System and method for iterative condition monitoring and fault diagnosis of electric machines

Patent metadata

Field	Value
Publication number	US-10429419-B2
Application number	US-201514874210-A
Country	US
Kind code	B2
Filing date	Oct 2, 2015
Priority date	Mar 26, 2015
Publication date	Oct 1, 2019
Grant date	Oct 1, 2019

Abstract

A system and method for iterative condition monitoring and fault diagnosis of an electric machine, such as an electric motor, is configured to iteratively analyze a pattern of multiple fault signatures in a motor current signal under harsh, noisy conditions. As such, the present invention evaluates patterns of multiple fault signatures that are spread over the wide motor current spectrum, by effectively rejecting signal distortions, harsh noise, and interference from the dynamic operation of the electric machine.

First claim

What is claimed is:

1. A method of detecting a fault comprising the steps of:
providing a 3-phase AC (alternating current) motor coupled to a voltage source inverter;
coupling a control feedback module to said voltage source inverter including:
a motor controller;
a stationary frame module to convert voltage output signals of said motor controller into frame voltage signals; and
a SVPWM (space vector pulse width modulation) generator to receive said frame voltage output signals, said SVPWM generator configured to generate a 3-phase signal for supplying to said 3-phase AC motor, and wherein said motor controller monitors 2 phases of said 3-phase signal;

detecting said 2 phases of said 3 phase signal to identify a motor current spectrum signature;
identifying n fault characteristic frequency signatures from said motor current spectrum signature, wherein n is a positive integer value;
identifying m random noise frequency signatures from said motor current spectrum signature, wherein m is a positive integer value equal to n;
comparing an amplitude of said n fault characteristic frequency signatures with an amplitude of said m random noise frequency signatures;
generating at least one initial detection matrix including at least one matrix element, such that each said matrix element is associated with said comparison of an amplitude of said n fault characteristic frequency signatures and an amplitude of said m random noise frequency signatures;
assigning a first binary number to said matrix element if said amplitude of said n fault characteristic frequency signatures is greater than said amplitude of said m random noise frequency signatures to identify the presence of a fault, or assigning a second binary number different from said first binary number to said matrix element if said amplitude of said fault characteristic frequency signatures is less than said amplitude of said random noise frequency signatures to identify the presence of a non-fault;
generating a probability distribution based on said at least one initial detection matrix;
applying a predetermined threshold value to said probability distribution of said at least one initial detection matrix to generate a binary decision element;
generating an intermediate detection matrix that includes said binary decision element, wherein said second generating step and said applying step are performed for said intermediate decision matrix to generate a final binary decision value, and wherein said threshold value applied to said intermediate decision matrix is different from said threshold value applied to said at least one initial detection matrix; and
determining whether a fault is present in the 3-phase AC motor based on said final binary decision value.

2. The method of claim 1 , wherein said probability distribution is generated by binomial distribution.

3. A system to detect a fault comprising:
a 3-phase AC (alternating current) motor including a voltage source inverter coupled thereto;
a control feedback module to said voltage source inverter including:
a motor controller;
a stationary frame module to convert voltage output signals of said motor controller into frame voltage signals; and
a SVPWM (space vector pulse width modulation) generator to receive said frame voltage output signals, said SVPWM generator configured to generate a 3-phase signal for delivery to said 3-phase AC motor;

a signature detection means for detecting signatures based on at least 2 of said 3 phases applied to the 3-phase AC motor;
a detection matrix generator means for generating a detection matrix of binary elements based on a signal output by said signature detection means;
a decision making means for determining whether there is a fault condition associated with the electric motor;
an update detection matrix means for updating said detection matrix, said update detection matrix means in bi-directional communication with said decision making means; and
a noise detection means for determining whether said signature includes noise.

Assignees

• Choi Seungdeog
• Univ Akron

Inventors

• Choi Seungdeog

Classifications

• G01R23/20 Primary

  Measurement of non-linear distortion · CPC title

• G01R31/343

  in operation · CPC title