CN109934100B - Amplitude-frequency time-varying process signal segmentation method based on sliding window - Google Patents
Amplitude-frequency time-varying process signal segmentation method based on sliding window Download PDFInfo
- Publication number
- CN109934100B CN109934100B CN201910069256.0A CN201910069256A CN109934100B CN 109934100 B CN109934100 B CN 109934100B CN 201910069256 A CN201910069256 A CN 201910069256A CN 109934100 B CN109934100 B CN 109934100B
- Authority
- CN
- China
- Prior art keywords
- amplitude
- frequency time
- segmentation
- sliding
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention discloses a sliding window based amplitude-frequency time-varying process signal segmentation method, which comprises the following steps: (1) setting a sliding window for an amplitude-frequency time-varying process signal to be analyzed; (2) calculating the average value of each process signal point in the sliding window, and dynamically setting a lower segmentation threshold point of a process curve according to the average value so as to obtain a lower segmentation line; (3) determining each lower segmentation point of the process curve according to the lower segmentation line, and segmenting the process curve according to the lower segmentation points; (4) and (4) solving the similarity of the segmentation curves by using the process signals in the segmentation curves, and judging the segmentation effectiveness. The invention divides the process period of the amplitude-frequency time-varying process curve, can adapt to the change of process parameters and has stronger adaptivity, stability and robustness.
Description
Technical Field
The invention relates to the technical field of process signal segmentation, in particular to a sliding window-based amplitude-frequency time-varying process signal segmentation method.
Background
In the field of fault diagnosis and equipment state monitoring, extraction of characteristic values of a process curve is relied on, and one way of characteristic value extraction is to segment the process curve and then extract characteristics of the segmented curve. Therefore, the process curve segmentation algorithm is the basis for equipment fault diagnosis and operation state monitoring.
At present, the research on the process curve segmentation algorithm of equipment operation is many, and the general method comprises the following steps: firstly, defining functional segment division in a signal period according to extreme points, monotonicity or actual meanings of signals, and then extracting signal division points by using time relations among the signal division points and extreme value characteristics of the signal division points and using methods of time domain analysis, cepstrum analysis, wavelet transformation and the like, thereby realizing the purpose of signal division.
However, in actual production and processing, the process parameters are changed in real time, so that the process curve of the equipment operation is a variable process parameter curve, the amplitude and the frequency of the process curve can be changed along with the change of the process parameters, the existing signal segmentation method cannot adapt to the process signals with changed periods and changed extreme values, and errors often occur.
Disclosure of Invention
The invention aims to provide a sliding window based amplitude-frequency time-varying process signal segmentation method aiming at the defects and shortcomings of the prior art.
In order to achieve the purpose, the invention adopts the technical scheme that:
1) setting a sliding window, and determining the window size and the sliding step length of the sliding window according to the process period of the amplitude-frequency-varying process curve;
2) traversing the amplitude-frequency time-varying process curve by using the sliding window set in the step 1) and calculating a sliding mean value, wherein the sliding mean value is a mean value corresponding to each signal point in the sliding window, setting a lower segmentation threshold point of the amplitude-frequency time-varying process curve according to the sliding mean value, and setting a lower segmentation line capable of continuously segmenting the amplitude-frequency time-varying process curve by using the lower segmentation threshold point;
3) and determining a lower division point of the amplitude-frequency time-varying process curve according to the minimum value point of the amplitude-frequency time-varying process curve signal below the lower division line, and performing process period division on the amplitude-frequency time-varying process curve according to the lower division point of the amplitude-frequency time-varying process curve.
Preferably, in the step 1), the size of the sliding window is 1.5-3 times of the process period, and the sliding step length of the sliding window is 1-5.
Preferably, in step 2), the position of the lower segmentation threshold point in the sampling time dimension is the center of the sliding window, and the position of the lower segmentation threshold point in the signal value dimension is equal to the sliding mean and ΔtDifference of (a)tIs the peak value and coefficient beta of the amplitude-frequency time-varying process curvetThe product of (a).
Preferably, said betatThe value range of (A) is 0.1-0.3.
Preferably, in the step 2), all the lower segmentation threshold points are connected to obtain a lower segmentation line.
Preferably, in the step 3), the determining of the lower segmentation point specifically includes the following steps: and extracting curve sections below the lower dividing line in the amplitude-frequency time-varying process curve, and determining the corresponding signal minimum value point on each section of curve section through traversal.
Preferably, the determining of the lower segmentation point further comprises the steps of: comparing the sampling interval distance of the adjacent signal minimum value points with a set minimum distance threshold, if the distance of the adjacent signal minimum value points is greater than the threshold, reserving the signal minimum value points, otherwise, discarding the signal minimum value points with the later sampling time, and taking all the reserved signal minimum value points as lower division points, wherein the minimum distance threshold is 2-10.
Preferably, the process signal dividing method further includes the steps of: and 3) solving the similarity of the corresponding segmentation curves for the process cycle signals segmented in the step 3), and screening effective process cycle signals according to the similarity.
Preferably, the similarity of the segmentation curve is obtained by a dynamic time warping algorithm (DTW), the similarity value of each segment of the segmented process cycle signal is obtained by using the algorithm and referring to a specified standard process cycle signal, and if the similarity value is greater than a set threshold (i.e. the shape is similar), the segment of the segmented process cycle signal is retained; otherwise, abandoning the divided process period signal.
Preferably, the standard process cycle signal is specified from all of the divided process cycle signals.
The invention has the beneficial effects that:
the method dynamically obtains the lower segmentation threshold value according to the sliding mean value of the sliding window aiming at the amplitude-frequency time-varying process curve, determines the lower segmentation line and the lower segmentation point, further performs periodic segmentation on the process curve, can adapt to the change of process parameters, and has strong adaptivity, stability and robustness. The invention has universality aiming at the segmentation of the process signals of which various periods and extreme values change along with the process parameters, can adapt to the change conditions of various signals in the process and realizes simple, convenient, rapid and effective segmentation of the process period signals.
Furthermore, the method carries out evaluation on the segmentation effect according to the similarity of the segmented curve, can better adapt to the change of the amplitude and the frequency of the process curve, and improves the adaptivity, the stability and the robustness of the segmentation result.
Drawings
FIG. 1 is a flow chart of a sliding window based amplitude-frequency time-varying process signal segmentation method in the present invention.
FIG. 2 is a process curve for normal operation of an intelligent device.
FIG. 3 shows the process curve segmentation result of an intelligent device.
Detailed Description
The invention is further described below with reference to the figures and examples. The examples are given solely for the purpose of illustration and are not intended to limit the scope of the invention.
As shown in fig. 1, the flow of the amplitude-frequency time-varying process signal segmentation method based on the sliding window is as follows:
(1) setting a sliding window for an amplitude-frequency time-varying process signal to be analyzed, and determining the size and the sliding step length of the window;
defining a process curve as shown in FIG. 2 with vector X ═ X1,x2,…,xi,…,xn]TRepresenting a data set acquired in the process, namely a process curve is composed of n original signal points (the signal points are sequentially connected into the process curve according to sampling time); the sliding window size N is determined to be 2.2 times the process cycle of certain intelligent equipment (e.g., a fast forging press) with a sliding step size of 1.
(2) Traversing signal points in the window, calculating the sliding mean value of the sliding window (the sliding mean value is the mean value of each point of the process curve in the sliding window), and setting the lower segmentation threshold point delta of the process curve according to the sliding mean valueb(i)Connecting all the lower segmentation threshold points to form a lower segmentation line; the lower segmentation threshold point is specifically obtained by reducing the sliding mean by a small amount deltatCalculated as the peak-to-peak value (Max (X) -Min (X)) of the process curve multiplied by a factor (beta)t) The result of (a), namely:
Δt＝βt(Max(X)-Min(X))
wherein X (i) is a middle point of the sliding window at the current position of the process curve, betatTake 0.2.
As can be seen from fig. 2, the obtained lower dividing line passes through each segment of the process curve, and the starting point and the ending point of each process cycle can be found according to the lower dividing line.
(3) The key of the process curve segmentation is that a process segmentation point (called as a lower segmentation point in the invention) is accurately found;
traversing all original signal points, if the original signal points are lower than the lower dividing line, keeping and adding the original signal points to a minimum judgment sequence, and if the original signal points are not lower than the lower dividing line, giving up the minimum judgment sequence; the lower division point of the process curve is the minimum value point SL of any interval below the lower division linepb：
Wherein the content of the first and second substances,
respectively representing the front and rear endpoints of the interval.
(4) Setting a minimum distance threshold value, and judging the distance between adjacent minimum value points (namely the distance of the abscissa between two points); if the distance is greater than the minimum distance threshold, adding the distance into the segmentation point sequence, otherwise, discarding the point with the larger abscissa, wherein the minimum distance threshold is 4.
(5) Carrying out process curve period division based on the division points; referring to fig. 2, the division of the process curve is performed every p (e.g., p ═ 2) lower division points from the designated first lower division point.
(6) Calculating the similarity of the segmentation curves by using the segmented process period signals, and judging the segmentation effectiveness;
the specific judgment process is as follows: designating the first process periodic signal in fig. 3 as a standard process periodic signal, performing DTW similarity solution on the segmentation curve segment by segment, and retaining the segmentation curve with high similarity according to a set threshold;
xxj＝[xx1,…,xxj,…,xxm(j)]and (3) a signal data set (k signal points) representing the j-th divided effective process period of the process curve, wherein j is 1, … and m. As can be seen,
ss is a process curveSampling start point, Se as process curve sampling end point, mSs(j)Starting point index number, m, for a process curveSe(j)Index number of process curve end point; and has:
as shown in fig. 2, the normal operating curve of an intelligent device exhibits a certain periodicity, but the amplitude and frequency are not fixed, but vary with the process parameters. The result obtained by applying the segmentation method of the present invention to the curve is shown in fig. 3, and through similarity determination, there is no segment to be discarded, and the segmentation result is ideal.
In a word, the amplitude-frequency time-varying process signal segmentation method based on the sliding window determines a lower segmentation threshold point according to a sliding mean value by arranging the sliding window, connects the lower segmentation threshold point to a lower segmentation line, performs process period segmentation on a process curve according to the lower segmentation point, and can also perform segmentation effect evaluation according to the similarity of the curve to obtain a better segmentation effect.
Claims (8)
1. A method for dividing amplitude-frequency time-varying process signals based on a sliding window is characterized in that: the process signal segmentation method comprises the following steps:
1) setting a sliding window, and determining the window size and the sliding step length of the sliding window according to the process period of the amplitude-frequency-varying process curve;
2) traversing the amplitude-frequency time-varying process curve by using the sliding window set in the step 1) and calculating a sliding mean value, wherein the sliding mean value is a mean value corresponding to each signal point in the sliding window, setting a lower segmentation threshold point of the amplitude-frequency time-varying process curve according to the sliding mean value, and setting a lower segmentation line capable of continuously segmenting the amplitude-frequency time-varying process curve by using the lower segmentation threshold point;
3) and determining a lower division point of the amplitude-frequency time-varying process curve according to the minimum value point of the amplitude-frequency time-varying process curve signal below the lower division line, and performing process period division on the amplitude-frequency time-varying process curve according to the lower division point of the amplitude-frequency time-varying process curve.
2. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 1, wherein: in the step 2), the position of the lower segmentation threshold point in the sampling time dimension is the center of the sliding window, and the position in the signal value dimension is equal to the sliding mean and deltatDifference of (a)tIs the peak value and coefficient beta of the amplitude-frequency time-varying process curvetThe product of (a).
3. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 2, wherein: beta is the same astThe value range of (A) is 0.1-0.3.
4. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 1 or 2, wherein: in the step 2), all the lower segmentation threshold points are connected to obtain a lower segmentation line.
5. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 1, wherein: in the step 3), the determination of the lower segmentation point specifically includes the following steps: and extracting curve sections below the lower dividing line in the amplitude-frequency time-varying process curve, and determining the corresponding signal minimum value point on each section of curve section through traversal.
6. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 1, wherein: the process signal dividing method further comprises the following steps: and 3) solving the similarity of the corresponding segmentation curves for the process cycle signals segmented in the step 3), and screening effective process cycle signals according to the similarity.
7. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 6, wherein: the similarity of the segmentation curves is obtained by adopting a dynamic time warping algorithm, the similarity value of each segment of the segmented process cycle signals is obtained by utilizing the algorithm and referring to the specified standard process cycle signals, and if the similarity value is greater than a set threshold value, the segmented segment of the process cycle signals are reserved; otherwise, abandoning the divided process period signal.
8. The sliding-window-based amplitude-frequency time-varying process signal segmentation method as claimed in claim 7, wherein: the standard process cycle signal is specified from all of the divided process cycle signals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910069256.0A CN109934100B (en) | 2019-01-24 | 2019-01-24 | Amplitude-frequency time-varying process signal segmentation method based on sliding window |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910069256.0A CN109934100B (en) | 2019-01-24 | 2019-01-24 | Amplitude-frequency time-varying process signal segmentation method based on sliding window |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109934100A CN109934100A (en) | 2019-06-25 |
| CN109934100B true CN109934100B (en) | 2021-07-13 |
Family
ID=66985198
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910069256.0A Active CN109934100B (en) | 2019-01-24 | 2019-01-24 | Amplitude-frequency time-varying process signal segmentation method based on sliding window |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109934100B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111080647B (en) * | 2019-11-26 | 2022-03-04 | 西安电子科技大学 | SAR image segmentation method based on adaptive sliding window filtering and FCM |
| CN111461081B (en) * | 2020-05-18 | 2022-01-18 | 江苏电力信息技术有限公司 | Current signal segmentation method |
| CN111738086B (en) * | 2020-05-22 | 2022-08-30 | 上海交通大学 | Composition method and system for point cloud segmentation and point cloud segmentation system and device |
| CN115659162B (en) * | 2022-09-15 | 2023-10-03 | 云南财经大学 | Method, system and equipment for extracting intra-pulse characteristics of radar radiation source signals |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1398118A (en) * | 2002-08-16 | 2003-02-19 | 清华大学 | Method based on slide window for estimating and equalizing channels of block signals containing pilot |
| CN101897578A (en) * | 2010-06-30 | 2010-12-01 | 重庆大学 | Method for segmenting arterial pressure signal by beats |
| CN107233093A (en) * | 2017-05-23 | 2017-10-10 | 安徽华米信息科技有限公司 | Detection method, device and the electronic equipment of R ripples |
| CN107527022A (en) * | 2017-08-04 | 2017-12-29 | 丁大良 | Signal period computational methods |
| CN108375472A (en) * | 2018-02-12 | 2018-08-07 | 武汉科技大学 | Based on the Method for Bearing Fault Diagnosis and system and device for improving experience wavelet transformation |
| CN108723895A (en) * | 2018-05-25 | 2018-11-02 | 湘潭大学 | A kind of signal dividing method monitored in real time for drilling machining state |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101866423B (en) * | 2010-05-12 | 2012-07-25 | 重庆大学 | Method for dividing arterial pressure signal beat by beat |
| CN103027667B (en) * | 2011-09-30 | 2017-01-18 | Ge医疗系统环球技术有限公司 | Characteristic parameter extraction of pulse wave |
| US9568513B2 (en) * | 2013-02-14 | 2017-02-14 | Schweitzer Engineering Laboratories, Inc. | Systems and methods to detect poorly damped oscillation modes |
| KR102026069B1 (en) * | 2013-08-05 | 2019-09-30 | 삼성전자 주식회사 | Segmentation system of sensor data in semiconductor manufacturing equipment and method thereof |
| CN105227763B (en) * | 2015-08-31 | 2018-03-20 | 武汉工程大学 | A kind of instrumental audio real time method for segmenting realized on Intelligent mobile equipment |
| CN106473750B (en) * | 2016-10-08 | 2019-03-26 | 西安电子科技大学 | Personal identification method based on photoplethysmographic optimal period waveform |
| CN107616791B (en) * | 2017-07-26 | 2021-02-19 | 浙江大学 | Standard 12-lead electrocardio piecewise linear reconstruction method based on self-adaptive electrocardiosignal region segmentation |
-
2019
- 2019-01-24 CN CN201910069256.0A patent/CN109934100B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1398118A (en) * | 2002-08-16 | 2003-02-19 | 清华大学 | Method based on slide window for estimating and equalizing channels of block signals containing pilot |
| CN101897578A (en) * | 2010-06-30 | 2010-12-01 | 重庆大学 | Method for segmenting arterial pressure signal by beats |
| CN107233093A (en) * | 2017-05-23 | 2017-10-10 | 安徽华米信息科技有限公司 | Detection method, device and the electronic equipment of R ripples |
| CN107527022A (en) * | 2017-08-04 | 2017-12-29 | 丁大良 | Signal period computational methods |
| CN108375472A (en) * | 2018-02-12 | 2018-08-07 | 武汉科技大学 | Based on the Method for Bearing Fault Diagnosis and system and device for improving experience wavelet transformation |
| CN108723895A (en) * | 2018-05-25 | 2018-11-02 | 湘潭大学 | A kind of signal dividing method monitored in real time for drilling machining state |
Non-Patent Citations (1)
| Title |
|---|
| 基于时域线性插值的信号周期计算方法及误差分析;周峰 等;《仪器仪表学报》;20110831;第32卷(第8期);1724-1730 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109934100A (en) | 2019-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109934100B (en) | Amplitude-frequency time-varying process signal segmentation method based on sliding window | |
| CN108804731B (en) | Time series trend feature extraction method based on important point dual evaluation factors | |
| CN106781486B (en) | Traffic state evaluation method based on floating car data | |
| CN108009122B (en) | Improved HHT method | |
| CN105139366A (en) | Image enhancement method based on space domain | |
| CN106653062A (en) | Spectrum-entropy improvement based speech endpoint detection method in low signal-to-noise ratio environment | |
| CN107688553B (en) | Method for detecting electrocardiographic waveform characteristics based on wavelet transform and logistic regression algorithm | |
| CN110775758B (en) | Elevator running health degree evaluation method based on car acceleration signal analysis | |
| CN113378661A (en) | Direct current electric energy signal denoising method based on improved wavelet threshold and related detection | |
| CN104991687A (en) | Method and system for acquiring curve operating track of touch-screen device | |
| CN108847252B (en) | Acoustic feature extraction method based on acoustic signal spectrogram texture distribution | |
| US20170075490A1 (en) | Method and device of optimizing signal-to-noise ratio parameter | |
| CN106228170A (en) | A kind of pointer position extracting method of pointer instrument | |
| CN113296154A (en) | Method for predicting base expansion frequency modulation synchronous extrusion seismic reservoir | |
| CN108226890B (en) | Airport foreign matter radar detection method based on time direction statistics | |
| CN106683405B (en) | Road traffic state evaluation method based on road traffic density | |
| CN114186596B (en) | Multi-window identification method and device for spectrogram peaks and electronic equipment | |
| CN110414442B (en) | Pressure time sequence data segmentation characteristic value prediction method | |
| CN109490776B (en) | Mobile phone vibration motor good and defective product detection method based on machine learning | |
| Azami et al. | Automatic signal segmentation using the fractal dimension and weighted moving average filter | |
| CN109544593B (en) | General image noise estimation method based on self-similarity measurement | |
| CN113567127B (en) | Rolling bearing degradation index extraction method based on time-frequency feature separation | |
| CN114114400B (en) | Microseism event effective signal pickup method | |
| CN115670397A (en) | PPG artifact identification method and device, storage medium and electronic equipment | |
| CN110490297B (en) | Intelligent segmentation method for railway turnout action power curve |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |