A Fault Diagnosis Method for an Internet of Things-Based GIS Online Monitoring System

Abstract

The present invention relates to a fault diagnosis method for an online monitoring system of GIS based on the Internet of Things, comprising the following steps: regularly collecting GIS device data, uniformly sampling the GIS device data to obtain GIS sampled data, and obtaining a GIS sampled signal according to the GIS sampled data; judging the stability and abnormality of the GIS sampled signal. If the GIS signal is an unstable abnormal signal, collecting the GIS device data through non-uniform interval compressive sampling to obtain GIS observation data, and obtaining a GIS observation signal according to the GIS observation data; transmitting the GIS observation signal to the system aggregation node through a wireless network, and reconstructing the GIS observation signal at the aggregation node to obtain a reconstructed signal; extracting waveform features in the time domain and frequency domain of the reconstructed signal, and performing fault diagnosis according to the extraction results. The present invention combines conventional sampling and non-uniform interval sampling to collect data for GIS, realizes the restoration and reconstruction of original data, and then performs feature extraction to improve the accuracy of fault features.

Description

Technical Field

[0001] The present invention relates to a fault diagnosis method for an on-line monitoring system of GIS based on the Internet of Things, and belongs to the technical field of insulation state monitoring of power equipment. Background Art

[0002] The partial discharge of the insulation of power equipment refers to the discharge occurring in a local area under the action of a strong electric field. The occurrence of partial discharge in some weak parts of the insulation under the action of a high electric field is a common problem, which will lead to insulation deterioration and even breakdown under certain conditions. The gas-insulated switchgear (GIS), with the advantages of small floor space and high reliability, has been more and more widely used in modern power systems. Although GIS has many advantages, during the production process in the factory, the warehousing and logistics process after leaving the factory, the installation process at the operation site, and the natural decline of the equipment state after entering the operation state, the GIS equipment will inevitably face the problem of insulation defects and their impact on long-term reliability. Since GIS is a fully enclosed combined power equipment, once an accident occurs inside the GIS equipment, due to the characteristics of equipment sealing and gas filling, the consequences are much more serious than those of separated open equipment. Its fault repair is particularly complex, and the power outage range is large, often involving non-fault components.

[0003] The on-line monitoring of GIS can well monitor the equipment for a long time and in real time, timely detect equipment faults, and ensure the normal operation of the equipment. However, due to the high sampling rate and large amount of data of ultra-high frequency, affected by the storage space at the sensor end and communication pressure, the conventional method will perform decimation processing on the sampling data, which will lose some useful characteristics of the original signal and cause the decline of the fault diagnosis rate. Summary of the Invention

[0004] In order to overcome the above problems, the present invention provides a fault diagnosis method for an on-line monitoring system of GIS based on the Internet of Things. This method combines conventional sampling and non-uniform sampling to collect data of GIS, realizes the restoration and reconstruction of the original data, and then performs feature extraction to improve the accuracy of fault features, providing high guiding significance for the development of an on-line monitoring system based on GIS, and can be applied to the on-line monitoring of GIS in different environments.

[0005] The technical solution of the present invention is as follows:

[0006] A fault diagnosis method for an on-line monitoring system of GIS based on the Internet of Things, comprising the following steps:

[0007] Conventionally collect data of GIS equipment, uniformly sample the GIS equipment data to obtain GIS sampled data, and obtain a GIS sampled signal according to the GIS sampled data;

[0008] Perform stability judgment and abnormality judgment on the GIS sampling signal. If the GIS signal is an unstable abnormal signal, collect GIS device data through non-uniform interval compressive sampling to obtain GIS observation data, and obtain a GIS observation signal based on the GIS observation data;

[0009] Transmit the GIS observation signal to the system aggregation node through a wireless network, and reconstruct the GIS observation signal at the aggregation node to obtain a reconstructed signal;

[0010] Extract waveform features in the time domain and frequency domain of the reconstructed signal, and perform fault diagnosis based on the extraction results.

[0011] Furthermore, the stability judgment and abnormality judgment on the GIS sampling signal are specifically as follows:

[0012] Calculate the average value and variance of each part of the GIS sampling signal through a sliding window;

[0013] Judge whether the GIS sampling signal is a stable signal according to each average value and the variance;

[0014] If the GIS sampling signal is an unstable signal, perform Weibull distribution parameter estimation on the GIS sampling signal, and perform abnormality judgment on the GIS sampling signal according to the Weibull distribution parameter estimation result.

[0015] Furthermore, the size of the sliding window is 50.

[0016] Furthermore, the judgment of whether the GIS sampling signal is a stable signal according to each average value and the variance is specifically as follows: If the variance within the same sliding window is less than 0.1 times the average value, the above GIS sampling signal is a stable signal; otherwise, the above GIS sampling signal is an unstable signal.

[0017] Furthermore, the abnormality judgment on the GIS sampling signal according to the Weibull distribution parameter estimation result is specifically as follows: If the Weibull distribution parameter estimation is less than 1.5, the GIS sampling signal is an abnormal signal; otherwise, the GIS sampling signal is a non-abnormal signal.

[0018] Furthermore, the collection of GIS device data through non-uniform interval compressive sampling to obtain GIS observation data is specifically as follows:

[0019] Determine the measurement matrix ψ, and perform non-uniform interval sampling on the GIS device data through the measurement matrix ψ to obtain the observation data U1; where U1 = ψ * x, and x is the signal to be restored.

[0020] Further, the measurement matrix is a random Bernoulli matrix.

[0021] Further, the reconstruction of the GIS observation signal at the aggregation node is specifically as follows: signal recovery is performed on each observation signal through a cosine transform orthogonal basis and a greedy algorithm.

[0022] Further, the extraction of waveform features in the time domain and frequency domain from the reconstructed signal includes extracting the standard deviation, peak value, root mean square, kurtosis, margin factor, and impulse factor.

[0023] The present invention has the following beneficial effects:

[0024] 1. This method combines conventional sampling and non-uniform sampling to collect data from GIS, realizes the recovery and reconstruction of original data, then performs feature extraction, improves the accuracy of fault features and the fault diagnosis rate, saves memory space and reduces the resources required for communication, improves the stability of data transmission, and reduces costs. Description of the Drawings

[0025] Figure 1 It is a schematic flowchart of an embodiment of the present invention. Detailed Embodiment

[0026] The present invention will be described in detail below with reference to the drawings and specific embodiments.

[0027] Embodiment 1

[0028] A fault diagnosis method for an on-line monitoring system of GIS based on the Internet of Things includes the following steps:

[0029] Conventionally collect GIS device data, uniformly sample the GIS device data to obtain GIS sampled data, and obtain a GIS sampled signal according to the GIS sampled data;

[0030] Perform stability judgment and abnormality judgment on the GIS sampled signal. If the GIS signal is a non-steady abnormal signal, collect GIS device data through non-uniform compressive sampling to obtain GIS observation data, and obtain a GIS observation signal according to the GIS observation data;

[0031] Transmit the GIS observation signal to the system aggregation node through a wireless network, and reconstruct the GIS observation signal at the aggregation node to obtain a reconstructed signal;

[0032] Extract waveform features in the time domain and frequency domain from the reconstructed signal, and perform fault diagnosis according to the extraction results.

[0033] Embodiment 2

[0034] A fault diagnosis method for an Internet of Things-based GIS online monitoring system. Further, on the basis of Embodiment 1, the stability judgment and abnormality judgment of the GIS sampling signal are specifically as follows:

[0035] Calculate the average value and variance of each part of the GIS sampling signal through a sliding window;

[0036] Judge whether the GIS sampling signal is a steady-state signal according to each of the average values and the variance;

[0037] If the GIS sampling signal is a non-steady-state signal, perform Weibull distribution parameter estimation on the GIS sampling signal, and perform abnormality judgment on the GIS sampling signal according to the Weibull distribution parameter estimation result.

[0038] In an embodiment of the present invention, the size of the sliding window is 50.

[0039] In an embodiment of the present invention, the judgment of whether the GIS sampling signal is a steady-state signal according to each of the average values and the variance is specifically as follows: If the variance within the same sliding window is less than 0.1 times the average value, the above GIS sampling signal is a steady-state signal; otherwise, the above GIS sampling signal is a non-steady-state signal.

[0040] Embodiment 3

[0041] A fault diagnosis method for an Internet of Things-based GIS online monitoring system. Further, on the basis of Embodiment 1, the abnormality judgment of the GIS sampling signal according to the Weibull distribution parameter estimation result is specifically as follows: If the Weibull distribution parameter estimation is less than 1.5, the GIS sampling signal is an abnormal signal; otherwise, the GIS sampling signal is a non-abnormal signal.

[0042] Embodiment 4

[0043] A fault diagnosis method for an Internet of Things-based GIS online monitoring system. Further, on the basis of Embodiment 1, the acquisition of GIS device data through non-uniform interval compressive sampling to obtain GIS observation data is specifically as follows:

[0044] Determine the measurement matrix ψ, and perform non-uniform interval sampling on the GIS device data through the measurement matrix ψ to obtain the observation data U1; where U1 = ψ * x, and x is the signal to be recovered.

[0045] In an embodiment of the present invention, the measurement matrix is a random Bernoulli matrix.

[0046] In an embodiment of the present invention, the reconstruction of the GIS observation signal at the aggregation node specifically includes: signal recovery of each observation signal by using the cosine transform orthogonal basis and the greedy algorithm.

[0047] In an embodiment of the present invention, the extraction of the waveform features in the time domain and frequency domain from the reconstructed signal includes extracting the standard deviation, peak value, root mean square, kurtosis, margin factor, and impulse factor.

[0048] The above are only the embodiments of the present invention, and thus do not limit the patent scope of the present invention. Any equivalent structures made by using the contents of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, are equally included in the patent protection scope of the present invention.

Claims

1. A fault diagnosis method for an Internet of Things-based GIS online monitoring system, characterized in that, it includes the following steps: Regularly collect GIS device data, uniformly sample the GIS device data to obtain GIS sampled data, and obtain a GIS sampled signal according to the GIS sampled data; Judge the stability and abnormality of the GIS sampled signal. If the GIS signal is an unstable abnormal signal, collect GIS device data through non-uniform interval compressive sampling to obtain GIS observed data, and obtain a GIS observed signal according to the GIS observed data; Transmit the GIS observed signal to the system aggregation node through a wireless network, and reconstruct the GIS observed signal at the aggregation node to obtain a reconstructed signal; Extract the waveform characteristics in the time domain and frequency domain of the reconstructed signal, and perform fault diagnosis according to the extraction results; The judgment of the stability and abnormality of the GIS sampled signal is specifically: Calculate the average value and variance of each part of the GIS sampled signal through a sliding window; Judge whether the GIS sampled signal is a steady-state signal according to each average value and the variance; If the GIS sampled signal is an unsteady signal, perform Weibull distribution parameter estimation on the GIS sampled signal, and judge the abnormality of the GIS sampled signal according to the Weibull distribution parameter estimation result.

2. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 1, characterized in that, the size of the sliding window is 50.

3. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 1, characterized in that, the judgment of whether the GIS sampled signal is a steady-state signal according to each average value and the variance is specifically: if the variance within the same sliding window is less than 0.1 times the average value, the above GIS sampled signal is a steady-state signal, otherwise, the above GIS sampled signal is an unsteady signal.

4. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 1, characterized in that, the judgment of the abnormality of the GIS sampled signal according to the Weibull distribution parameter estimation result is specifically: if the Weibull distribution parameter estimation is less than 1.5, the GIS sampled signal is an abnormal signal, otherwise, the GIS sampled signal is a non-abnormal signal.

5. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 1, characterized in that, the collection of GIS device data through non-uniform interval compressive sampling to obtain GIS observed data is specifically: Determine the measurement matrix ψ, and perform non-uniform interval sampling on the GIS device data through the measurement matrix ψ to obtain the observed data U1; where U1 = ψ * x, and x is the signal to be restored.

6. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 5, characterized in that, the measurement matrix is a random Bernoulli matrix.

7. The fault diagnosis method for an Internet of Things-based GIS online monitoring system according to claim 5, characterized in that, The reconstruction of the GIS observation signal at the convergence node is specifically as follows: signal recovery is performed on each observation signal through a cosine transform orthogonal basis and a greedy algorithm.

8. The fault diagnosis method for the GIS online monitoring system based on the Internet of Things according to claim 5, characterized in that the extraction of waveform features in the time domain and frequency domain from the reconstructed signal includes extracting the standard deviation, peak value, root mean square, kurtosis, margin factor, and impulse factor.

Images