Method and system for power equipment diagnosis based on windowed feature and Hilbert visualization

Patent metadata

Field	Value
Publication number	US-11520676-B2
Application number	US-202117161680-A
Country	US
Kind code	B2
Filing date	Jan 29, 2021
Priority date	May 25, 2020
Publication date	Dec 6, 2022
Grant date	Dec 6, 2022

Abstract

A method and a system for power equipment diagnosis based on windowed feature and Hilbert visualization are provided, which belong to the field of power equipment fault diagnosis. The method includes: obtaining an original data set of monitoring data containing power equipment fault features; introducing windowed feature calculation considering logarithmic constraints to process data to obtain a feature sequence; using Hilbert visualization method for further processing to obtain a Hilbert image data set used to train and verify a convolutional neural network; and finally directly inputting newly obtained test sample data after windowed feature calculation and Hilbert visualization processing into the trained network for fault diagnosis and location. The disclosure uses windowed feature calculation and Hilbert visualization to process the monitoring data of a power equipment to fully extract fault features and effectively improve diagnostic accuracy, and uses the convolutional neural network for diagnosis to improve the intelligence of diagnosis.

First claim

What is claimed is:

1. A method for power equipment diagnosis based on windowed feature and Hilbert visualization, comprising:
step 1 of obtaining a plurality of sets of monitoring data comprising different fault locations, fault types, and severity of a power equipment, analyzing each set of the monitoring data, obtaining a corresponding fault status label, and using each set of the monitoring data and the con - esponding fault status label to constitute a target data set, wherein the fault status label comprises the fault type and the fault location;
step 2 of performing windowed feature calculation on the target data set considering logarithmic constraints, extracting fault features contained therein, and obtaining a feature sequence data set;
step 3 of performing Hilbert visualization processing on each set of data in the feature sequence data set, and obtaining a Hilbert image data set;
step 4 of constructing a fault diagnosis and location model, dividing the Hilbert image data set into a training set and a verification set, training the fault diagnosis and location model using the training set, and verifying the trained fault diagnosis and location model using the verification set; and
step 5 of obtaining a final diagnosis result from the trained fault diagnosis and location model after performing the same windowed feature calculation and Hilbert visualization on diagnosis data.

2. The method according to claim 1 , wherein the plurality of sets of the monitoring data of the power equipment is data 1 ={a i,1 , a i,2 , . . . , a i,j , . . . , a i,N , s i }i∈[1, K], where K is K sets of the monitoring data of the power equipment, a i,j is a monitoring value corresponding to a j(j∈[1, N]) -th point in an i -th set of the monitoring data of the power equipment, s i is the fault status label of the power equipment corresponding to the i -th set of the monitoring data of the power equipment, and N is a number of monitoring data points.

3. The method according to claim 2 , wherein step 2 comprises:
taking a monitoring curve of the power equipment in a normal state as a reference curve, and performing the windowed feature calculation considering the logarithmic constraints based on the reference curve to obtain a feature sequence data set, wherein a feature value is a self-defined evaluation indicator or an evaluation indicator provided by different research work.

4. The method according to claim 3 , wherein step 2 comprises:
step 2.1 of defining a window with a specified length W w , wherein the window scans a frequency band step by step with a step size s from a starting point x 0 of a scanning frequency band range to an end x n , each time the window moves one step, the feature value is calculated within a frequency region of the length W w of the window, and a finally obtained feature sequence length is

N
′

=

floor
(

(

x
n

-

x
0

)

-

W
w

s

)

+
1

,

where s and W w , are positive integer multiples of a shortest interval between monitoring points, all monitoring data is to be scanned, thus W w ≥2s , floor represents a round-down function, and returns to a largest previous integer;
step 2.2 of determining an x-coordinate of a symmetry axis of a k(k=1, 2, . . . , N′)-th window from

x
k
c

=

x
0

+

W
w

2

+

(

k
-
1

)

·
s

,

determining an x-coordinate of a left boundary from

x
k
left

=

x
k
c

-

W
w

2

,

determining an x-coordinate of a right boundary from

x
k
right

=

x
k
c

+

W
w

2

,

and determining a number of points contained in each window from

n
w

=

W
w

Δ
⁢
f

+
1

,

where Δf is an interval between two monitoring points; and
step 2.3 of obtaining a windowed feature value considering the logarithmic constraints from the x-coordinate of the symmetry axis, the x-coordinate of the left boundary, the x-coordinate of the right boundary, and the number of points contained in each window based on a selected feature value evaluation indicator, so as to obtain the feature sequence data set.

5. The method according to claim 4 , wherein a feature sequence data obtained after the windowed feature calculation is data′ i ={b i,1 , b i,2 , . . . , b i,j , . . . b i,N′ , s i }i∈[1,K], where K is K sets of the feature sequence da

Assignees

• Univ Wuhan

Inventors

• He Yigang
• Wu Xiaoxin
• Duan Jiajun
• Liu Xiaoyan
• Li Lie
• Zeng Zhaorong

Classifications

• G06V10/82 Primary

  using neural networks · CPC title

• G06F11/3058 Primary

  Monitoring arrangements for monitoring environmental properties or parameters of the computing system or of the computing system component, e.g. monitoring of power, currents, temperature, humidity, position, vibrations (thermal management in cooling arrangements of a computing system G06F1/206) · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06N3/045

  Combinations of networks · CPC title