Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Fault diagnosis method of equipment analog circuit based on probabilistic neural network

A probabilistic neural network, a technology for simulating circuit faults, applied in biological neural network models, neural architectures, electrical and digital data processing, etc. Effect

Active Publication Date: 2022-05-03
NAVAL AVIATION UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In order to improve the performance of the existing probabilistic neural network pattern classification methods, relevant scholars have conducted a lot of research in this area, and the representative ones are swarm intelligence algorithms such as ant colony algorithm, fruit fly algorithm, and fireworks algorithm, but these algorithms have not shown Absolute advantage, the effect is not significant

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Fault diagnosis method of equipment analog circuit based on probabilistic neural network
  • Fault diagnosis method of equipment analog circuit based on probabilistic neural network
  • Fault diagnosis method of equipment analog circuit based on probabilistic neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0144] In this example specifying ψ 1 、ψ 2 The role of the two parameters, with image 3 The circuit shown is the research object, the selected state set is shown in Table 1 below, and the fault feature at the output node OUT is selected as the BCM value. Since there is only one output feature—the BCM value of the voltage at the OUT node, the sensitivity matrix becomes a 1×11-dimensional row vector Each element represents the sensitivity of the corresponding component out-of-tolerance condition when the fault feature is BCM value. The measured values ​​are shown in Table 2 below.

[0145] Table 1 Circuit state set

[0146]

[0147] Table 2 Sensitivity of measured values ​​with BCM value as fault feature

[0148]

[0149] Combining Table 1 and Table 2, it can be seen that:

[0150] (1) Except state X 2 、X 3 、X 4 、X 5 、X 6 、X 7 In addition, the sensitivities are too small, if the first type of threshold parameter ψ 1 =1%, then the remaining states with too lo...

Embodiment 2

[0158] In this embodiment, the diagnosis effect of the fault diagnosis method of the present invention is verified. Still use the attached image 3 In the circuit diagram in the circuit, the fault state of the components in the circuit is set as positive (↑) and negative (↓) 20% of the nominal value, and the node OUT is the output test point. The simulation platform is Orcad CIS 7.0 and Matlab 8.5, and the number of sampling points is 1024. Take the circuit state set as shown in the part of the circuit state in Table 3 below.

[0159] Table 3 Partial circuit status

[0160]

[0161] When the input excitation signal is set as v=10sin(ωt), the BCM value is calculated for different fault states of the circuit. Furthermore, 100 sets of data in each state of ω=100 are taken, among which 40 sets are used as training samples and 60 sets are used as test samples. Utilize the algorithm setting Z=80% among the present invention, obtain parameter σ=0.8638, ψ 1 =0.813%, ψ 2 =0.09...

Embodiment 3

[0167] to attach Figure 5 Take part of the circuit in a certain type of missile as an example to verify the effectiveness of the algorithm in the present invention. Assume that the fault has been located to the circuit part shown in the figure, but the specific location and parameters are unknown. Through the parameter optimization method in the present invention, the variable resistor RP3 has low sensitivity to the output characteristics at the test point OUT, so it is eliminated, and the remaining measurable circuit faults are sorted into Table 5, together with the normal state, there are 19 categories in total.

[0168] Table 5 The measurable fault state table of the active filter circuit of a certain missile

[0169]

[0170] Adjust the corresponding components in the circuit according to the fault values ​​of each component shown in Table 5, collect signals at OUT and process the samples using the SVD-LCD-BCM algorithm in the present invention, and input the processe...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a probabilistic neural network-based equipment analog circuit fault diagnosis method, comprising the following steps: S1: constructing a probabilistic neural network model for analog circuit fault diagnosis; Decompose and screen out the useful components; S3: calculate the BCM value of the useful components, and construct the corresponding fault feature vector; S4: use the fault feature vector based on the training samples to train the probabilistic neural network model, and set The minimum optimal probability value Z, optimize the three types of parameters until the fault state set output by the model meets the predetermined requirements, and save the training results; S5: Construct the corresponding fault feature vector of the test samples according to the operations of steps S2 and S3, and input Fault diagnosis is carried out in the probabilistic neural network model trained in step S4. The circuit fault diagnosis method in the present invention provides a technical approach for the selection of fault categories in fault diagnosis.

Description

technical field [0001] The invention relates to the technical field of analog circuit fault diagnosis, in particular to a probabilistic neural network-based fault diagnosis method for equipment analog circuits. Background technique [0002] Fault diagnosis of analog circuits is essentially equivalent to the problem of pattern recognition. Pattern recognition is usually concerned with the recognition range and recognition accuracy, the two are interrelated and opposite to each other. However, it seems that many scholars often ignore or deliberately omit the selection of fault categories when conducting research on this type of subject. In the case that the screening method of the fault category is not clearly explained, it has tried to prove that its research results have a very high diagnostic coverage and accuracy rate, and its experimental data and results will inevitably have a certain degree of subjective color. In fact, with the current technical means, it is impossib...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F30/3308G06K9/00G06K9/62G06N3/04
CPCG06F30/3308G06N3/045G06F2218/04G06F2218/08G06F18/214
Inventor 徐学文盛沛郑振戴永军盖炳良白玉张广法
Owner NAVAL AVIATION UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com