KR20220100310A - Error detection apparatus and method of motor - Google Patents

Abstract

An apparatus for detecting the failure of a motor according to a preferred embodiment of the present invention includes: a resampling part for performing resampling on a phase current signal of the motor and converting it into a resampling signal; a conversion part for preprocessing the resampling signal and converting the resampling signal into a preprocessing signal; a learning part for learning failure characteristic data through the preprocessing signal; and a determination part for determining whether the motor is out of order by using the failure characteristic data.

Description

Fault detection device and method of electric motor

The present invention relates to an apparatus and method for detecting a failure of an electric motor.

An electric motor is a device that converts electrical energy into mechanical energy and is used in various industrial fields including fans, robots, electric vehicles, and pumps. In the industrial field, AC motors using AC power are mainly used and are widely used because of their advantages of relatively high efficiency and long lifespan compared to DC motors.

In order to ensure smooth operation and safety of the motor, fault diagnosis is essential. In particular, if a breakdown of some motors in an industrial line occurs, it can cause a huge economic loss by causing the entire line to be stopped. Accordingly, it is considered an important issue to develop a reliable and effective motor failure diagnosis technique.

In general, a fault diagnosis algorithm based on an industrial robot running at a constant speed and an electric vehicle is used for fault diagnosis of an electric motor.

In the conventional fault diagnosis algorithm, the motor signature current analysis (MCSA) method is applied, which analyzes the stator current of the motor to extract the characteristics related to the fault frequency and then uses the extracted characteristic frequency to diagnose the fault of the motor. .

The conventional fault diagnosis algorithm is applicable only to a constant velocity and identical load signal that can be converted into a frequency domain. This is because the frequency of the current also changes according to the variable rotation speed caused by the variable speed operation of the motor, so it is impossible to obtain the failure characteristic frequency through the frequency analysis method. In addition, information on parameters (eg, number of poles, bearing information, etc.) of the motor is required to obtain the failure characteristic frequency. In addition, as the magnitude and frequency of the electric current continuously change, there is a problem in that accurate fault diagnosis is difficult because in many cases similar fault characteristic frequencies occur despite different faults.

To solve this problem, a deep learning-based fault diagnosis method is being developed recently. Deep learning is a model created by stacking several neural networks in mathematical form, and performs the function of automatically learning the characteristics of data and classifying the learned data.

The deep learning-based fault diagnosis method has high diagnostic performance and is widely used for fault diagnosis. However, a large amount of data is required to learn the model, and there is a possibility that generalization performance is not good because it is excessively limited only to the data learned by the model. However, there is a disadvantage in that interpretation of diagnostic results is difficult.

Republic of Korea Patent No. 10-2189269

Accordingly, the present invention has been devised in consideration of the above circumstances, and it is possible to diagnose a failure using the magnitude of the phase current signal and the failure characteristic frequency of the electric motor, which varies according to the operating conditions of the electric motor, and failure using a convolution-based deep learning algorithm. An object of the present invention is to provide a motor fault detection device to which a current data preprocessing method capable of increasing classification and generalization performance is applied.

In accordance with a preferred embodiment of the present invention for achieving the above object, there is provided an apparatus for detecting a failure of a motor, comprising: a resampling unit for performing resampling on a phase current signal of a motor and converting it into a resampling signal; a conversion unit that pre-processes the resampling signal and converts it into a pre-processed signal; a learning unit for learning failure characteristic data through the preprocessing signal; and a determination unit configured to determine whether the motor has failed by using the failure characteristic data.

The resampling unit may convert the phase current signal into an analysis signal, estimate an instantaneous time from the analysis signal, and separate the analysis signal for each period based on the change point of the instantaneous time.

The resampling unit may generate the resampling signal by resampling the analysis signal separated by period through a pre-prepared interpolation method into a signal of the same interval, and connecting the signals of the same interval.

The transform unit may perform a log-scale fast Fourier transform on the resampling signal.

The transform unit may generate the preprocessed signal by performing scaling on the magnitude of the signal according to the fast Fourier transform.

The scaling unit may have a value of 0 to 1.

The learning unit may learn the preprocessing signal using a pre-prepared convolution-based deep learning learning algorithm.

The determination unit may determine an axial imbalance or winding deterioration failure of the electric motor by using the failure characteristic data.

The determination unit may interpret the failure determination result for the motor using a pre-prepared CAM (Class Activation Map)-based analysis algorithm.

A fault detection method of a motor according to a preferred embodiment of the present invention for achieving the above object, a resampling step of performing resampling on the phase current signal of the motor to convert it into a resampling signal; a pre-processing step of pre-processing the resampling signal and converting it into a pre-processing signal; a learning step of learning failure characteristic data through the preprocessing signal; and a failure determination step of determining whether the electric motor has failed using the failure characteristic data.

The resampling step may include converting the phase current signal into an analysis signal; estimating an instantaneous time from the analysis signal; separating the analysis signal for each period based on the change point of the instantaneous time; Resampling the analysis signal separated for each period through a pre-prepared interpolation method into a signal having the same interval; and generating the resampling signal by connecting the signals of the same interval.

The pre-processing step includes: a fast Fourier transform step of performing a log scale fast Fourier transform on the resampling signal; and a scaling step of performing scaling on the magnitude of the signal according to the fast Fourier transform.

In the learning step, the preprocessing signal may be learned using a pre-prepared convolution-based deep learning learning algorithm.

In the failure determination step, it is possible to determine an axial imbalance or winding deterioration failure of the electric motor using the failure characteristic data.

In the failure determination step, the failure determination result of the motor may be analyzed using a pre-prepared CAM (Class Activation Map)-based analysis algorithm.

According to the apparatus and method for detecting a failure of an electric motor according to a preferred embodiment of the present invention, it is possible to diagnose a failure using the magnitude of the electric current signal and the frequency of the failure characteristic that vary according to the operating conditions of the electric motor, and a convolution-based deep learning learning algorithm By applying the current data preprocessing method that can improve fault classification and generalization performance using

In addition, since the characteristic frequency of the motor is automatically extracted through a convolution-based deep learning learning algorithm, prior knowledge related to the motor parameters is not required. Since the diagnosis is made, a high-performance current measuring sensor is not required, which has the effect of reducing the cost.

In addition, the result of the convolution-based deep learning learning algorithm can be interpreted through a post-processing algorithm, so that an expert can judge the consistency of the algorithm.

In addition, there is an effect applicable to dynamic motor systems (eg, electric vehicles, and industrial robots) that are difficult to diagnose with vibration signals.

In addition, there is an effect of identifying the signal failure mode without disassembling the faulty motor.

1 is a block diagram of an apparatus for detecting a failure of an electric motor according to a preferred embodiment of the present invention.
2 is a view showing an example of a speed profile of an electric motor.
3 is a view showing a signal processing result for a phase current of an electric motor operating at a constant speed according to the speed profile of FIG. 2 .
FIG. 4 is a view showing a signal processing result for a phase current of an electric motor operating in a speed change according to the speed profile of FIG. 2 .
5 is a diagram showing the structure of a convolutional neural network of a learning unit according to a preferred embodiment of the present invention.
6 is a view showing the analysis result of analyzing the failure characteristic data related to the internal winding deterioration failure of the electric motor.
7 is a diagram showing analysis results for failure characteristic data on which resampling is not performed in a matrix form.
8 is a diagram showing the analysis result of the resampling performed failure characteristic data in the form of a matrix.
9 is a flowchart of a method for detecting a failure of an electric motor according to a preferred embodiment of the present invention.
FIG. 10 is a diagram for explaining the resampling step of FIG. 9 in detail.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, it should be noted that in adding reference numerals to the components of each drawing, the same components are given the same reference numerals as much as possible even though they are indicated on different drawings. In addition, preferred embodiments of the present invention will be described below, but the technical spirit of the present invention is not limited thereto or may be variously implemented by those skilled in the art without being limited thereto.

1 is a block diagram of an apparatus for detecting a failure of an electric motor according to a preferred embodiment of the present invention.

Referring to Figure 1, the motor failure detection apparatus 100 according to a preferred embodiment of the present invention obtains the phase current signal of the motor, performs resampling on the phase current signal, and fast Fourier transform and scaling for the resampling signal It is characterized in that the failure characteristic data is learned through the fast Fourier transform and scaled preprocessing signal, and the failure of the motor is determined in consideration of the failure characteristic data.

The motor failure detection apparatus 100 includes a receiving unit 110 , a resampling unit 120 , a converting unit 130 , a learning unit 140 , and a determining unit 150 .

The receiver 110 may receive the phase current signal of the electric motor from the sensing device 200 . Here, the sensing device 200 may sense the phase current signal flowing through the winding of the motor. The motor may be, but is not limited to, a surface type permanent magnet synchronous motor. The sensing device 200 may be a non-contact sensing device that is easy to install. In one embodiment, the sensing device 200 may measure the phase current by biting the AC current probe (Probe) to the three-phase current line of the surface type permanent magnet motor.

The resampling unit 120 may be converted into a resampling signal by resampling the phase current signal of the motor. To this end, the resampling unit 120 may convert the phase current signal of the motor into an analytic signal using a previously prepared Hilbert transform. The resampling unit 120 may estimate the instantaneous time from the analysis signal. The resampling unit 120 may detect a change point of the estimated instantaneous time. The resampling unit 120 may detect whether the corresponding change point is greater than or equal to a pre-prepared boundary line. The resampling unit 120 may separate the analysis signal for each period based on the change point when the corresponding change point is greater than or equal to a pre-prepared boundary line. The resampling unit 120 may resample each analysis signal separated for each period into a signal of the same interval through a cubic spline interpolation method. The resampling unit 120 may generate a resampling signal by connecting signals of the same interval to each other. A detailed process and the resampling process of the resampling unit 120 can be confirmed with reference to FIG. 10 .

The transform unit 130 may decompose the resampling signal by extracting it in real time in a sliding window method in order to use it as input data of a convolution-based deep learning algorithm. The transform unit 130 may convert a signal-decomposed resampling signal using a fast Fourier transform (FFT) into a fast Fourier transform signal. In this case, the transform unit 130 may set a scale of the fast Fourier transform signal based on a log unit. To this end, the converter 130 may select a relatively low sampling frequency region. Here, the selected frequency range may be approximately 4 times or less of the electrical angular frequency. The transform unit 130 may perform min-max scaling of the fast Fourier transform signal. In an embodiment, the minimum-maximum scaling is to set a minimum and maximum value of failure characteristic data, which will be described later, and may be set to a value between approximately 0 and 1. Through this, the size of the failure characteristic frequency region having a relatively small intensity compared to the running component of the motor may be increased. Thereafter, the scaled fast Fourier transform signal is defined as a preprocessing signal.

When the preprocessing signal is input to the convolution-based deep learning learning algorithm, the learning unit 140 may learn the failure characteristic data through this. Parameter information used in the convolution-based deep learning learning algorithm may be shown in Table 1 below.

layer name Parameter information number of parameters Convld 1 kernel size= 8, channel : 64 576 BatchNormalization 1 moment=0.01 128 Convld 2 kernel size= 5, channel : 128 41088 BatchNormalization 2 moment=0.01 256 Convld 3 kernel size= 3, channel : 64 24640 BatchNormalization 3 moment=0.01 128 Dense Node number: 3 195 all 67011

In Table 1, the parameters of the final Dense layer are variable according to the number of failure modes to be diagnosed. In one embodiment, the convolution-based deep learning learning algorithm according to the present invention may set a parameter to 3 to classify two failure modes and a normal situation. Here, the two failure modes may include axis misalignment and inter-turn short.

The determination unit 150 may determine whether the electric motor has failed by using the failure characteristic data. The determination unit 150 may interpret the failure determination result through a CAM (Class Activation Map)-based analysis algorithm to provide reliability of the failure determination result. In one embodiment, the failure characteristic data related to the internal winding deterioration failure of the electric motor may be confirmed through FIG. 6 .

Hereinafter, effects that can be obtained through resampling of the phase current of the motor will be described.

2 is a view showing an example of a speed profile of an electric motor.

Referring to FIG. 2 , an example of revolution per minute (RPM) according to the speed profile of the motor can be confirmed. When the motor operates with such a speed profile, the frequency of the phase current of the motor may be changed by the PWM (Pulse Width Modulation) inverter.

3 is a view showing a signal processing result for a phase current of an electric motor operating at a constant speed according to the speed profile of FIG. 2 . The scale unit of FIG. 3 may be logarithmic.

3B is a view showing a general fast Fourier transform result with respect to the phase current of an electric motor operating at a constant speed according to the speed profile of FIG. 2 . That is, if the fast Fourier transform is performed in the frequency domain without additional signal processing for the phase current of the motor, a plurality of frequencies generated by the speed change are mixed with each other as shown in FIG. Here, the failure frequency is composed of a function related to the rotation speed of the electric motor. If the time axis of the phase current is changed to the angle axis through the fast Fourier transform, the fault frequency and the normal frequency of the phase current can be aligned even in various speed changes of the motor.

FIG. 3C is a diagram showing a result of fast Fourier transform of a resampling signal after resampling the phase current of an electric motor operating at a constant speed according to the speed profile of FIG. 2 . The resampling unit 120 may change the phase current data that is previously configured as a function of time and current into a function of the rotation angle and current by resampling the phase current of the motor. At this time, frequency components are gathered according to the rotational speed of the motor, so that it is possible to determine the failure of the motor through the components.

FIG. 4 is a view showing a signal processing result for a phase current of an electric motor operating in a speed change according to the speed profile of FIG. 2 .

FIG. 4A is a view showing a general fast Fourier transform result with respect to the phase current of an electric motor operating in a speed change according to the speed profile of FIG. 2 .

As shown in (a) of FIG. 4 , when the magnitude is expressed on a general scale after performing the fast Fourier transform on the phase current of the motor, since the operating component is the main component, only the operating component has a very large value and other harmonics The surrounding component (Sideband) is relatively close to 0, so the learning performance is poor during deep learning learning.

FIG. 4B is a diagram showing a result of fast Fourier transform of a resampling signal after resampling the phase current of an electric motor operating in a speed change according to the speed profile of FIG. 2 . The scale unit of FIG. 4B may be logarithmic.

As shown in (b) of FIG. 4 , when scaling is performed in logarithmic units on the fast Fourier transform signal, the peripheral frequency and harmonic components, which are smaller in magnitude than the motion component, may be emphasized. Thereafter, for normalization, the magnitude of the fast Fourier transform signal may be re-transformed between 0 and 1 based on the maximum and minimum values of the entire data.

In order to use the preprocessing signal for fault diagnosis, the magnitude and existence of the fault characteristic frequency should be checked. However, there is a problem in that it is difficult to determine where the failure characteristic frequency is located and how large it is to be determined as a failure unless you are an expert in managing the motor. In addition, when the load is changed, the size of the FFT component changes, and it is almost impossible to diagnose the size of all fault components according to the load change.

Accordingly, the apparatus 100 for detecting a failure of an electric motor according to a preferred embodiment of the present invention uses a convolution-based deep learning learning model that automatically extracts characteristics for each data and performs learning in order to solve this point.

5 is a diagram showing the structure of a convolutional neural network of a learning unit according to a preferred embodiment of the present invention.

Referring to FIG. 5, the convolution-based deep learning learning algorithm can learn the frequency characteristics of the phase current through three one-dimensional convolution layers (Conv 1d), and the bottleneck of parameters through the global average pooling layer (GAP). In addition, through FC (Fully Connected Layer), it is possible to classify the failure frequency related to the failure of the motor. Through this, failure characteristic data including the classified failure frequency may be generated.

6 is a view showing the analysis result of analyzing the failure characteristic data related to the internal winding deterioration failure of the electric motor.

Referring to FIG. 6 , the CAM-based analysis algorithm according to the preferred embodiment of the present invention is a 1D CAM algorithm utilizing the existing CAM technique and is configured to interpret the frequency domain of the failure characteristic data. Through this, the CAM-based analysis algorithm can represent the frequency domain centered on the failure characteristic data in red through the heat map, and the rarely utilized part in the failure characteristic data in blue through the heat map. Through this heat map, the user can interpret the judgment result of the CAM-based analysis algorithm, and the reliability of the judgment result in the industrial field can be increased by securing interpretability.

7 is a view showing the analysis result of the failure characteristic data for which resampling is not performed in a matrix form.

Referring to FIG. 7 , the fusion matrix may represent the verification result for the normal and two failure modes (winding deterioration, shaft imbalance) of the motor. At this time, as learning proceeds on the preprocessed signal on which resampling is not performed, the normal state of the matrix data appears as approximately 50%, the inter-turn short state appears as approximately 64%, and the axis imbalance ( It can be seen that the classification performance is about 80% misaligned. That is, it can be seen that the learning and classification performance is deteriorated.

8 is a diagram showing the analysis result of the resampling performed failure characteristic data in the form of a matrix.

Referring to FIG. 8 , the fusion matrix may represent the verification result for the normal and two failure modes (winding deterioration, shaft imbalance) of the motor. At this time, as learning proceeds with respect to the pre-processed signal on which the resampling is performed, it can be confirmed that all of the normal, axis misalign, and inter-turn short states of the matrix data are more than 95% classification performance. . In other words, high learning and verification performance can be confirmed by properly extracting features for each failure frequency.

9 is a flowchart of a method for detecting a failure of an electric motor according to a preferred embodiment of the present invention.

Referring to FIG. 9 , the fault detection method of an electric motor according to a preferred embodiment of the present invention includes a phase current signal acquisition step (S910), a resampling step (S920), a fast Fourier transform step (S930), a scaling step (S940), learning It may include a step (S950), and a failure determination step (S960).

In the phase current signal acquisition step ( S910 ), the receiver 110 may acquire a phase current signal of the motor. Here, the phase current signal of the motor may be measured by a separate sensing device 200 .

In the resampling step (S920), the resampling unit 120 may be converted into a resampling signal by resampling the phase current signal of the motor. The resampling step ( S920 ) will be described in detail later with reference to FIG. 10 .

In the fast Fourier transform step (S930), the transform unit 130 may perform fast Fourier transform on the resampling signal. The transform unit 130 may decompose the resampling signal in a sliding window method prior to the fast Fourier transform. The transform unit 130 may perform fast Fourier transform on the decibel (DB) scale on the signal-decomposed resampling signal.

In the scaling step S940 , the transform unit 130 may perform min-maximum scaling on the size of a relatively low frequency domain of the fast Fourier transform signal. The min-max scaling may be set to a log value between approximately 0 and 1.

In the learning step ( S950 ), the learning unit 140 may learn the scaled preprocessed signal using a pre-prepared convolution-based deep learning learning algorithm. The learning unit 140 may learn the failure characteristic data through the preprocessing signal.

In the failure determination step ( S960 ), the determination unit 150 may determine whether the motor has failed using the failure characteristic data. The determination unit 150 may interpret the failure determination result using a pre-prepared CAM-based analysis algorithm.

FIG. 10 is a diagram for explaining in detail the resampling step of FIG. 9 .

Referring to FIG. 10 , in the resampling step S920 of FIG. 9 , first, the resampling unit 120 may convert the original phase current signal into an analysis signal by using the Hilbert transform. The resampling unit 120 may estimate the instantaneous time Φ(t) from the analysis signal. The resampling unit 120 may separate or decompose the original current signal (analysis signal) for each period (f0, f1) based on the change point of the instantaneous time. The resampling unit 120 may resample each analysis signal separated for each period into a signal of the same interval through a cubic spline interpolation method. The resampling unit 120 may generate a resampling signal by connecting signals of the same interval.

The above description is merely illustrative of the technical idea of the present invention, and various modifications, changes and substitutions are possible within the scope that does not depart from the essential characteristics of the present invention by those of ordinary skill in the art to which the present invention pertains. will be. Accordingly, the embodiments disclosed in the present invention and the accompanying drawings are for explanation rather than limiting the technical spirit of the present invention, and the scope of the technical spirit of the present invention is not limited by these embodiments and the accompanying drawings. .

Steps and/or operations according to the present invention may occur concurrently in different embodiments, either in a different order, or in parallel, or for different epochs, etc., as would be understood by one of ordinary skill in the art. can

Depending on the embodiment, some or all of the steps and/or operations run instructions, programs, interactive data structures, clients and/or servers stored in one or more non-transitory computer-readable media. At least some may be implemented or performed using one or more processors. The one or more non-transitory computer-readable media may be illustratively software, firmware, hardware, and/or any combination thereof. Further, the functionality of a “module” discussed herein may be implemented in software, firmware, hardware, and/or any combination thereof.

100: fault detection device
110: receiver
120: resampling unit
130: conversion unit
140: study unit
150: judgment unit
200: sensing device

Claims (15)

a resampling unit that performs resampling on the phase current signal of the motor and converts it into a resampling signal;
a conversion unit that pre-processes the resampling signal and converts it into a pre-processed signal;
a learning unit for learning failure characteristic data through the preprocessing signal; and
a determination unit configured to determine whether the motor has failed by using the failure characteristic data;
A fault detection device for an electric motor comprising a. The method of claim 1,
The resampling unit,
Converting the phase current signal into an analysis signal, estimating an instantaneous time from the analysis signal, and separating the analysis signal for each period based on the change point of the instantaneous time. 3. The method of claim 2,
The resampling unit,
Resampling the analysis signal separated for each period through a pre-prepared interpolation method into a signal of the same interval, and generating the resampling signal by connecting the signals of the same interval. The method of claim 1,
The conversion unit,
A failure detection device for a motor, characterized in that performing a log-scale fast Fourier transform on the resampling signal. 5. The method of claim 4,
The conversion unit,
The apparatus for detecting a failure of an electric motor, characterized in that the pre-processing signal is generated by scaling the magnitude of the signal according to the fast Fourier transform. 6. The method of claim 5,
The scaling unit is an apparatus for detecting a failure of an electric motor, characterized in that it has a value of 0 to 1. The method of claim 1,
The learning unit,
A failure detection device for a motor, characterized in that it learns the pre-processed signal using a pre-prepared convolution-based deep learning learning algorithm. The method of claim 1,
The judging unit,
An apparatus for detecting a failure of an electric motor, characterized in that it determines an axial imbalance or a winding deterioration failure of the electric motor by using the failure characteristic data. 9. The method of claim 8,
The judging unit,
A failure detection apparatus of a motor, characterized in that the analysis of the failure determination result for the motor using a pre-prepared CAM (Class Activation Map) based analysis algorithm. A resampling step of performing resampling on the phase current signal of the motor and converting it into a resampling signal;
a pre-processing step of pre-processing the resampling signal and converting it into a pre-processing signal;
a learning step of learning failure characteristic data through the preprocessing signal; and
a failure determination step of determining whether the electric motor has failed using the failure characteristic data;
A fault detection method of an electric motor comprising a. 11. The method of claim 10,
The resampling step is
converting the phase current signal into an analysis signal;
estimating the instantaneous time from the analysis signal;
separating the analysis signal for each period based on the change point of the instantaneous time;
Resampling the analysis signal separated by period through a pre-prepared interpolation method into a signal of the same interval; and
generating the resampling signal by connecting the signals of the same interval;
A fault detection method of an electric motor comprising a. 11. The method of claim 10,
The pre-processing step is
A fast Fourier transform step of performing a log scale fast Fourier transform on the resampling signal; and
a scaling step of scaling a signal according to the fast Fourier transform;
A fault detection method of an electric motor comprising a. 11. The method of claim 10,
The learning step is
A failure detection method of an electric motor, characterized in that the pre-processing signal is learned using a pre-prepared convolution-based deep learning learning algorithm. 11. The method of claim 10,
The failure determination step is
A failure detection method of an electric motor, characterized in that for determining an axial imbalance or winding deterioration failure of the electric motor by using the failure characteristic data. 11. The method of claim 10,
The failure determination step is
A failure detection method of a motor, characterized in that the analysis of the failure determination result of the motor using a pre-prepared CAM (Class Activation Map) based analysis algorithm.

Images