CN111160454A - Method and device for detecting speed change signal - Google Patents
Method and device for detecting speed change signal Download PDFInfo
- Publication number
- CN111160454A CN111160454A CN201911379540.4A CN201911379540A CN111160454A CN 111160454 A CN111160454 A CN 111160454A CN 201911379540 A CN201911379540 A CN 201911379540A CN 111160454 A CN111160454 A CN 111160454A
- Authority
- CN
- China
- Prior art keywords
- determining
- speed change
- working condition
- data
- neural network
- 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.)
- Granted
Links
- 230000008859 change Effects 0.000 title claims abstract description 155
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000003062 neural network model Methods 0.000 claims abstract description 124
- 238000001514 detection method Methods 0.000 claims abstract description 34
- 239000011159 matrix material Substances 0.000 claims description 152
- 238000012549 training Methods 0.000 claims description 105
- 230000002159 abnormal effect Effects 0.000 claims description 88
- 238000012360 testing method Methods 0.000 claims description 35
- 238000013528 artificial neural network Methods 0.000 claims description 24
- 238000000605 extraction Methods 0.000 claims description 16
- 230000009191 jumping Effects 0.000 claims description 12
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 description 17
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 11
- 229910052760 oxygen Inorganic materials 0.000 description 11
- 239000001301 oxygen Substances 0.000 description 11
- 238000002474 experimental method Methods 0.000 description 10
- 238000010801 machine learning Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 4
- 230000037433 frameshift Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 241000282414 Homo sapiens Species 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Educational Administration (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Biomedical Technology (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Probability & Statistics with Applications (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method for detecting a speed change signal, which comprises the following steps: acquiring a speed change signal generated by equipment to be detected, and determining characteristic data to be detected according to the acquired speed change signal; inputting the determined characteristic data to be detected into a pre-trained first probability neural network model, and determining the working condition of the equipment to be detected; determining a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested; and inputting the determined characteristic data to be detected into the determined second probabilistic neural network model, and determining the state of the speed change signal. The invention also discloses a speed change signal detection device.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method for detecting abnormal speed change signals based on machine learning.
Background
The PHM (fault prediction and health management) is a system that uses various advanced sensors to monitor equipment operating state parameters and characteristic signals in real time, evaluates equipment health states with the help of intelligent algorithms and models, predicts remaining service life, diagnoses fault types, and provides a series of fault maintenance decisions before a fault occurs. The PHM technology is a product of the combination of advanced diagnosis technology and testing technology equipment maintenance management theory. The type of fault is identified by the fault diagnosis capability of this technique. The equipment user and the maintenance personnel can clearly determine the equipment fault type, so that an effective maintenance mode is adopted, the fault risk is effectively reduced, the equipment resource is saved, and the economic loss caused by misdiagnosis of equipment faults is reduced.
At present, the PHM technology mainly includes the following methods for implementing process anomaly detection of equipment speed change signals: the anomaly detection method based on the support vector machine is based on a particle filter method, a gray model detection method and a neural network detection method.
The abnormity detection method based on the support vector machine has the defects that: the method is difficult to realize for large-scale training samples and difficult to solve various problems. The particle filter based method has the defects that: the method has no capability of coping with interference change, and filtering divergence is easy to occur to uncertain disturbance, so that the detection precision is reduced. The gray model-based detection method has the defects that: the periodicity and various trend characteristics of the original time sequence are easily damaged, and the robustness of the model is reduced.
Compared with the methods, the neural network detection method based on the neural network detection method has the advantages that: the method has the advantages of strong nonlinear problem processing capability, distributed information storage, parallel processing, adaptive learning and the like.
However, the detection result of the speed change signal is susceptible to the change of the working condition, and the characteristic of the abrupt change of the time sequence of the procedural speed change signal increases the difficulty of detecting the abnormal signal. The existing detection method is difficult to carry out effective signal abnormality detection on the existing detection method. There is therefore a need to improve existing algorithms or to develop new algorithms to address this difficult problem.
Disclosure of Invention
In order to solve the technical problems, the invention provides a machine learning-based speed change signal detection method, which can accurately detect process signal abnormality, effectively avoid falling into local optimum, has strong inference capability, and can avoid the problems of detection precision and model robustness reduction.
The invention provides a method for detecting a speed change signal, which comprises the following steps:
acquiring a speed change signal generated by equipment to be detected, and determining characteristic data to be detected according to the acquired speed change signal;
inputting the determined characteristic data to be detected into a pre-trained first probability neural network model, and determining the working condition of the equipment to be detected;
determining a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested;
and inputting the determined characteristic data to be detected into the determined second probabilistic neural network model, and determining the state of the speed change signal.
Optionally, the determining, according to the working condition of the device under test, a corresponding second probabilistic neural network model includes:
according to the working condition of the equipment to be tested, determining a normal feature matrix by using normal sample data under the working condition; determining an abnormal feature matrix by using the abnormal sample data under the working condition; inputting the normal characteristic matrix and the working condition label thereof, and the abnormal characteristic matrix and the working condition label thereof into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model;
or, the determining a corresponding second probabilistic neural network model according to the working condition of the device under test includes:
selecting a second probabilistic neural network model of the working condition from at least one second probabilistic neural network model which is trained in advance according to the working condition of the equipment to be tested,
wherein the at least one pre-trained second probabilistic neural network model comprises: according to different working conditions, respectively executing the following steps to obtain second probabilistic neural network models with different working conditions:
determining a normal characteristic matrix by using normal sample data under the current working condition; and determining an abnormal characteristic matrix by using the abnormal sample data under the current working condition, inputting the normal characteristic matrix, the current working condition label, the abnormal characteristic matrix and the current working condition label into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model.
Optionally, wherein the inputting the determined feature data to be detected into the determined second probabilistic neural network to determine the state of the speed change signal includes:
and inputting the characteristic data to be detected into a determined second probabilistic neural network, and determining the state of the speed change signal according to the training model parameters of the second probabilistic neural network model.
Optionally, wherein the first probabilistic neural network model is trained in advance according to the following method:
collecting corresponding speed change signals generated by the equipment to be tested under different working conditions in advance;
extracting normal signal data as normal sample data and extracting abnormal signal data as abnormal sample data;
the following treatment is respectively carried out on different working conditions: determining a normal feature matrix under the working condition according to the normal sample data under the working condition, and determining an abnormal feature matrix under the working condition according to the abnormal sample data under the working condition;
inputting the first probability neural network model for training according to the normal feature matrix and the corresponding label under different working conditions, and the abnormal feature matrix and the corresponding label under different working conditions, and determining the training model parameters of the first probability neural network model.
Optionally, the determining the feature data to be detected according to the acquired speed change signal includes:
determining whether more than one corresponding working condition exists according to the acquired speed change signals;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
Optionally, wherein the determining the normal feature matrix includes:
according to a preset first moving window width and a first time shifting step number, carrying out time-sharing time shifting on the normal sample data to obtain a first window time shifting matrix, and carrying out cloud feature extraction on the first window time shifting matrix to obtain the normal feature matrix;
wherein the determining the abnormal feature matrix comprises:
and according to a preset second moving window width and a second time shifting step number, carrying out time-sharing time shifting on the abnormal sample data to obtain a second sub-window time shifting matrix, and carrying out cloud feature extraction on the second sub-window time shifting matrix to obtain the abnormal feature matrix.
Optionally, the determining the feature data to be detected includes:
and according to a preset third moving window width and a third time shifting step number, carrying out time-sharing time shifting on the speed change signal to obtain a third window time shifting matrix, and carrying out cloud feature extraction on the third window time shifting matrix to obtain a data feature matrix to be detected.
Optionally, the determining, according to the acquired speed change signal, whether more than one corresponding operating condition exists includes:
according to a preset jump threshold, searching a data sequence number position whether the absolute value of the change value between two adjacent data points is greater than the jump threshold from the acquired speed change signal, and recording the position as a data signal jump point;
if the acquired speed change signal data does not have the data signal jumping points, determining that the working condition corresponding to the acquired speed change signal only comprises one type;
if the acquired speed change signals have such data signal jumping points, determining that the working conditions corresponding to the acquired speed change signals comprise more than one; wherein the dividing the acquired rate-change signals into a corresponding plurality of more than one rate-change signal subsets comprises: and carrying out data segmentation on the acquired speed change signals according to the positions of the jumping points to obtain more than one speed change signal subset.
The present invention also provides a speed change signal detection apparatus, comprising:
the signal acquisition module is used for acquiring a speed change signal generated by the equipment to be detected and determining the characteristic data to be detected according to the acquired speed change signal;
the working condition determining module is used for inputting the determined characteristic data to be detected into a first probability neural network model which is trained in advance and determining the working condition of the equipment to be detected;
the second probabilistic neural network model determining module is set to determine a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested;
and the signal state determining module is used for inputting the determined characteristic data to be detected into the determined second probabilistic neural network model and determining the state of the speed change signal.
Optionally, the acquisition module is further configured to:
determining whether more than one corresponding working condition is available according to the speed change signal acquired by the signal;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
Drawings
Fig. 1 is a flowchart of a method for detecting a speed change signal according to an embodiment;
fig. 2 is a flowchart of a method for detecting a speed change signal according to a second embodiment;
fig. 3 is a flowchart of a method for detecting a speed change signal according to a second embodiment;
FIG. 4 is a diagram of a probabilistic neural network model according to a second embodiment;
fig. 5 shows a structure diagram of a detecting apparatus for detecting a speed change signal according to a third embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be arbitrarily combined with each other without conflict.
First, the relevant definitions in the art are described below, but not limited to the following individual cases:
machine learning: the method is a multi-field cross discipline and relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. The machine learning in the related art is classified from the perspective of learning strategies as: machine learning simulating the human brain and machine learning using mathematical methods. The machine learning by mathematics mainly comprises statistical machine learning, and is characterized in that an appropriate mathematical model is selected based on preliminary knowledge of data and analysis of learning purposes, parameters are formulated, sample data is input, the model is trained by using an appropriate learning algorithm according to a certain strategy, and finally the trained model is used for analyzing and predicting the data.
A probabilistic neural network: (PNN) is a Neural network which is proposed by D.F. Specht in 1989, has a simple structure and is widely applied to the problem of mode classification.
Example one
The invention provides a method for detecting a speed change signal, the flow of which is shown in figure 1, and the method comprises the following steps:
step 101, acquiring a speed change signal generated by equipment to be detected, and determining characteristic data to be detected according to the acquired speed change signal;
step 102, inputting the determined characteristic data to be detected into a first probability neural network model trained in advance, and determining the working condition of the equipment to be detected;
103, determining a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested;
and 104, inputting the determined characteristic data to be detected into the determined second probabilistic neural network model, and determining the state of the speed change signal.
Optionally, in step 103, the determining a corresponding second probabilistic neural network model according to the working condition of the device under test includes:
according to the working condition of the equipment to be tested, determining a normal feature matrix by using normal sample data under the working condition; determining an abnormal feature matrix by using the abnormal sample data under the working condition; inputting the normal characteristic matrix and the working condition label thereof, and the abnormal characteristic matrix and the working condition label thereof into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model;
optionally, in step 103, the determining a corresponding second probabilistic neural network model according to the working condition of the device under test includes:
selecting a second probabilistic neural network model of the working condition from at least one second probabilistic neural network model which is trained in advance according to the working condition of the equipment to be tested,
wherein the at least one pre-trained second probabilistic neural network model comprises: according to different working conditions, respectively executing the following steps to obtain second probabilistic neural network models with different working conditions:
determining a normal characteristic matrix by using normal sample data under the current working condition; and determining an abnormal characteristic matrix by using the abnormal sample data under the current working condition, inputting the normal characteristic matrix, the current working condition label, the abnormal characteristic matrix and the current working condition label into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model.
Optionally, in step 104, the inputting the determined feature data to be detected into the determined second probabilistic neural network to determine the state of the speed change signal includes:
and inputting the characteristic data to be detected into a determined second probabilistic neural network, and determining the state of the speed change signal according to the training model parameters of the second probabilistic neural network model.
Optionally, the first probabilistic neural network model in step 102 is obtained by training in advance according to the following method:
collecting corresponding speed change signals generated by the equipment to be tested under different working conditions in advance;
extracting normal signal data as normal sample data and extracting abnormal signal data as abnormal sample data;
the following treatment is respectively carried out on different working conditions: determining a normal feature matrix under the working condition according to the normal sample data under the working condition, and determining an abnormal feature matrix under the working condition according to the abnormal sample data under the working condition;
and inputting the first probabilistic neural network model for training according to the normal feature matrix and the corresponding label under different working conditions, and the abnormal feature matrix and the corresponding label under different working conditions, and determining the training model parameters of the first probabilistic neural network model, namely finishing the training of the first probabilistic neural network model.
Optionally, in step 101, the determining, according to the acquired speed change signal, feature data to be detected includes:
determining whether more than one corresponding working condition exists according to the acquired speed change signals;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
When there are more than one working conditions, the collected speed change signal is divided into more than one corresponding speed change signal data subsets, and then the steps 101 and 104 are respectively executed for each speed change signal data subset.
Optionally, the determining a normal feature matrix includes:
according to a preset first moving window width and a first time shifting step number, carrying out time-sharing time shifting on the normal sample data to obtain a first window time shifting matrix, and carrying out cloud feature extraction on the first window time shifting matrix to obtain the normal feature matrix;
the determining of the abnormal feature matrix comprises:
and according to a preset second moving window width and a second time shifting step number, carrying out time-sharing time shifting on the abnormal sample data to obtain a second sub-window time shifting matrix, and carrying out cloud feature extraction on the second sub-window time shifting matrix to obtain the abnormal feature matrix.
Optionally, the determining the feature data to be detected includes:
and according to a preset third moving window width and a third time shifting step number, carrying out time-sharing time shifting on the speed change signal to obtain a third window time shifting matrix, and carrying out cloud feature extraction on the third window time shifting matrix to obtain a data feature matrix to be detected.
Optionally, the determining, according to the acquired speed change signal, whether more than one corresponding operating condition exists includes:
according to a preset jump threshold, searching a data sequence number position whether the absolute value of the change value between two adjacent data points is greater than the jump threshold from the acquired speed change signal, and recording the position as a data signal jump point;
if the acquired speed change signal data does not have the data signal jumping points, determining that the working condition corresponding to the acquired speed change signal only comprises one type;
if the acquired speed change signals have such data signal jumping points, determining that the working conditions corresponding to the acquired speed change signals comprise more than one; wherein the dividing the acquired rate-change signals into a corresponding plurality of more than one rate-change signal subsets comprises: and carrying out data segmentation on the acquired speed change signals according to the positions of the jumping points to obtain more than one speed change signal subset.
Example two
The invention provides a method for detecting a speed change signal, the flow of which is shown in figure 2,
the embodiment of the invention takes the detection of the oxygen pump vibration signal of the engine as an example, and the oxygen pump vibration signal of the engine is a speed change signal. The method for acquiring the signal is implemented according to the related technical scheme in the field, and is not limited to a specific scheme.
Step 201: acquiring a speed change signal generated by equipment to be detected, and determining characteristic data to be detected according to the acquired speed change signal;
step 202: inputting the determined characteristic data to be detected into a pre-trained first probability neural network model, and determining the working condition of the equipment to be detected;
step 203: training a second neural network model according to the working condition of the equipment to be tested;
step 204: and inputting the determined characteristic data to be detected into a trained second neural network model, and determining the state of the speed change signal.
Optionally, in step 201, the determining the feature data to be detected includes:
according to a preset third moving window width and a preset third time shifting step number, time-division time shifting is carried out on the collected speed change signal data to obtain a third time-division window time shifting matrix, cloud feature extraction is carried out on the third time-division window time shifting matrix, and a feature matrix of the data to be detected is obtained and serves as feature data to be detected.
Optionally, the first probabilistic neural network model in step 202 is pre-trained by:
step 2021, pre-collecting an oxygen pump vibration signal of the engine as raw data, wherein the raw data includes corresponding oxygen pump vibration signals generated by the engine under various different operating conditions. Wherein, different operating conditions include at least: high efficiency working conditions, medium efficiency working conditions, low efficiency working conditions, etc.; wherein, the working condition is called working condition for short, and the working condition is marked as high working condition, medium working condition and low working condition correspondingly; the determination of different working conditions is determined according to the related data, and the corresponding determination standard is not specifically limited in this embodiment.
Step 2022, extracting different working condition data from the original data, at least including: normal signal data under high, medium and low working conditions, and abnormal signal data under high, medium and low working conditions. Optionally, the normal signal data set and the abnormal signal data set for each operating condition take the same sample size.
And 2023, performing a segmented time shifting algorithm on the normal data and the abnormal data under different working conditions respectively, and obtaining window time shifting matrixes corresponding to different working conditions according to the width of the moving window and the time shifting step number.
2024, executing a cloud feature extraction algorithm on the obtained windowing time-shifting matrix to obtain a feature matrix; the characteristic matrix is composed of the average value, skewness and root mean square characteristic value of data.
2025, taking the normal and abnormal feature matrices under each working condition and the corresponding label (the working condition corresponding to the label identification) as the input of the first probabilistic neural network training algorithm, training the input to identify the working condition, and determining the corresponding training model parameters after training to obtain the trained first probabilistic neural network model.
For example, when the raw data at least includes data of high operating condition, medium operating condition and low operating condition, the high operating condition normal feature matrix, the high operating condition abnormal feature matrix, the medium operating condition normal feature matrix, the medium operating condition abnormal feature matrix, the low operating condition normal feature matrix and the low operating condition abnormal feature matrix are obtained correspondingly. And inputting the 6 feature matrices and the corresponding label vectors into a first probability neural network model, training the first probability neural network model to identify working conditions, and determining corresponding training model parameters after training to obtain the trained first probability neural network model.
Optionally, the working condition may be further divided into: primary, secondary and tertiary; or otherwise. And is not particularly limited to the present embodiment example. Correspondingly, according to different working conditions, the normal characteristic matrix and the abnormal characteristic matrix are correspondingly determined, the corresponding characteristic matrix and the corresponding label vector are input into the first probability neural network model, working condition recognition training is carried out, corresponding training model parameters are determined, and the trained first probability neural network model is obtained.
Optionally, in step 202, the determining the working condition of the device under test includes:
and inputting the characteristic data to be tested into the trained first probability neural network model, and determining the working condition of the equipment to be tested according to the determined parameters of the training model, wherein the corresponding result is a high working condition, a medium working condition or a low working condition. Alternatively, or in response to the result: primary working condition, secondary working condition and tertiary working condition.
Optionally, in step 203, the training of the second probabilistic neural network model according to the working condition of the device under test includes:
inputting normal sample data and abnormal sample data under the working condition into a second probabilistic neural network model for training according to a result determined by the working condition, wherein the training comprises the following steps: respectively carrying out segmented time shifting and cloud feature extraction on the normal sample data and the abnormal sample data under the determined working condition, and determining a corresponding normal feature matrix and a corresponding abnormal feature matrix; and taking the normal characteristic matrix and the corresponding label (the working condition corresponding to the label identification) and the abnormal characteristic matrix and the corresponding label (the working condition corresponding to the label identification) as the input of a second probabilistic neural network training algorithm, performing second probabilistic neural network model training, determining the training model parameters of the model, and finishing the training of the second probabilistic neural network model.
For example, if the result of the determination of the working condition is a high working condition, inputting the corresponding normal feature matrix and abnormal feature matrix respectively determined according to the normal sample data and the abnormal sample data under the high working condition and the corresponding labels thereof into the second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model; and if the probability neural network model is in a low working condition, respectively determining a corresponding normal characteristic matrix and an abnormal characteristic matrix according to normal sample data and abnormal sample data under the low working condition, inputting the normal characteristic matrix and the abnormal characteristic matrix into the second probability neural network model together with the corresponding labels, training, and determining the training model parameters of the second probability neural network model.
When the normal sample data and the abnormal sample data are respectively subjected to the segmented time shifting algorithm, the configured moving window widths are the same, and the time shifting steps are the same.
Further, in step 204, the inputting the determined feature data to be detected into the trained second probabilistic neural network model to determine the state of the speed change signal includes:
inputting the data to be detected determined in step 201 into the trained second probabilistic neural network model, and determining a detection result of the signal according to the training model parameters of the second probabilistic neural network model.
Optionally, in step 201, the determining, according to the acquired speed change signal, the feature data to be detected includes:
judging whether the working conditions corresponding to the collected speed change signals are more than one according to the collected speed change signals,
when more than one working condition, for example, corresponding to 3 working conditions, the collected speed change signal is divided into 3 speed change signal subsets, subset 1, subset 2 and subset 3; and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset, namely obtaining a characteristic matrix 1 for the speed change signal subset 1, obtaining a characteristic matrix 2 for the speed change signal subset 2, obtaining a characteristic matrix 3 for the speed change signal subset 3, and further respectively executing subsequent steps 202, 203 and 204 for the characteristic data to be detected corresponding to each speed change signal subset. Finally, the state of the variable-speed signal corresponding to each variable-speed signal subset is determined.
For example, for 3 operating conditions, including: high working condition, medium working condition and high working condition; alternatively, it comprises: medium, low and high operating conditions. Corresponding to 2 working conditions, comprising: high working condition and low working condition; alternatively, it comprises: low operating mode and medium operating mode.
And when only one working condition exists, the speed change signal subsets are not divided, and the collected speed change signals are used as a set to determine the corresponding characteristic data to be detected.
Optionally, the windowing time shift algorithm involved in steps 201, 203 and step 2023 includes:
first, signal data is input, and a set window length win (i.e., a moving window width) and a frame shift inc (i.e., a time-shift step number) are set. Calculating the time shift times of the window on the data according to the set window length, the signal length and the frame shift, wherein the calculation formula is as follows:
where nf is the time shift number, nx is the data length, and fix () is the rounding function.
And next, generating an all-zero matrix with a line nf and a column win, and sequentially storing data intercepted by the window at each time into the all-zero matrix until the last time of window movement is reached.
Thereby, a windowed time shift matrix may be obtained. The moving window width and the number of time-shifting steps corresponding to different steps are respectively set and can be the same or different.
Optionally, the cloud feature extraction algorithm involved in steps 201, 203 and 2023 includes:
inputting data processed by a window time shifting algorithm (namely a window time shifting matrix), respectively calculating the average value, skewness and root mean square of each row in the matrix, and forming the average value, skewness and root mean square into a feature matrix.
Wherein the average value of each row
The formula of (1) is:
(x1,……,xNfor N data elements in each row of data of the input matrix, N being an integer greater than 1)
The skewness α is calculated as:
root mean square xrmsThe calculation formula is as follows:
optionally, the determining, according to the acquired speed change signal, whether more than one working condition corresponding to the acquired speed change signal exists includes:
firstly, inputting a collected speed change signal, setting a preset jump threshold value of adjacent signal data points, and searching a data sequence number position of whether the absolute value of a change value between two adjacent data points is greater than the jump threshold value, wherein the position is a data signal jump point. If no such data signal trip point exists, the acquired speed change signal corresponds to a single operating condition. If such a data signal trip point exists, the acquired speed change signal corresponds to multiple operating conditions, including more than one operating condition.
Wherein the first probabilistic neural network model and the second probabilistic neural network model involved in steps 202, 203 are both neural network models; after different training data are respectively trained, different training model parameters are determined, and different technical targets are achieved. In this embodiment, the trained first probabilistic neural network model is used to determine the operating condition, and the determined (trained) second probabilistic neural network model is used to determine the state of the variable signal.
The structure of the probabilistic neural network is shown in fig. 4, and the probabilistic neural network includes four layers, which are respectively: input layer, sample layer, summation layer, competition layer.
Wherein, the effect of each layer is respectively: an input layer: accept input data sample layer: data collection collates summing layers: judging a category competition layer of the data: and outputting the result data of the category judgment.
Wherein the probabilistic neural network training algorithm for training the first probabilistic neural network model or the second probabilistic neural network model in steps 2025 and 203 comprises:
firstly, generating a probabilistic neural network training model parameter, and inputting characteristic data as training data of the training network model parameter. Wherein, the formula is adopted:
calculating an attribute probability parameter w of training datatrainAs training model parameters for the probabilistic network model.
Wherein, the train data is used as training characteristic data, which is a matrix with the size of M × N, M is the number of training samples, and N is the number of characteristic attributes. I is an identity matrix of size M x 1. w is atrainAnd training the generated N x M probability matrix for the probabilistic neural network model. Wherein wtrainRepresents a probability value for each attribute in each training sample.
Next, generating training labels relative to the training feature data: generating a training label matrix B, wherein B is a matrix with the size of M C, wherein M is the number of samples, and C is the number of training label categories. And marking the label category corresponding to each training sample serial number in the training label matrix, and marking the value of the label category as 1.
Wherein, the step 202 determines the working condition by using a pre-trained first probability neural network model and the step 204 inputs the determined feature data to be detected into a trained second neural network model to determine the probability neural network test algorithm of the state of the speed change signal, which comprises the following steps:
firstly, inputting characteristic data to be tested into an input layer, and calculating the probability value of each attribute of testdata to be tested. The feature data to be tested here refers to the feature data to be tested for steps 202 and 204.
Wherein, a formula is adopted
Calculating attribute probability parameter w of data to be testedtestThe testdata is used as characteristic data to be tested, and is a matrix with the size of P x N, P is the number of data samples to be tested, and N is the number of characteristic attributes. I is an identity matrix of size P x 1. w is atestAn N x P probability matrix is generated for the probabilistic neural network model test, each value of which represents a probability value for each attribute in each input sample.
Next, training model parameters w which are trained and generated are usedtrainAttribute probability w generated from input test datatestAnd multiplying to obtain the joint probability of the data attributes of the sample layer, wherein the joint probability comprises the following steps:
using the formula wnew＝(wtrain)T*wtestGenerating a joint probability in the sample layer,
wherein wnewIs a matrix of size M x P, which is expressed as the joint probability of each parameter probability of the training samples and the probability of each attribute sample being input.
Next, the joint probabilities output from the sample layer are input to a summation layer, where the summation probability of the input data attribute weights received from the sample layer is calculated, thereby obtaining probability values judged as various categories.
Generating a d matrix according to the generated training label matrix B, and adopting a formula d ═ BMj*I1(j ═ 1,2,3,4 …, n) where B isMjThe training label matrix is the j-th class, and the size of the training label matrix is M x 1. I is1The training label matrix is an identity matrix with the size of 1 × P, n is an integer greater than 1, the number of categories of the training label matrix is represented, for example, the training label matrix has high, medium and low working conditions, the training label matrix respectively corresponds to a normal feature matrix, and if the training label matrix is an abnormal feature matrix, n is 6.
Using a formula
And calculating the summation probability of each type of sample, wherein cp represents the summation probability of each type of sample. Sigma is a smoothing parameter, and the value is generally 0-1. D is an identity matrix with the size of M x P, and represents a category label mark matrix corresponding to the joint probability of each category.
And comparing the summed probability values of various categories according to the summed probability values of various categories, and finding out the maximum value in the summed probability values, wherein the category corresponding to the maximum value is the classification result finally obtained by judging the test data. The first probability neural network model determines the working condition corresponding to the collected characteristic data to be detected through the probability neural network test algorithm; and the second probabilistic neural network model determines whether the collected characteristic data to be detected is abnormal or not through the probabilistic neural network test algorithm.
Optionally, the accuracy of the signal detection result is verified by using a decision matrix method.
The relevant experimental data in example two are as follows:
the training set data in experiment 1 below was used to perform the training of the first probabilistic neural network model; the test set data is used for verifying whether the working condition is accurate or not by using the trained first probabilistic neural network. The training set in these experimental data corresponds to the raw data in the above-mentioned step 2022-2025, which is used to train the first probabilistic neural network. The test set in the experimental data is used for checking the training result to show that the working condition determined after the data to be tested is input into the trained first probability neural network after the relevant training can reach certain accuracy.
Experiment 1:
in the experiment, the data set selects the signal data acquired under the conditions of low, medium and high oxygen pump vibration signals, and the acquired data set is directly divided into a training set and a testing set according to the proportion of 2:1 of the number of samples, namely the training set and the testing set
The number of the signal samples collected under each working condition is equal in the training set and the testing set respectively.
In this experiment, the smoothing parameter σ was 0.00684. The window width win has a value of 50 and the frame shift has a value of 8. In low operating conditions, the number of training samples collected is 60, and the number of test samples collected is 30.
TABLE 1 signal condition discrimination results
The accuracy rate of the overall working condition discrimination is 95.56 percent.
The training set data in the following experiments 2,3 are used for training the second probabilistic neural network model; the test set is used for verifying whether the data to be detected is abnormal or not by using the trained second probabilistic neural network model. The training set in these experimental data corresponds to the sample data in step 203 above, and is used to train the second probabilistic neural network model. The test set in the experimental data is used for testing the training result to show that the result of the normal or abnormal speed change signal determined after the data to be tested is input into the trained second probabilistic neural network model after the relevant training can achieve certain accuracy.
Experiment 2:
in the experiment, the data set selects normal and abnormal signal data collected under the conditions of low, medium and high oxygen pump vibration signals, and the collected data set is directly divided into a training set and a testing set according to the proportion of 2:1 of the number of samples, namely the training set and the testing set
Normally, the number of abnormal signal samples is equal in the training set and the testing set respectively.
Under low operating conditions, the number of training samples collected is 100, and the number of test samples is 50.
In this experiment, the value of the smoothing parameter σ is 0.00684, the value of the window width win is 50, the value of the frame shift is 8, the value of the learning rate α is 0.01, and the value of the momentum factor eta is 0.05.
TABLE 2 detection results of oxygen pump vibration signals under low operating conditions
The overall detection accuracy is as follows: 96 percent.
In the middle working condition, the number of training samples collected is 200, and the number of test samples collected is 100.
Table 3 shows the detection results of the vibration signals of the oxygen pump under the working conditions
The overall detection accuracy is 93%.
In high operating conditions, the number of training samples collected is 120 and the number of test samples is 60.
TABLE 4 detection results of oxygen pump vibration signals under high operating conditions
The overall detection accuracy is 85 percent.
Experiment 3:
in the experiment, the data set selects normal and abnormal jump signal data acquired under three mixed working conditions of medium-high, medium-low and high-low of oxygen pump vibration. The number of the normal and abnormal signal samples adopted under each mixed working condition is respectively equal. The quantity of the normal and abnormal jump signal data samples collected under the medium-high mixed working condition is 50, the quantity of the normal and abnormal jump signal data samples collected under the medium-low mixed working condition is 70, and the quantity of the normal and abnormal jump signal data samples collected under the high-low mixed working condition is 30. The test results are as follows:
TABLE 5 oxygen Pump vibration Signal detection results under Mixed operating conditions
The comprehensive detection accuracy rate under the medium-high mixed working condition is 91 percent;
the comprehensive detection accuracy rate under the medium-low mixed working condition is 93.93 percent;
the comprehensive detection accuracy rate of the high-low mixed working condition is 88.33 percent.
EXAMPLE III
The present invention provides a speed change signal detection device 50, which has a structure as shown in fig. 5, and includes:
the signal acquisition module 501 is configured to acquire a speed change signal generated by the device to be detected and determine characteristic data to be detected according to the acquired speed change signal;
a working condition determining module 502 configured to input the determined feature data to be detected into a pre-trained first probability neural network model, and determine a working condition of the device to be detected;
a second probabilistic neural network model determining module 503 configured to determine a corresponding second probabilistic neural network model according to a working condition of the device under test;
a signal state determination module 504 configured to input the determined feature data to be detected into the determined second probabilistic neural network model to determine a state of the transient signal.
Optionally, the signal acquisition module 501 is further configured to:
determining whether more than one corresponding working condition exists according to the acquired speed change signals;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
Optionally, the working condition determining module 502 is configured to determine a corresponding second probabilistic neural network model, including:
according to the working condition of the equipment to be tested, determining a normal feature matrix by using normal sample data under the working condition; determining an abnormal feature matrix by using the abnormal sample data under the working condition; inputting the normal characteristic matrix and the working condition label thereof, and the abnormal characteristic matrix and the working condition label thereof into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model;
optionally, the working condition determining module 502 is configured to determine a corresponding second probabilistic neural network model, including:
selecting a second probabilistic neural network model of the working condition from at least one second probabilistic neural network model which is trained in advance according to the working condition of the equipment to be tested,
wherein the at least one pre-trained second probabilistic neural network model comprises: according to different working conditions, respectively executing the following steps to obtain second probabilistic neural network models with different working conditions:
determining a normal characteristic matrix by using normal sample data under the current working condition; and determining an abnormal characteristic matrix by using the abnormal sample data under the current working condition, inputting the normal characteristic matrix, the current working condition label, the abnormal characteristic matrix and the current working condition label into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model.
Optionally, the signal state determining module 504 is configured to input the feature data to be detected into a determined second probabilistic neural network, and determine the state of the speed change signal according to a training model parameter of the second probabilistic neural network model.
Optionally, the first probabilistic neural network model is obtained by training in advance according to the following method:
collecting corresponding speed change signals generated by the equipment to be tested under different working conditions in advance;
extracting normal signal data as normal sample data and extracting abnormal signal data as abnormal sample data;
the following treatment is respectively carried out on different working conditions: determining a normal feature matrix under the working condition according to the normal sample data under the working condition, and determining an abnormal feature matrix under the working condition according to the abnormal sample data under the working condition;
inputting the first probability neural network model for training according to the normal feature matrix and the corresponding label under different working conditions, and the abnormal feature matrix and the corresponding label under different working conditions, and determining the training model parameters of the first probability neural network model.
Optionally, wherein the determining the normal feature matrix includes:
according to a preset first moving window width and a first time shifting step number, carrying out time-sharing time shifting on the normal sample data to obtain a first window time shifting matrix, and carrying out cloud feature extraction on the first window time shifting matrix to obtain the normal feature matrix;
wherein the determining the abnormal feature matrix comprises:
and according to a preset second moving window width and a second time shifting step number, carrying out time-sharing time shifting on the abnormal sample data to obtain a second sub-window time shifting matrix, and carrying out cloud feature extraction on the second sub-window time shifting matrix to obtain the abnormal feature matrix.
Optionally, the signal acquisition module 501 is further configured to,
and according to a preset third moving window width and a third time shifting step number, carrying out time-sharing time shifting on the speed change signal to obtain a third window time shifting matrix, and carrying out cloud feature extraction on the third window time shifting matrix to obtain a data feature matrix to be detected.
Optionally, the determining, according to the acquired speed change signal, whether more than one corresponding operating condition exists includes:
according to a preset jump threshold, searching a data sequence number position whether the absolute value of the change value between two adjacent data points is greater than the jump threshold from the acquired speed change signal, and recording the position as a data signal jump point;
if the acquired speed change signal data does not have the data signal jumping points, determining that the working condition corresponding to the acquired speed change signal only comprises one type;
if the acquired speed change signals have such data signal jumping points, determining that the working conditions corresponding to the acquired speed change signals comprise more than one; wherein the dividing the acquired rate-change signals into a corresponding plurality of more than one rate-change signal subsets comprises: and carrying out data segmentation on the acquired speed change signals according to the positions of the jumping points to obtain more than one speed change signal subset.
Compared with the method for detecting the process variable speed signal at the present stage, the method for detecting the variable speed signal based on the deep learning greatly improves the detection accuracy rate of the abnormal signal in the variable speed signal, effectively avoids the situation of falling into local optimum, has stronger inference capability, and can avoid the problems of reduction of the detection precision and the model robustness.
It will be understood by those of ordinary skill in the art that all or part of the steps of the above embodiments may be implemented using a computer program flow, which may be stored in a computer readable storage medium and executed on a corresponding hardware platform (e.g., system, apparatus, device, etc.), and when executed, includes one or a combination of the steps of the method embodiments.
Alternatively, all or part of the steps of the above embodiments may be implemented by using an integrated circuit, and the steps may be respectively manufactured as an integrated circuit module, or a plurality of the blocks or steps may be manufactured as a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.
The devices/functional modules/functional units in the above embodiments may be implemented by general-purpose computing devices, and they may be centralized on a single computing device or distributed on a network formed by a plurality of computing devices.
Each device/function module/function unit in the above embodiments may be implemented in the form of a software function module and may be stored in a computer-readable storage medium when being sold or used as a separate product. The computer readable storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A method for detecting a variable signal, comprising:
acquiring a speed change signal generated by equipment to be detected, and determining characteristic data to be detected according to the acquired speed change signal;
inputting the determined characteristic data to be detected into a pre-trained first probability neural network model, and determining the working condition of the equipment to be detected;
determining a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested;
and inputting the determined characteristic data to be detected into the determined second probabilistic neural network model, and determining the state of the speed change signal.
2. The method of claim 1, wherein:
determining a corresponding second probabilistic neural network model according to the working condition of the device to be tested, wherein the determining of the corresponding second probabilistic neural network model comprises the following steps:
according to the working condition of the equipment to be tested, determining a normal feature matrix by using normal sample data under the working condition; determining an abnormal feature matrix by using the abnormal sample data under the working condition; inputting the normal characteristic matrix and the working condition label thereof, and the abnormal characteristic matrix and the working condition label thereof into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model;
or, the determining a corresponding second probabilistic neural network model according to the working condition of the device under test includes:
selecting a second probabilistic neural network model of the working condition from at least one second probabilistic neural network model which is trained in advance according to the working condition of the equipment to be tested,
wherein the at least one pre-trained second probabilistic neural network model comprises: according to different working conditions, respectively executing the following steps to obtain second probabilistic neural network models with different working conditions:
determining a normal characteristic matrix by using normal sample data under the current working condition; and determining an abnormal characteristic matrix by using the abnormal sample data under the current working condition, inputting the normal characteristic matrix, the current working condition label, the abnormal characteristic matrix and the current working condition label into a second probabilistic neural network model for training, and determining the training model parameters of the second probabilistic neural network model.
3. The method of claim 1, wherein:
wherein, the inputting the determined characteristic data to be detected into the determined second probabilistic neural network to determine the state of the speed change signal comprises:
and inputting the characteristic data to be detected into a determined second probabilistic neural network, and determining the state of the speed change signal according to the training model parameters of the second probabilistic neural network model.
4. The method of claim 1,
wherein the first probabilistic neural network model is obtained by training in advance according to the following method:
collecting corresponding speed change signals generated by the equipment to be tested under different working conditions in advance;
extracting normal signal data as normal sample data and extracting abnormal signal data as abnormal sample data;
the following treatment is respectively carried out on different working conditions: determining a normal feature matrix under the working condition according to the normal sample data under the working condition, and determining an abnormal feature matrix under the working condition according to the abnormal sample data under the working condition;
inputting the first probability neural network model for training according to the normal feature matrix and the corresponding label under different working conditions, and the abnormal feature matrix and the corresponding label under different working conditions, and determining the training model parameters of the first probability neural network model.
5. The method of claim 1,
wherein, according to the speed change signal that gathers, confirm the characteristic data that awaits measuring, include:
determining whether more than one corresponding working condition exists according to the acquired speed change signals;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
6. The method according to claim 2 or 4,
wherein the determining the normal feature matrix includes:
according to a preset first moving window width and a first time shifting step number, carrying out time-sharing time shifting on the normal sample data to obtain a first window time shifting matrix, and carrying out cloud feature extraction on the first window time shifting matrix to obtain the normal feature matrix;
wherein the determining the abnormal feature matrix comprises:
and according to a preset second moving window width and a second time shifting step number, carrying out time-sharing time shifting on the abnormal sample data to obtain a second sub-window time shifting matrix, and carrying out cloud feature extraction on the second sub-window time shifting matrix to obtain the abnormal feature matrix.
7. The method according to claim 1 or 5,
wherein, the determining the characteristic data to be detected comprises:
and according to a preset third moving window width and a third time shifting step number, carrying out time-sharing time shifting on the speed change signal to obtain a third window time shifting matrix, and carrying out cloud feature extraction on the third window time shifting matrix to obtain a data feature matrix to be detected.
8. The method of claim 5,
wherein, the determining whether the corresponding working condition is more than one according to the collected speed change signals comprises:
according to a preset jump threshold, searching a data sequence number position whether the absolute value of the change value between two adjacent data points is greater than the jump threshold from the acquired speed change signal, and recording the position as a data signal jump point;
if the acquired speed change signal data does not have the data signal jumping points, determining that the working condition corresponding to the acquired speed change signal only comprises one type;
if the acquired speed change signals have such data signal jumping points, determining that the working conditions corresponding to the acquired speed change signals comprise more than one; wherein the dividing the acquired rate-change signals into a corresponding plurality of more than one rate-change signal subsets comprises: and carrying out data segmentation on the acquired speed change signals according to the positions of the jumping points to obtain more than one speed change signal subset.
9. A speed change signal detection device, comprising:
the signal acquisition module is used for acquiring a speed change signal generated by the equipment to be detected and determining the characteristic data to be detected according to the acquired speed change signal;
the working condition determining module is used for inputting the determined characteristic data to be detected into a first probability neural network model which is trained in advance and determining the working condition of the equipment to be detected;
the second probabilistic neural network model determining module is set to determine a corresponding second probabilistic neural network model according to the working condition of the equipment to be tested;
and the signal state determining module is used for inputting the determined characteristic data to be detected into the determined second probabilistic neural network model and determining the state of the speed change signal.
10. The apparatus of claim 9,
wherein the acquisition module is further configured to:
determining whether more than one corresponding working condition is available according to the speed change signal acquired by the signal;
when more than one working condition exists, dividing the acquired speed change signal into more than one corresponding speed change signal data subsets, and respectively determining corresponding characteristic data to be detected according to each speed change signal data subset;
and when only one working condition exists, determining the corresponding characteristic data to be detected according to the collected speed change signals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911379540.4A CN111160454B (en) | 2019-12-27 | 2019-12-27 | Quick change signal detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911379540.4A CN111160454B (en) | 2019-12-27 | 2019-12-27 | Quick change signal detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111160454A true CN111160454A (en) | 2020-05-15 |
| CN111160454B CN111160454B (en) | 2023-12-22 |
Family
ID=70558602
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911379540.4A Active CN111160454B (en) | 2019-12-27 | 2019-12-27 | Quick change signal detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111160454B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103886405A (en) * | 2014-02-20 | 2014-06-25 | 东南大学 | Boiler combustion condition identification method based on information entropy characteristics and probability nerve network |
| CN105913026A (en) * | 2016-04-12 | 2016-08-31 | 江苏大学 | Passenger detecting method based on Haar-PCA characteristic and probability neural network |
| US20170323202A1 (en) * | 2016-05-06 | 2017-11-09 | Fujitsu Limited | Recognition apparatus based on deep neural network, training apparatus and methods thereof |
| CN108334907A (en) * | 2018-02-09 | 2018-07-27 | 哈尔滨工业大学（威海） | A kind of complex equipment point method for detecting abnormality and system based on deep learning |
| CN108647707A (en) * | 2018-04-25 | 2018-10-12 | 北京旋极信息技术股份有限公司 | Probabilistic neural network creation method, method for diagnosing faults and device, storage medium |
| CN109308522A (en) * | 2018-09-03 | 2019-02-05 | 河海大学常州校区 | A kind of GIS failure prediction method based on Recognition with Recurrent Neural Network |
| CN110110740A (en) * | 2019-03-26 | 2019-08-09 | 中国地质大学（武汉） | Drilling process operating mode's switch method based on Multiple Time Scales feature and neural network |
| CN110598845A (en) * | 2019-08-13 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Data processing method, data processing device, computer equipment and storage medium |
-
2019
- 2019-12-27 CN CN201911379540.4A patent/CN111160454B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103886405A (en) * | 2014-02-20 | 2014-06-25 | 东南大学 | Boiler combustion condition identification method based on information entropy characteristics and probability nerve network |
| CN105913026A (en) * | 2016-04-12 | 2016-08-31 | 江苏大学 | Passenger detecting method based on Haar-PCA characteristic and probability neural network |
| US20170323202A1 (en) * | 2016-05-06 | 2017-11-09 | Fujitsu Limited | Recognition apparatus based on deep neural network, training apparatus and methods thereof |
| CN108334907A (en) * | 2018-02-09 | 2018-07-27 | 哈尔滨工业大学（威海） | A kind of complex equipment point method for detecting abnormality and system based on deep learning |
| CN108647707A (en) * | 2018-04-25 | 2018-10-12 | 北京旋极信息技术股份有限公司 | Probabilistic neural network creation method, method for diagnosing faults and device, storage medium |
| CN109308522A (en) * | 2018-09-03 | 2019-02-05 | 河海大学常州校区 | A kind of GIS failure prediction method based on Recognition with Recurrent Neural Network |
| CN110110740A (en) * | 2019-03-26 | 2019-08-09 | 中国地质大学（武汉） | Drilling process operating mode's switch method based on Multiple Time Scales feature and neural network |
| CN110598845A (en) * | 2019-08-13 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Data processing method, data processing device, computer equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 刘学;梁红;玄志武;: "基于小波模极大值模糊熵的遥测振动信号异常检测", 振动与冲击, no. 09, pages 147 - 152 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111160454B (en) | 2023-12-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110132598B (en) | Fault noise diagnosis algorithm for rolling bearing of rotating equipment | |
| KR101822829B1 (en) | Method of identifying partial discharge and noise of switchgear using machine learning | |
| CN111505424A (en) | Large experimental device power equipment fault diagnosis method based on deep convolutional neural network | |
| CN111079348B (en) | Method and device for detecting slowly-varying signal | |
| CN108647707B (en) | Probabilistic neural network creation method, failure diagnosis method and apparatus, and storage medium | |
| CN110247910A (en) | A kind of detection method of abnormal flow, system and associated component | |
| CN102339347A (en) | A method for computer-assisted analyzing of a technical system | |
| SE542335C2 (en) | Method and system for vehicle analysis | |
| CN111126820A (en) | Electricity stealing prevention method and system | |
| CN112116002A (en) | Determination method, verification method and device of detection model | |
| CN115793590A (en) | Data processing method and platform suitable for system safety operation and maintenance | |
| CN115660262A (en) | Intelligent engineering quality inspection method, system and medium based on database application | |
| CN117056678B (en) | Machine pump equipment operation fault diagnosis method and device based on small sample | |
| CN117589233A (en) | Intelligent detection method and system for running state of relay protection device | |
| CN112199295A (en) | Deep neural network defect positioning method and system based on frequency spectrum | |
| US20190294988A1 (en) | Plant abnormality prediction system and method | |
| CN111160454B (en) | Quick change signal detection method and device | |
| CN116400168A (en) | Power grid fault diagnosis method and system based on depth feature clustering | |
| JP3234742B2 (en) | Method and apparatus for estimating remaining life of electronic component | |
| JP2002099319A (en) | Plant diagnosing device | |
| CN113052060B (en) | Bearing residual life prediction method and device based on data enhancement and electronic equipment | |
| CN114863210A (en) | Method and system for resisting sample attack of bridge structure health monitoring data driving model | |
| CN114022698A (en) | Multi-tag behavior identification method and device based on binary tree structure | |
| Miron et al. | Artificial neural network approach for fault recognition in a wastewater treatment process | |
| Singh et al. | Predicting the remaining useful life of ball bearing under dynamic loading using supervised learning |
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 |