equipment of GIS is the same, resonance will occur, causing abnormal vibration of GIS.
  • the application number CN201810245653 is a gas-insulated combination switchgear mechanical fault online diagnosis method, which discloses a time-frequency vibration signal based on the S-transformation of equalized energy based on the opening and closing operation of the circuit breaker as the excitation source Plane segmentation ideas, and extract the energy value of each feature sub-interval under normal and fault states, and use SVM optimized by genetic algorithm to classify and identify faults.
  • the patent does not coordinate the diagnosis of mechanical faults under the action of other excitation sources (such as isolation switch excitation, grounding switch excitation and electromagnetic force excitation).
  • the purpose of the present invention is to overcome the shortcomings of the prior art and provide a coordinated plan for the vibration signals under the action of four kinds of excitation forces: circuit breaker, isolating switch, grounding switch, and electromagnetic force, and realize GIS machinery by extracting wavelet packet characteristic entropy and BP neural network.
  • Fault diagnosis is a method for mechanical fault diagnosis of gas-insulated switchgear based on abnormal vibration.
  • the vibration signal includes a non-stationary vibration signal excited by the action of an operating mechanism and a steady vibration signal excited by the electromagnetic force of the operating state, wherein the operating mechanism includes a circuit breaker Device, isolating switch, grounding switch;
  • the abnormal vibration signal is judged to enter the feature extraction; the abnormal vibration signal is processed by the three-layer wavelet packet decomposition, and the wavelet packet feature entropy of each frequency band of the third layer is obtained, and the feature entropy vector is constructed; according to the feature The entropy vector constructs the data set, and uses BP neural network for fault pattern recognition.
  • the vibration signal includes the non-stationary vibration signal excited by the action of the operating mechanism and the steady vibration signal excited by the electromagnetic force in the operating state, including the following steps:
  • step (6) adopts BP neural network for fault pattern recognition.
  • the start time of collecting non-stationary vibration signals excited by the operating mechanism in the vibration signal library in steps (1) and (2) is the moment when the operating mechanism starts, and the end time is 0.5 seconds after the completion of the operation of the operating mechanism.
  • the acquisition time of the steady vibration signal excited by the electromagnetic force in the operating state described in steps (1) and (2) is 2 seconds, and the sampling frequency is 20 kHz.
  • step (5) the comparison of E S and E S0 in step (5) is performed under the same operating mechanism excitation source.
  • the excitation source of the operating mechanism includes circuit breaker excitation, isolating switch excitation and grounding switch excitation.
  • step (5) the comparison between E Q and E Q0 in step (5) is performed under the same electromagnetic force excitation source.
  • step (6) the wavelet packet decomposition processing described in step (6) is: selecting the dbl0 wavelet of the Daubechies wavelet series to perform three-layer wavelet packet decomposition and transformation.
  • the vibration signal of the normal state GIS is used as the reference, after the vibration signal is judged to be abnormal, it enters the feature extraction; the abnormal vibration signal is processed by wavelet packet decomposition, the wavelet packet feature entropy of each frequency band is obtained, and the feature entropy vector is constructed; according to the feature entropy
  • the vector is used to construct the data set, and the BP neural network is used for fault pattern recognition.
  • the invention utilizes wavelet packet energy entropy to have the advantages of multi-resolution, and has obvious advantages in time-frequency localization. It is very suitable for processing stable and non-stationary vibration signals, and at the same time co-ordinates the four types of circuit breakers, isolating switches, grounding switches, and electromagnetic forces.
  • the vibration signal under the excitation force has comprehensive data information, high diagnostic reliability, and accurate calculation results.
  • Figure 1 is a working flow chart of the present invention
  • Figure 2 is a time-domain diagram of the vibration signal of the GIS circuit breaker's transmission rod under normal and jammed states; among them, (a) is normal and (b) is jammed;
  • Fig. 3 is a wavelet packet energy entropy spectrum diagram in the normal or jammed state of Fig. 2;
  • Figure 4 is a neural network performance graph containing three results of training, verification, and testing;
  • Figure 5 shows the fault classification results under four operations of training, verification, testing, and overall data.
  • the innovation of the present invention is that: the vibration signal contains the non-stationary vibration signal excited by the action of the operating mechanism and the electromagnetic force excited by the operating state.
  • Stable vibration signal the operating mechanism includes circuit breaker, isolating switch, grounding switch; based on the vibration signal of normal GIS, after judging the abnormal vibration signal, enter the feature extraction; using 3-layer wavelet packet decomposition to process the abnormal vibration signal to obtain
  • the wavelet packet feature entropy of each frequency band in the third layer is used to construct the feature entropy vector; the data set is constructed according to the feature entropy vector, and the BP neural network is used for fault pattern recognition.
  • the vibration signal includes the non-stationary vibration signal excited by the operation of the operating mechanism and the steady vibration signal excited by the electromagnetic force in the operating state.
  • the vibration signal of the GIS includes the non-stationary vibration signal excited by the operation of the operating mechanism and the steady vibration signal excited by the electromagnetic force in the operating state.
  • the action signal of the operating mechanism is a non-stationary signal, that is, the vibration energy changes greatly with time; the energy of the vibration excited by the electromagnetic force is relatively stable, which is a stable signal.
  • Signal acquisition needs to use three kinds of instruments: vibration acceleration sensor, signal acquisition instrument, and computer.
  • the installation position of the specific sensor and the connection with the signal acquisition instrument and computer are all existing technologies.
  • the above diagnosis method includes the following steps:
  • the Hilbert transform is used to extract the envelope curve of the vibration signal, and the area of the envelope of the non-stationary vibration signal is calculated.
  • the stable vibration signal uses FFT to extract the vibration energy of 100 Hz.
  • the signals in the vibration signal library of the normal GIS are measured in advance.
  • the signals in the normal state are called and compared with the monitored signals.
  • step (6) adopts BP neural network for fault pattern recognition.
  • the start time of collecting non-stationary vibration signals excited by the operating mechanism in the vibration signal library in steps (1) and (2) is the moment when the operating mechanism starts, and the end time is 0.5 seconds after the completion of the operating mechanism.
  • the acquisition time of the steady vibration signal excited by the electromagnetic force in the operating state described in steps (1) and (2) is 2 seconds, and the sampling frequency is 20 kHz.
  • the comparison of E S and E S0 in step (5) is carried out under the same excitation source of the operating mechanism.
  • the operating mechanism excitation source includes circuit breaker excitation, isolating switch excitation and grounding switch excitation.
  • the comparison between E Q and E Q0 in step (5) is performed under the same electromagnetic excitation source.
  • the same meaning of the excitation source of the operating mechanism means: the vibration signal in the normal GIS vibration signal library is the circuit breaker excitation source, then the vibration signal in the monitored GIS vibration signal library is also the circuit breaker excitation source, that is, two
  • the comparison of the envelope area is only comparable when the excitation source is the same.
  • the same electromagnetic force excitation source also has this meaning.
  • the wavelet packet decomposition process described in step (6) is: selecting the dbl0 wavelet of the Daubechies wavelet series to perform three-layer wavelet packet decomposition and transformation.
  • the simulation uses the wavelet packet energy entropy method to process the data obtained from the circuit breaker jamming experiment, and the obtained vibration signal measured diagram is shown in Figure 2.
  • the excitation source of Fig. 2 a and b is the excitation of the circuit breaker in the GIS.
  • the difference between the two states is calculated. After the jam fault occurs, the vibration envelope area is significantly reduced, reaching the 5% threshold. Then, using the wavelet packet energy entropy method to process the data, the characteristic entropy vector under the jammed fault of the circuit breaker is shown in Table 1 below.
  • Table 1 Characteristic entropy vector table of circuit breaker jamming fault
  • the vibration signal of the normal state GIS is used as the reference, after the vibration signal is judged to be abnormal, it enters the feature extraction; the abnormal vibration signal is processed by wavelet packet decomposition, the wavelet packet feature entropy of each frequency band is obtained, and the feature entropy vector is constructed; according to the feature entropy
  • the vector is used to construct the data set, and the BP neural network is used for fault pattern recognition.
  • the invention utilizes wavelet packet energy entropy to have the advantages of multi-resolution, and has obvious advantages in time-frequency localization. It is very suitable for processing stable and non-stationary vibration signals, and at the same time co-ordinates the four types of circuit breakers, isolating switches, grounding switches, and electromagnetic forces.
  • the vibration signal under the excitation force has comprehensive data information, high diagnostic reliability, and accurate calculation results.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Testing Electric Properties And Detecting Electric Faults (AREA)
  • Arc-Extinguishing Devices That Are Switches (AREA)

Abstract

本发明涉及一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，首先以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用小波包分解处理异常振动信号，获取各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。本发明利用小波包能量熵具有多分辨率的优点，时频局部化的刻画优势明显，非常适用于处理平稳和非平稳振动信号，同时统筹断路器、隔离开关、接地开关、以及电磁力四种激振力作用下的振动信号，数据信息全面，诊断可靠性高，计算结果精确。

Description

基于异常振动的气体绝缘组合开关设备机械故障诊断方法 技术领域
本发明属于智能变电站技术领域，尤其是一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法。
背景技术
GIS设备自20世纪60年代实用化以来，已经广泛运用于世界各地。当前在国内，高压断路器使用SF6气体来代替绝缘油和空气已成为主流，新建或者已经改造的220KV及以上等级变压站基本全部采用开关设备。截止到2000年前，全国的220KV及以上电压等级断路器共12162台，其中，100％的500KV开关设备和接近100％的330KV开关设备为断路器和GIS；而220KV电压等级的断路器和GIS占装运开关设备的50％～60％。
尽管具有很多优点，GIS也依旧存在着一些缺点。例如，因为GIS设备的内部元件全部都封装在金属壳体内，一旦GIS发生故障，很难确定故障具体发生位置，且由于GIS结构复杂，内部元件难以拆卸，所以很难实现现场维护，检修工作繁杂，停电检修时间长。因此，有必要对GIS的运行进行实时监测，从而提高供电的可靠性。
GIS在运行时，运行设备会产生一定的振动，但是这些振动都属于正常振动。当GIS内部元件出现缺陷时就会产生有别于正常振动时的异常振动，从而引发机械故障。通常引起GIS机械故障异常振动的因素有以下几个方面：①GIS内部出现局部放电；②断路器或隔离开关等操作机构在操作时产生的振动；③GIS筒体紧固件松动变形等静止机构异常振动；④在电动力的激励下，某些部件可能产生异常振动；⑤当GIS相邻设备间振动频率相同时，会发生共振，使得GIS出现异常振动。
经检索发现，申请号为CN201810245653的一种气体绝缘组合开关设备的机械故障在线诊断方法，公开了一种以断路器的分合闸操作为激振源，基于S变换 能量均等的振动信号时频平面分割思想，并提取正常与故障状态下的每个特征子区间的能量值，采用遗传算法优化的SVM对故障进行分类识别。但是该专利未统筹其他激振源(如隔离开关激振、接地开关激振和电磁力激振)作用下的机械故障诊断问题。
发明内容
本发明的目的在于克服现有技术的不足，提供统筹断路器、隔离开关、接地开关、以及电磁力4种激振力作用下的振动信号，通过提取小波包特征熵和BP神经网络实现GIS机械故障诊断的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法。
本发明所采用的具体技术方案如下：
一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，其中操动机构包括断路器、隔离开关、接地开关；
以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用3层小波包分解处理异常振动信号，获取第3层各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。
再有，振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，包括以下步骤：
⑴建立正常GIS的振动信号库；
⑵建立被监测GIS的振动信号库；
⑶计算步骤⑴的振动信号库中的操动机构激发的非平稳振动信号的包络面积E S0，计算步骤⑴的运行状态电磁力激发的平稳振动信号的振动能量E Q0；
计算步骤⑵的振动信号库中的操动机构激发的非平稳振动信号的包络面积E S；计算步骤⑵的运行状态电磁力激发的平稳振动信号的振动能量E Q；
⑷分别对E S和E S0、E Q和E Q0进行比较；
⑸三个操动机构的E S均满足|E S-E S0|/E S0<5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0<5％，则GIS工作 正常；
任意两个操动机构的E S均满足|E S-E S0|/E S0≥5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0≥5％，则判定振动信号异常，进入下一步；
⑹对被监测GIS的振动信号进行小波包分解处理，并构造特征熵向量；
⑺对步骤⑹的结果采用BP神经网络进行故障模式识别。
再有，步骤⑴和⑵所述振动信号库中的操动机构激发的非平稳振动信号的采集起始时间为操动机构动作开始瞬间，结束时间为操动机构动作完毕后延时0.5秒。
再有，步骤⑴和⑵所述的运行状态电磁力激发的平稳振动信号的采集时长为2秒，采样频率为20kHz。
再有，步骤⑸中的E S与E S0的比较为相同的操动机构激振源下进行的。
再有，所述操动机构激振源包括断路器激振、隔离开关激振和接地开关激振。
再有，步骤⑸中的E Q与E Q0的比较为相同的电磁力激振源下进行的。
再有，步骤⑹所述的小波包分解处理为：选用Daubechies小波系列的dbl0小波进行三层小波包分解变换。
本发明的优点和有益效果是：
本发明中，以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用小波包分解处理异常振动信号，获取各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。本发明利用小波包能量熵具有多分辨率的优点，时频局部化的刻画优势明显，非常适用于处理平稳和非平稳振动信号，同时统筹断路器、隔离开关、接地开关、以及电磁力四种激振力作用下的振动信号，数据信息全面，诊断可靠性高，计算结果精确。
附图说明
图1是本发明的工作流程图；
图2是GIS的断路器的传动杆正常和卡涩两种状态下的振动信号时域图；其中，(a)为正常，(b)为卡涩；
图3是图2的正常或卡涩两种状态时的小波包能量熵谱图；
图4是包含训练、验证、测试三个结果的神经网络性能图；
图5是训练、验证、测试、全体数据四种运算下的故障分类结果。
具体实施方式
本发明通过以下实施例进一步详述，但本实施例所叙述的技术内容是说明性的，而不是限定性的，不应依此来局限本发明的保护范围。
一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，如图1～5所示，本发明的创新在于：振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，其中操动机构包括断路器、隔离开关、接地开关；以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用3层小波包分解处理异常振动信号，获取第3层各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。
更优选的方案是：
振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，GIS的振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号。操动机构动作信号就是非平稳信号，即振动能量随着时间有较大变化；电磁力激发的振动其能量比较平稳，为平稳信号。
信号采集需要用到振动加速度传感器、信号采集仪、计算机三种仪器，具体传感器的安装位置以及与信号采集仪、计算机之间的连接均为现有技术。
上述诊断方法包括以下步骤：
⑴建立正常GIS的振动信号库。
⑵建立被监测GIS的振动信号库。
⑶非平稳振动信号，采用Hilbert变换提取振动信号的包络曲线，计算非平稳振动信号包络的面积。平稳振动信号采用FFT提取100Hz的振动能量。
计算步骤⑴的振动信号库中的操动机构激发的非平稳振动信号的包络面积E S0，计算步骤⑴的运行状态电磁力激发的平稳振动信号的振动能量E Q0。
计算步骤⑵的振动信号库中的操动机构激发的非平稳振动信号的包络面积 E S；计算步骤⑵的运行状态电磁力激发的平稳振动信号的振动能量E Q。
⑷分别对E S和E S0、E Q和E Q0进行比较。
正常GIS的振动信号库中的信号是提前测量好的，在进行监测的时候调取正常状态的信号，与被监测的信号进行比对。
⑸三个操动机构的E S均满足|E S-E S0|/E S0<5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0<5％，则GIS工作正常；
任意两个操动机构的E S均满足|E S-E S0|/E S0≥5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0≥5％，则判定振动信号异常，进入下一步；
⑹对被监测GIS的振动信号进行小波包分解处理，并构造特征熵向量；
⑺对步骤⑹的结果采用BP神经网络进行故障模式识别。
其中，步骤⑴和⑵所述振动信号库中的操动机构激发的非平稳振动信号的采集起始时间为操动机构动作开始瞬间，结束时间为操动机构动作完毕后延时0.5秒。步骤⑴和⑵所述的运行状态电磁力激发的平稳振动信号的采集时长为2秒，采样频率为20kHz。
步骤⑸中的E S与E S0的比较为相同的操动机构激振源下进行的。操动机构激振源包括断路器激振、隔离开关激振和接地开关激振。步骤⑸中的E Q与E Q0的比较为相同的电磁力激振源下进行的。操动机构激振源相同的含义是指：正常GIS振动信号库中的振动信号为断路器激振源，那么被监测GIS振动信号库中的振动信号也同样是断路器激振源，即两个包络面积的比较在激振源相同的情况下才有可比性。同样的电磁力激振源也是这个含义。
步骤⑹所述的小波包分解处理为：选用Daubechies小波系列的dbl0小波进行三层小波包分解变换。
实施例
以西安西开高压电气股份有限公司(西安高压开关厂)生产的126kV的SF6气体绝缘金属封闭开关设备(GIS)，型号为120-SFMT-32CA，模拟断路器传动杆卡涩故障。模拟用小波包能量熵法对断路器卡涩实验得到的数据进行处理，得到 的振动信号实测图见附图2。图2的a、b的激振源为GIS中的断路器激振。
首先，根据采集的正常和被监测的数据，计算两种状态的差异，卡涩故障发生后振动包络面积明显减小，达到5％的阈值。继而，采用小波包能量熵法处理数据得到的断路器卡涩故障下的特征熵向量如下表1所示。
表1：断路器卡涩故障的特征熵向量表
Figure PCTCN2020096669-appb-000001
对以上两种情况画出小波包能量熵谱图如附图3所示，图3是小波包的能量熵谱图，是提取的特征信息。
将经过小波包能量熵法提取出的特征熵向量放入MATLAB中的神经网络工具箱中进行训练。将隐含层神经元数目设为10，采用L-M优化算法进行处理，得到的各项性能见附图4。
从附图4可以看出，误差曲线具有较好的收敛速度(图4包含训练、验证、测试三个结果，图中可以看出运算的误差收敛效果较好)。从附图5可以看出，拟合效果较好。从中发现，BP网络能够较好地对GIS机械故障异常振动信号进行分类识别(图5的曲线表示训练、验证、测试、全体数据四种运算下的故障分类结果，均达到95％以上的准确率)。同时，这也证明了小波包能量熵法在GIS机械故障异常振动诊断上的有效性。
本发明中，以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用小波包分解处理异常振动信号，获取各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。本发明利用小波包能量熵具有多分辨率的优点，时频局部化的刻画优势明显，非常适用于处理平稳和非平稳振动信号，同时统筹断路器、隔离开关、接地开关、以及电磁力四种激振力作用下的振动信号，数据信息全面，诊断可靠性高， 计算结果精确。

Claims (8)

  1. 一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，其中操动机构包括断路器、隔离开关、接地开关；
    以正常状态GIS的振动信号为基准，判定振动信号异常后，进入特征提取；采用3层小波包分解处理异常振动信号，获取第3层各频段的小波包特征熵，构造特征熵向量；根据特征熵向量构造数据集，采用BP神经网络进行故障模式识别。
  2. 根据权利要求1所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：振动信号包含操动机构动作激发的非平稳振动信号和运行状态电磁力激发的平稳振动信号，包括以下步骤：
    ⑴建立正常GIS的振动信号库；
    ⑵建立被监测GIS的振动信号库；
    ⑶计算步骤⑴的振动信号库中的操动机构激发的非平稳振动信号的包络面积E S0，计算步骤⑴的运行状态电磁力激发的平稳振动信号的振动能量E Q0；
    计算步骤⑵的振动信号库中的操动机构激发的非平稳振动信号的包络面积E S；计算步骤⑵的运行状态电磁力激发的平稳振动信号的振动能量E Q；
    ⑷分别对E S和E S0、E Q和E Q0进行比较；
    ⑸三个操动机构的E S均满足|E S-E S0|/E S0<5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0<5％，则GIS工作正常；
    任意两个操动机构的E S均满足|E S-E S0|/E S0≥5％且被监测GIS的100Hz振动能量E Q与正常GIS的100Hz振动能量E Q0之间满足|E Q-E Q0|/E Q0≥5％，则判定振动信号异常，进入下一步；
    ⑹对被监测GIS的振动信号进行小波包分解处理，并构造特征熵向量；
    ⑺对步骤⑹的结果采用BP神经网络进行故障模式识别。
  3. 根据权利要求2所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：步骤⑴和⑵所述振动信号库中的操动机构激发的非 平稳振动信号的采集起始时间为操动机构动作开始瞬间，结束时间为操动机构动作完毕后延时0.5秒。
  4. 根据权利要求2所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：步骤⑴和⑵所述的运行状态电磁力激发的平稳振动信号的采集时长为2秒，采样频率为20kHz。
  5. 根据权利要求2所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：步骤⑸中的E S与E S0的比较为相同的操动机构激振源下进行的。
  6. 根据权利要求5所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：所述操动机构激振源包括断路器激振、隔离开关激振和接地开关激振。
  7. 根据权利要求2所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：步骤⑸中的E Q与E Q0的比较为相同的电磁力激振源下进行的。
  8. 根据权利要求2所述的一种基于异常振动的气体绝缘组合开关设备机械故障诊断方法，其特征在于：步骤⑹所述的小波包分解处理为：选用Daubechies小波系列的dbl0小波进行三层小波包分解变换。
PCT/CN2020/096669 2019-11-29 2020-06-17 基于异常振动的气体绝缘组合开关设备机械故障诊断方法 WO2021103496A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/117,106 US20210167584A1 (en) 2019-11-29 2020-12-09 Gis mechanical fault diagnosis method and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911205810.X 2019-11-29
CN201911205810.XA CN110926778B (zh) 2019-11-29 2019-11-29 基于异常振动的气体绝缘组合开关设备机械故障诊断方法

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/117,106 Continuation-In-Part US20210167584A1 (en) 2019-11-29 2020-12-09 Gis mechanical fault diagnosis method and device

Publications (1)

Publication Number Publication Date
WO2021103496A1 true WO2021103496A1 (zh) 2021-06-03

Family

ID=69847040

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/096669 WO2021103496A1 (zh) 2019-11-29 2020-06-17 基于异常振动的气体绝缘组合开关设备机械故障诊断方法

Country Status (2)

Country Link
CN (1) CN110926778B (zh)
WO (1) WO2021103496A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113705637A (zh) * 2021-08-12 2021-11-26 西安交通大学 一种断路器机械故障检测方法、系统、设备及可读存储介质
CN113933038A (zh) * 2021-10-11 2022-01-14 国网山东省电力公司菏泽供电公司 一种基于多源信息融合算法的gis机械性能监测方法
CN114034381A (zh) * 2021-11-12 2022-02-11 广东电网有限责任公司江门供电局 一种基于小波信息熵的配电变压器振动提取方法及系统
CN114743089A (zh) * 2021-12-22 2022-07-12 宁波送变电建设有限公司 一种基于ssa-svm的图像识别gis故障诊断装置及方法
CN115510925A (zh) * 2022-11-17 2022-12-23 中国人民解放军国防科技大学 一种机械故障诊断方法、装置、介质
CN116415509A (zh) * 2023-06-12 2023-07-11 华东交通大学 一种轴承性能退化预测方法、系统、计算机及存储介质
CN114743089B (zh) * 2021-12-22 2024-06-04 宁波送变电建设有限公司 一种基于ssa-svm的图像识别gis故障诊断装置及方法

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110926778B (zh) * 2019-11-29 2021-08-20 国网天津市电力公司电力科学研究院 基于异常振动的气体绝缘组合开关设备机械故障诊断方法
CN111638051B (zh) * 2020-06-15 2022-05-27 山东电工电气集团有限公司 一种刀闸状态在线监测及故障预判方法
CN112710972A (zh) * 2020-12-18 2021-04-27 云南电网有限责任公司丽江供电局 Gis触头接触不良故障诊断方法、装置、系统及存储介质
CN113408458B (zh) * 2021-06-29 2023-10-17 北京交通大学 一种基于振动信号特征熵融合的断路器故障诊断方法
CN114167270B (zh) * 2021-11-30 2023-09-19 广东电网有限责任公司 一种基于边缘计算的刀闸状态识别系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102736027A (zh) * 2012-07-18 2012-10-17 南京因泰莱配电自动化设备有限公司 一种基于断路器动特性测试仪的断路器故障诊断方法
CN106574863A (zh) * 2014-08-21 2017-04-19 湖原大学校产学协力团 电力设备声音诊断系统
CN109839263A (zh) * 2019-02-26 2019-06-04 北京航空航天大学 一种基于振动信息的gis特征提取及机械缺陷诊断方法
CN110132565A (zh) * 2019-05-28 2019-08-16 河海大学 一种基于小波包和神经网络结合的oltc故障诊断方法
CN110926778A (zh) * 2019-11-29 2020-03-27 国网天津市电力公司电力科学研究院 基于异常振动的气体绝缘组合开关设备机械故障诊断方法

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100380308B1 (ko) * 2000-12-12 2003-04-16 한국전기연구원 신경회로망을 이용한 동기 차단기의 개폐 제어장치
KR100853725B1 (ko) * 2007-06-20 2008-08-22 (주) 피에스디테크 Ｐｒｐｓ 알고리즘을 이용한 가스절연부하개폐장치의부분방전 원인분석 방법 및 그 장치
CN101404408B (zh) * 2008-11-13 2011-08-31 昆明理工大学 利用相对能量的配网缆-线混合线路故障选线方法
CN103308292B (zh) * 2013-05-16 2015-07-29 国家电网公司 基于振动信号分析的真空断路器机械状态检测方法
CN104111108A (zh) * 2014-07-16 2014-10-22 云南电力试验研究院（集团）有限公司电力研究院 一种旋转机械扭振冲击信号特征提取方法
KR101574615B1 (ko) * 2015-06-18 2015-12-11 지투파워(주) 통계적 위상패턴 기반의 신호검출 방법을 이용한 전력설비 부분 방전 감시진단 시스템
CN107766816A (zh) * 2017-10-18 2018-03-06 河海大学 一种基于lvq神经网络的高压断路器机械故障识别方法
CN110320467B (zh) * 2019-06-19 2021-11-09 武汉船用电力推进装置研究所（中国船舶重工集团公司第七一二研究所） 一种低压直流断路器故障诊断方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102736027A (zh) * 2012-07-18 2012-10-17 南京因泰莱配电自动化设备有限公司 一种基于断路器动特性测试仪的断路器故障诊断方法
CN106574863A (zh) * 2014-08-21 2017-04-19 湖原大学校产学协力团 电力设备声音诊断系统
CN109839263A (zh) * 2019-02-26 2019-06-04 北京航空航天大学 一种基于振动信息的gis特征提取及机械缺陷诊断方法
CN110132565A (zh) * 2019-05-28 2019-08-16 河海大学 一种基于小波包和神经网络结合的oltc故障诊断方法
CN110926778A (zh) * 2019-11-29 2020-03-27 国网天津市电力公司电力科学研究院 基于异常振动的气体绝缘组合开关设备机械故障诊断方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SUN LAI-JUN , HU XIAO-GUANG , JI YAN-CHAO: "Fault Diagnosis for High Voltage Circuit Breakers With Improved Characteristic Entropy of Wavelet Packet", PROCEEDINGS OF THE CSEE, vol. 27, no. 12, 25 April 2007 (2007-04-25), pages 103 - 108, XP055816582, ISSN: 0258-8013, DOI: 10.13334/j.0258-8013.pcsee.2007.12.019 *
SUN LAI-JUN , HU XIAOGWANG , JI YANCHAO , LU CHAO: "Fault Diagnosis for HV Circuit Breakers with Characteristic Entropy of Wavelet Packet", AUTOMATION OF ELECTRIC POWER SYSTEMS, vol. 30, no. 14, 25 July 2006 (2006-07-25), XP055817255, ISSN: 1000-1026 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113705637A (zh) * 2021-08-12 2021-11-26 西安交通大学 一种断路器机械故障检测方法、系统、设备及可读存储介质
CN113705637B (zh) * 2021-08-12 2024-04-02 西安交通大学 一种断路器机械故障检测方法、系统、设备及可读存储介质
CN113933038A (zh) * 2021-10-11 2022-01-14 国网山东省电力公司菏泽供电公司 一种基于多源信息融合算法的gis机械性能监测方法
CN113933038B (zh) * 2021-10-11 2023-11-17 国网山东省电力公司菏泽供电公司 一种基于多源信息融合算法的gis机械性能监测方法
CN114034381A (zh) * 2021-11-12 2022-02-11 广东电网有限责任公司江门供电局 一种基于小波信息熵的配电变压器振动提取方法及系统
CN114034381B (zh) * 2021-11-12 2023-10-10 广东电网有限责任公司江门供电局 一种基于小波信息熵的配电变压器振动提取方法及系统
CN114743089A (zh) * 2021-12-22 2022-07-12 宁波送变电建设有限公司 一种基于ssa-svm的图像识别gis故障诊断装置及方法
CN114743089B (zh) * 2021-12-22 2024-06-04 宁波送变电建设有限公司 一种基于ssa-svm的图像识别gis故障诊断装置及方法
CN115510925A (zh) * 2022-11-17 2022-12-23 中国人民解放军国防科技大学 一种机械故障诊断方法、装置、介质
CN116415509A (zh) * 2023-06-12 2023-07-11 华东交通大学 一种轴承性能退化预测方法、系统、计算机及存储介质
CN116415509B (zh) * 2023-06-12 2023-08-11 华东交通大学 一种轴承性能退化预测方法、系统、计算机及存储介质

Also Published As

Publication number Publication date
CN110926778B (zh) 2021-08-20
CN110926778A (zh) 2020-03-27

Similar Documents

Publication Publication Date Title
WO2021103496A1 (zh) 基于异常振动的气体绝缘组合开关设备机械故障诊断方法
CN106482937B (zh) 一种高压断路器机械状态的监测方法
US20210167584A1 (en) Gis mechanical fault diagnosis method and device
CN103713237B (zh) 一种电力系统输电线路短路故障诊断方法
CN103116090B (zh) 一种基于小波包分析和支持向量机的三相pwm整流器故障诊断方法
CN104281982B (zh) 一种基于电网拓扑结构的变电站设备状态评估方法
CN106324490A (zh) 变压器有载分接开关机械故障诊断方法
CN103575525A (zh) 一种断路器机械故障的智能诊断方法
WO2022105287A1 (zh) 一种基于声纹诊断来监测异常状态的系统及方法
CN107728509A (zh) 一种基于多维数据模型的断路器机械特性在线专家诊断系统
CN102135560A (zh) 雷电波侵入变电站的扰动识别方法
CN110703151A (zh) 基于振动盲源分离和贝叶斯模型的变压器故障诊断方法
CN109800740A (zh) 一种基于样本熵和svm的oltc机械故障诊断方法
CN112083328A (zh) 一种高压断路器故障诊断方法及系统和装置
CN105241643A (zh) 采用hs变换和单类支持向量机的高压断路器机械状态监测方法
Yi et al. Insulation fault diagnosis of disconnecting switches based on wavelet packet transform and PCA-IPSO-SVM of electric fields
CN106980051B (zh) 一种间歇性串联型故障电弧识别方法
WO2022105285A1 (zh) 一种基于冲击事件来监测异常状态的系统及方法
CN117110798B (zh) 智能配电网的故障检测方法和系统
Ke et al. A novel power transformer condition monitoring system based on wide-band measurement of core earth signals and correlation analysis with multi-source data
CN110018391B (zh) 一种基于振动信号的gis故障精确定位方法及定位系统
CN116881825A (zh) 基于gis设备的刀闸状态和三相刀闸动作的同步性检测方法
CN116910470A (zh) 一种gis组合电器局部放电故障模式识别方法
CN101930049A (zh) 超高压电缆在线局放的监测方法
CN108152733B (zh) 一种用于高压断路器机械特性的带电检测分合闸时间方法

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: 20894675

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20894675

Country of ref document: EP

Kind code of ref document: A1