CN109739197A - A kind of multi-state failure prediction method of chemical spent material processing equipment - Google Patents
A kind of multi-state failure prediction method of chemical spent material processing equipment Download PDFInfo
- Publication number
- CN109739197A CN109739197A CN201910036826.6A CN201910036826A CN109739197A CN 109739197 A CN109739197 A CN 109739197A CN 201910036826 A CN201910036826 A CN 201910036826A CN 109739197 A CN109739197 A CN 109739197A
- Authority
- CN
- China
- Prior art keywords
- module
- data
- material processing
- processing equipment
- spent material
- 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.)
- Pending
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention belongs to chemical industry equipment failure predication technical fields, a kind of multi-state failure prediction method of chemical spent material processing equipment is disclosed, the multi-state failure prediction system of the chemical spent material processing equipment includes: current detection module, voltage detection module, noise detection module, main control module, fault database building module, fault diagnosis module, warning module, detection data memory module, display module.The present invention constructs module by fault database and establishes a kind of efficient multi-level state detection data combined index model, improves the accuracy of chemical spent material processing equipment detection data O&M, the cost of reduced O&M improves the working efficiency of O&M;Meanwhile the diagnosis of chemical spent material processing equipment component failure is realized using assembled classification method by fault diagnosis module, it can be used for online real-time fault diagnosis, find mechanical equipment fault in time, prevent the generation of major accident;Significantly improve the precision of fault diagnosis.
Description
Technical field
The invention belongs to chemical industry equipment failure predication technical field more particularly to a kind of multiplexings of chemical spent material processing equipment
Condition failure prediction method.
Background technique
Chemical spent material is exactly industrial chemicals quilt in remaining leftover pieces or production process during Commercial cultivation
Pollution, the waste that cannot be recycled.Currently, development of the chemical production processes in chemical production is constantly in exploitation rank
Section, and the research and development of chemical technology were becoming gradually intimately in recent years, with holding for energy conservation and environmental protection and low-carbon life concept
It is continuous burning hot, the attention degree of environment is also gradually increased, so, with regard to needing in time to change it in Chemical Manufacture
Become.In the past, it is generally difficult to obtain effective solution since chemical industry leads to the problem of the disposal of pollutants generated in the middle, chemical spent material
Discharge is very serious, this just brings very big pollution to our living environment.Therefore, the processing of chemical spent material is very heavy
It wants.However, existing chemical spent material processing equipment O&M low efficiency;Meanwhile it is existing to chemical spent material processing equipment fault diagnosis essence
It spends not high, it is difficult to meet demand.
In conclusion problem of the existing technology is:
(1) existing chemical spent material processing equipment O&M low efficiency;Meanwhile it is existing to chemical spent material processing equipment fault diagnosis
Precision is not high, it is difficult to meet demand.
(2) it is utilized during current sensor detection chemical spent material processing equipment operating current data in the prior art
The section with low-current ground faults localization method of similarity relationships between the upstream and downstream fault transient state current of fault point, due to lacking
There are certain blind areas to easily lead to positioning mistake in stringent theoretical proof and principle, cannot eliminate blind location area, reduce transient state
The reliability and adaptability of positioning principle.
(3) noise transducer uses traditional algorithm to detect chemical spent material processing equipment work noise data in the prior art
During, not can be reduced number of training mesh while guaranteeing voice training sample quality, reduce trained efficiency and
Voice recognition accuracy rate.
(4) it during surveying chemical spent material processing equipment operating voltage data by voltage sensor in the prior art, surveys
Influence of the accuracy vulnerable to environment temperature is measured, voltage sensor measurement result is modified by using existing algorithm, no
It can effectively realize the temperature-compensating to sensor.
Summary of the invention
In view of the problems of the existing technology, the present invention provides a kind of multi-state failure of chemical spent material processing equipment is pre-
Survey method.
The invention is realized in this way a kind of multi-state failure prediction method of chemical spent material processing equipment, the chemical industry
The multi-state failure prediction method of disposal unit includes:
The first step, to chemical spent material processing equipment operating current data, chemical spent material processing equipment operating voltage data and
Chemical spent material processing equipment work noise data carry out data acquisition;
Second step according to the collected data judges the operating condition and fault type of chemical spent material processing equipment;
Third step, according to the data of fault diagnosis and as a result, building fault case library;
4th step, electric current, voltage, noise, fault case library, the fault diagnosis result of detection are stored, and display is passed through
Device shows relevant data information.
Further, chemical spent material processing equipment operating current data are detected by current sensor, using fault area
The realization process of section location algorithm is as follows:
Step 1, substation's terminal change according to zero mode voltage and start, and select faulty line, and by route selection result and failure
Zero mould electric current of line outlet reports main website;
Step 2, the line feed terminals of each outlet start according to the catastrophe of zero mould electric current of place test point, and by failure
Zero mould electric current reports main website;
Step 3, main website receive terminal data, are filtered to each terminal zero mould current data of faulty line, extract respectively
Its transient state component and power frequency component then disregard sound circuit terminal data;
Step 4 successively calculates the transient zero mode current phase relation between each adjacent test point since faulty line outlet
Several and power frequency zero-sequence current related coefficient, synthesizes revised transient zero mode current related coefficient;
Step 5 successively judges each section two sides transient zero mode current related coefficient since faulty line outlet.If certain
The revised transient zero mode current related coefficient in section two sides is less than setting threshold value ρT, then the section is fault section;
Step 6, if the transient current related coefficient of all section two sides is all larger than threshold value ρTThen most end test point downstream
Section is fault section.
Further, chemical spent material processing equipment work noise data are detected by noise transducer, using based on cluster mark
The training sample selection algorithm of note, comprising the following steps:
Step 1 selects a large amount of sample as the initial set of training sample, is then by manually discrimination rude classification
Coarse sample set 1, coarse sample set 2 ..., coarse sample set w, each coarse sample set represent same class sound;
Step 2 carries out pretreatment and hybrid feature extraction to each sample in coarse sample set, is mixed accordingly
Close eigenvectors matrix;
Step 3 asks its mean vector to indicate sample each composite character vector matrix, can obtain w altogether
It is worth vector set;
Step 4 clusters this w mean vector collection, depending on the selection gist actual conditions of classification number, often respectively
One mean vector collection can be polymerized to PwThen class selects sample corresponding to a small number of mean vectors as final from every one kind
Training sample presenting set.
Further, chemical spent material processing equipment operating voltage data are surveyed by voltage sensor, by calibration factor to electricity
Pressure sensor measurement result is modified, and realizes the temperature-compensating to sensor, and specific algorithm is as follows:
Step 1, it is f that reference voltage source, which generates frequency,2, virtual value U2Reference voltage signal, two times transfer device is from electricity
The received reference voltage signal u of pressure sensor calibration voltage output end2After demodulated processing, high-temperature stability, height are obtained
The voltage signal of accuracy, is expressed as
N is the counting of data sample in formula;tnFor the sampling time of nth data;For the acquisition of distal end acquisition module
Reference voltage signal u2, initial phase;
Step 2, two times transfer device receive optical voltage sensing unit sensitivity from the inductive signal output end of voltage sensor
The induction measured voltage signal u ' of acquisition1With induction reference voltage signal u '2, and to the signal carry out data processing, obtain vulnerable to
The induction measured voltage and induction reference voltage of ambient temperature effect, are expressed as
Δ k is the output factor variable quantity that the external influence factors such as environment temperature cause optical voltage sensing unit in formula,
It is unrelated with sense voltage signal frequency;For the induction measured voltage signal u ' of optical voltage sensing unit sensitivity1Initial phase
Position;U1For the virtual value of measured voltage source output voltage signal;For the induction reference voltage of optical voltage sensing unit sensitivity
Signal u '2Initial phase;
Step 3, using triangle window weighting algorithm and discrete fourier algorithm, two times transfer device realizes the induction to acquisition
Reference voltage signal u '2With reference voltage signal u2Multicycle data virtual value calculate;
The external influence factors such as environment temperature cause the output factor variable quantity k of optical voltage sensing unit to pass through following formula meter
It obtains;
U ' in formula2The induction reference voltage signal u ' obtained for optical voltage sensing unit sensitivity2Virtual value;
Step 4 utilizes the above-mentioned coefficient induction measured voltage signal u ' sensitive to optical voltage sensing unit1It is repaired
Just, obtaining output voltage signal hardly influenced by ambient temperature is
1+ Δ k is the self calibration coefficient of voltage sensor output signal in formula;
Measured voltage source output voltage signal u1In addition to comprising fundamental frequency, it is also possible to it include other order harmonic frequencies,
Above-mentioned calculation method is equally applicable.
Another object of the present invention is to provide a kind of multi-state failure predications for executing the chemical spent material processing equipment
The multi-state failure of the multi-state failure prediction system of the chemical spent material processing equipment of method, the chemical spent material processing equipment is pre-
Examining system includes:
Current detection module is connect with main control module, for detecting chemical spent material processing equipment work by current sensor
Make current data;
Voltage detection module is connect with main control module, for detecting chemical spent material processing equipment work by voltage sensor
Make voltage data;
Noise detection module is connect with main control module, for detecting chemical spent material processing equipment work by noise transducer
Make noise data;
Main control module constructs module, failure with current detection module, voltage detection module, noise detection module, fault database
Diagnostic module, warning module, detection data memory module, display module connection, for controlling modules just by single-chip microcontroller
Often work;
Fault database constructs module, connect with main control module, for constructing fault case library by data construction procedures;
Fault diagnosis module is connect with main control module, for diagnosing chemical industry according to noise data by fault diagnostic program
Disposal unit fault type;
Warning module is connect with main control module, for carrying out pre-alert notification according to diagnostic result by alarm;
Detection data memory module, connect with main control module, for storing the electric current of detection by memory, voltage, making an uproar
Sound, fault case library, fault diagnosis result;
Display module is connect with main control module, when for showing detection chemical spent material processing equipment work by display
Electric current, voltage, noise, fault case library, fault diagnosis result.
Another object of the present invention is to provide a kind of multi-state failure predications using the chemical spent material processing equipment
The chemical industry equipment failure predication platform of method.
Advantages of the present invention and good effect are as follows: the present invention constructs module by fault database and establishes a kind of efficient multistage
State-detection aggregation of data index model constructs a kind of effective fault case library, and with fault tree, the method for fault spectrum
Case library is indexed, the information content and availability of live detection data is improved, improves chemical spent material processing equipment
The accuracy of detection data O&M, the cost of reduced O&M improve the working efficiency of O&M;Meanwhile passing through fault diagnosis
Module realizes the diagnosis of chemical spent material processing equipment component failure using assembled classification method, can be used for online real time fail
Diagnosis, finds mechanical equipment fault in time, prevents the generation of major accident;Significantly improve the precision of fault diagnosis.
Current detection module detects chemical spent material processing equipment operating current data by current sensor in the present invention
In the process, the section with low-current ground faults positioning side of similarity relationships between the upstream and downstream fault transient state current of fault point is utilized
Method is avoided since there are certain blind areas to easily lead to positioning mistake in the stringent theoretical proof of shortage and principle, in order to eliminate positioning
The reliability and adaptability of transient state positioning principle are improved, using a kind of fault section location algorithm in blind area.
Noise detection module is used to detect chemical spent material processing equipment work noise number by noise transducer in the present invention
During, in order to reduce training sample number while guaranteeing voice training sample quality, trained efficiency is improved, is mentioned
High sound recognition accuracy, using the training sample selection algorithm based on cluster mark.
Voltage detection module surveys the mistake of chemical spent material processing equipment operating voltage data by voltage sensor in the present invention
Cheng Zhong, influence of the measuring accuracy vulnerable to environment temperature are modified voltage sensor measurement result by calibration factor, real
Now to the temperature-compensating of sensor.
Detailed description of the invention
Fig. 1 is the multi-state failure prediction method flow chart of chemical spent material processing equipment provided in an embodiment of the present invention.
Fig. 2 is the multi-state failure prediction system structural representation of chemical spent material processing equipment provided in an embodiment of the present invention
Figure;
In figure: 1, current detection module;2, voltage detection module;3, noise detection module;4, main control module;5, fault database
Construct module;6, fault diagnosis module;7, warning module;8, detection data memory module;9, display module.
Specific embodiment
In order to further understand the content, features and effects of the present invention, the following examples are hereby given, and cooperate attached drawing
Detailed description are as follows.
Structure of the invention is explained in detail with reference to the accompanying drawing.
As shown in Figure 1, the multi-state failure prediction method of chemical spent material processing equipment provided by the invention includes following step
It is rapid:
S101: to chemical spent material processing equipment operating current data, chemical spent material processing equipment operating voltage data and change
Work disposal unit work noise data carry out data acquisition;
S102;According to the data of above-mentioned acquisition, the operating condition and fault type of chemical spent material processing equipment are sentenced
It is disconnected, it such as breaks down, alarm carries out pre-alert notification according to diagnostic result;
S103: according to the data of fault diagnosis and as a result, building fault case library;
S104: electric current, voltage, noise, fault case library, the fault diagnosis result of detection are stored, and display is passed through
Show relevant data information.
As shown in Fig. 2, the multi-state failure prediction system of chemical spent material processing equipment provided by the invention includes: electric current inspection
Survey module 1, voltage detection module 2, noise detection module 3, main control module 4, fault database building module 5, fault diagnosis module 6,
Warning module 7, detection data memory module 8, display module 9.
Current detection module 1 is connect with main control module 4, for detecting chemical spent material processing equipment by current sensor
Operating current data;
Voltage detection module 2 is connect with main control module 4, for surveying chemical spent material processing equipment work by voltage sensor
Make voltage data;
Noise detection module 3 is connect with main control module 4, for detecting chemical spent material processing equipment by noise transducer
Work noise data;
Main control module 4 constructs module with current detection module 1, voltage detection module 2, noise detection module 3, fault database
5, fault diagnosis module 6, warning module 7, detection data memory module 8, display module 9 connect, for being controlled by single-chip microcontroller
Modules work normally;
Fault database constructs module 5, connect with main control module 4, for constructing fault case library by data construction procedures;
Fault diagnosis module 6 is connect with main control module 4, for passing through fault diagnostic program according to noise data diagnosisization
Work disposal unit fault type;
Warning module 7 is connect with main control module 4, for carrying out pre-alert notification according to diagnostic result by alarm;
Detection data memory module 8 is connect with main control module 4, for by memory storage detection electric current, voltage,
Noise, fault case library, fault diagnosis result;
Display module 9 is connect with main control module 4, for showing detection chemical spent material processing equipment work by display
When electric current, voltage, noise, fault case library, fault diagnosis result.
The current detection module 1 detects the process of chemical spent material processing equipment operating current data by current sensor
In, using the section with low-current ground faults localization method of similarity relationships between the upstream and downstream fault transient state current of fault point,
It avoids since there are certain blind areas to easily lead to positioning mistake in the stringent theoretical proof of shortage and principle, in order to eliminate positioning blind
Area, improve transient state positioning principle reliability and adaptability, using a kind of fault section location algorithm realization process such as
Under:
Step 1, substation's terminal change according to zero mode voltage and start, and select faulty line, and by route selection result and failure
Zero mould electric current of line outlet reports main website;
Step 2, the line feed terminals of each outlet start according to the catastrophe of zero mould electric current of place test point, and by failure
Zero mould electric current reports main website;
Step 3, main website receive terminal data, are filtered to each terminal zero mould current data of faulty line, extract respectively
Its transient state component and power frequency component then disregard sound circuit terminal data;
Step 4 successively calculates the transient zero mode current phase relation between each adjacent test point since faulty line outlet
Several and power frequency zero-sequence current related coefficient, synthesizes revised transient zero mode current related coefficient;
Step 5 successively judges each section two sides transient zero mode current related coefficient since faulty line outlet.If certain
The revised transient zero mode current related coefficient in section two sides is less than setting threshold value ρ T (including negative value), then the section is failure
Section;
Step 6, if the transient current related coefficient of all section two sides is all larger than threshold value ρTThen most end test point downstream
Section is fault section.
The noise detection module 3, for detecting chemical spent material processing equipment work noise data by noise transducer
During, in order to reduce training sample number while guaranteeing voice training sample quality, trained efficiency is improved, is improved
Voice recognition accuracy rate, using the training sample selection algorithm based on cluster mark, comprising the following steps:
Step 1 selects a large amount of sample as the initial set of training sample, is then by manually discrimination rude classification
Coarse sample set 1, coarse sample set 2 ..., coarse sample set w, each coarse sample set represent same class sound;
Step 2 carries out pretreatment and hybrid feature extraction to each sample in coarse sample set, is mixed accordingly
Close eigenvectors matrix;
Step 3 asks its mean vector to indicate sample each composite character vector matrix, can obtain w altogether
It is worth vector set;
Step 4 clusters this w mean vector collection, depending on the selection gist actual conditions of classification number, often respectively
One mean vector collection can be polymerized to PwThen class selects sample corresponding to a small number of mean vectors as final from every one kind
Training sample presenting set.
The voltage detection module 2 surveys the process of chemical spent material processing equipment operating voltage data by voltage sensor
In, influence of the measuring accuracy vulnerable to environment temperature is modified voltage sensor measurement result by calibration factor, realizes
Temperature-compensating to sensor, specific algorithm are as follows:
Step 1, it is f that reference voltage source, which generates frequency,2, virtual value U2Reference voltage signal, two times transfer device is from electricity
The received reference voltage signal u of pressure sensor calibration voltage output end2After demodulated processing, high-temperature stability, height are obtained
The voltage signal of accuracy, is expressed as
N is the counting of data sample in formula;tnFor the sampling time of nth data;For the acquisition of distal end acquisition module
Reference voltage signal u2, initial phase;
Step 2, two times transfer device receive optical voltage sensing unit sensitivity from the inductive signal output end of voltage sensor
The induction measured voltage signal u ' of acquisition1With induction reference voltage signal u '2, and to the signal carry out data processing, obtain vulnerable to
The induction measured voltage and induction reference voltage of ambient temperature effect, are expressed as
Δ k is the output factor variable quantity that the external influence factors such as environment temperature cause optical voltage sensing unit in formula,
It is unrelated with sense voltage signal frequency;For the induction measured voltage signal u ' of optical voltage sensing unit sensitivity1Initial phase
Position;U1For the virtual value of measured voltage source output voltage signal;For the induction reference voltage of optical voltage sensing unit sensitivity
Signal u '2Initial phase;
Step 3, using triangle window weighting algorithm and discrete fourier algorithm, two times transfer device realizes the induction to acquisition
Reference voltage signal u '2With reference voltage signal u2Multicycle data virtual value calculate;
The external influence factors such as environment temperature cause the output factor variable quantity k of optical voltage sensing unit to pass through following formula meter
It obtains;
U ' in formula2The induction reference voltage signal u ' obtained for optical voltage sensing unit sensitivity2Virtual value;
Step 4 utilizes the above-mentioned coefficient induction measured voltage signal u ' sensitive to optical voltage sensing unit1It is repaired
Just, obtaining output voltage signal hardly influenced by ambient temperature is
1+ Δ k is the self calibration coefficient of voltage sensor output signal in formula;
Measured voltage source output voltage signal u1In addition to comprising fundamental frequency, it is also possible to it include other order harmonic frequencies,
Above-mentioned calculation method is equally applicable.
Fault database building 5 construction method of module provided by the invention is as follows:
1) the multi-state fault data of chemical spent material processing equipment is detected by fault test set;
2) high-efficiency multi-stage state-detection aggregation of data index model is established, detection data is modeled from multiple dimensions;
3) carry out expertise, case library and the intelligentized research of rule base;
4) using the state-detection data based on all kinds of standardization analyzing and diagnosing technology, using multi-information merging technology come
Improve the accuracy of analyzing and diagnosing;
5) for the multi-state fault case library of chemical spent material processing equipment and correlativity library research standard, intelligence
Typical fault case library constructing technology, construct Fault tree and fault spectrum based on fault case;
6) the typical case library of transformer equipment scene O&M is constructed, on this basis, is matched using the quick-searching of case
Technology realizes standardization storage, the Rapid matching of typical case, assists the O&M decision at scene.
In step 2) provided by the invention, state-detection data from power equipment basic data, power equipment operation data,
The dimensions such as detecting instrument, detection data, weather information carry out data modeling, and carry out clustering using K-prototypes,
The relevance between each dimension is studied, establishes the higher-dimension blended data mark mould of type state detection data on this basis
Type;State-detection data sample library is constructed with this identification model, constructs the state-detection data fast indexing based on fingerprint search
Model, the model is by extracting data to be tested fingerprint, in state-detection data sample library, carries out fingerprint search matching.
In step 3) provided by the invention, the expert system of transformer equipment status assessment and diagnosis can be become using having
The failure tree node of amount realizes transformer equipment fault diagnosis typical case by the algorithm of compact fusion fuzzy set and fault tree
Regular expression and storage.
In step 4) provided by the invention, the input source or information source of accident analysis are determined by multi-information merging technology,
The fault case library and correlativity library of main information and support equipment diagnostic analysis including equipment;Then according to diagnostic device
Whether failure is stopped transport, and in fortune then by fault case library and correlativity library, it is latent to carry out pre- measurement equipment using failure predication technology
In failure, if stopping transport, then by fault case library and correlativity library, is diagnosed using fault diagnosis technology and carry out positioning device
Failure.
Mainly include following sub-step in step 5) provided by the invention:
Using standardization, the constructing technology of intelligentized typical fault case library, the equipment event based on fault case is constructed
Barrier tree and fault spectrum, fault tree use the node with variable, can intuitively reflect the system failure and various basic faults
Logical relation, and then the weakest link of system is found out by fault spectrum again, reinforces inspection and maintenance to weak link;
Utilize equipment account, historical failure defect, historical test data, online monitoring data, gas based on mass data
Image data etc. constructs the fault prediction model based on fault history and condition monitoring, and the model is by by monitoring state and history
Failure operation state compares, and whether judging that failure occurs, and then realizes equipment potentiality fault analysis and diagnosis and prediction;
Using the method for diagnosing faults of data driven type (statistical analysis method and artificial intelligence quantitative Diagnosis method), electric power is realized
The fault diagnosis of equipment;
For the deficiency of existing diagnosis algorithm, the electric power apparatus integrated diagnosis scheme based on fault tree frame, the party are proposed
Case utilizes the fault tree with variable node, is checked layer by layer step by step, final positioning failure position and failure cause.
6 diagnostic method of fault diagnosis module provided by the invention is as follows:
(1) it is special that the chemical spent material processing equipment failure for including in sensor acquisition signal is extracted using signal analysis method
Sign, and form fault feature vector;
(2) fault feature vector off-line training fault diagnosis classifier is utilized, the fault diagnosis classifier is using combination
Classifier, the assembled classifier is using Adaboost boosting algorithm as combined method;
(3) with the fault diagnosis of trained fault diagnosis classifier real-time perfoming chemical spent material processing equipment.
The step of training provided by the invention and test failure diagnostic classification device, is as follows:
(a) data set D includes d data group: (x1,y1),(x2,y2),…,(xd,yd), wherein xjIndicate j-th of failure
Feature vector, yjIndicate class label, j=1,2 ..., d are initialized as the weight wj of each data group in data set D
(b) k wheel is carried out from data set D to sample with putting back to, and obtains training set Di, wherein DiIndicate that the i-th wheel is sampled
The training set arrived, i=1,2 ..., k, k are the number of base classifier in assembled classifier;
(c) according to training set DiObtain corresponding base classifier Ti;
(d) T is calculatediError rate error (Ti), work as TiError rate error (Ti) when being more than preset threshold t, abandon instruction
Practice collection Di, regenerate new training set DiAnd corresponding new base classifier Ti;
(e) for each data group correctly classified, according to TiError rate error (Ti) update weight wj, and to institute
There is the weight of data group to standardize, so that the sum of updated weight is identical as the sum of weight before update;
(f) each base classifier T is assignediVoting weight Wi, TiError rate error (Ti) value it is lower, WiValue get over
Height obtains the assembled classifier comprising k base classifier;
(g) classified using assembled classifier to test data x, the weight wj of all classes is initialized as 0, from first base
Classifier starts, and successively calculates the classification results of each base classifier, then the output result of each base classifier is as follows:
ci=Wi·Ti(x);
Output result for each classification number, to all base classifiers that same class alias is distributed to assembled classifier
ciSummation, using classification number corresponding to summing value maximum as the predicted value returned.
The above is only the preferred embodiments of the present invention, and is not intended to limit the present invention in any form,
Any simple modification made to the above embodiment according to the technical essence of the invention, equivalent variations and modification, belong to
In the range of technical solution of the present invention.
Claims (6)
1. a kind of multi-state failure prediction method of chemical spent material processing equipment, which is characterized in that the chemical spent material processing is set
Standby multi-state failure prediction method includes:
The first step, to chemical spent material processing equipment operating current data, chemical spent material processing equipment operating voltage data and chemical industry
Disposal unit work noise data carry out data acquisition;
Second step according to the collected data judges the operating condition and fault type of chemical spent material processing equipment;
Third step, according to the data of fault diagnosis and as a result, building fault case library;
4th step, electric current, voltage, noise, fault case library, the fault diagnosis result of detection are stored, aobvious by display
Show relevant data information.
2. the multi-state failure prediction method of chemical spent material processing equipment as described in claim 1, which is characterized in that pass through electricity
Flow sensor detect chemical spent material processing equipment operating current data, using fault section location algorithm realization process such as
Under:
Step 1, substation's terminal change according to zero mode voltage and start, and select faulty line, and by route selection result and faulty line
It exports zero mould electric current and reports main website;
Step 2, the line feed terminals of each outlet start according to the catastrophe of zero mould electric current of place test point, and by zero mould of failure
Electric current reports main website;
Step 3, main website receive terminal data, are filtered to each terminal zero mould current data of faulty line, extract it respectively temporarily
State component and power frequency component then disregard sound circuit terminal data;
Step 4, since faulty line outlet, successively calculate transient zero mode current related coefficient between each adjacent test point and
Power frequency zero-sequence current related coefficient synthesizes revised transient zero mode current related coefficient;
Step 5 successively judges each section two sides transient zero mode current related coefficient since faulty line outlet;If certain section
The revised transient zero mode current related coefficient in two sides is less than setting threshold value ρ T, then the section is fault section;
Step 6, the most end test point downstream section if the transient current related coefficient of all section two sides is all larger than threshold value ρ T
For fault section.
3. the multi-state failure prediction method of chemical spent material processing equipment as described in claim 1, which is characterized in that by making an uproar
Sonic transducer detect chemical spent material processing equipment work noise data, using based on cluster mark training sample selection algorithm,
The following steps are included:
Step 1 selects a large amount of sample as the initial set of training sample, then by manually distinguishing that rude classification is coarse
Sample set 1, coarse sample set 2 ..., coarse sample set w, each coarse sample set represent same class sound;
Step 2 carries out pretreatment and hybrid feature extraction to each sample in coarse sample set, obtains mixing accordingly special
Levy vector matrix;
Step 3, to each composite character vector matrix ask its mean vector indicate sample, can obtain altogether w mean value to
Quantity set;
Step 4 clusters this w mean vector collection respectively, depending on the selection gist actual conditions of classification number, each
Mean vector collection can be polymerized to PwThen class selects sample corresponding to a small number of mean vectors as final instruction from every one kind
Practice sample representation collection.
4. the multi-state failure prediction method of chemical spent material processing equipment as described in claim 1, which is characterized in that pass through electricity
Pressure sensor surveys chemical spent material processing equipment operating voltage data, is repaired by calibration factor to voltage sensor measurement result
Just, the temperature-compensating to sensor is realized, specific algorithm is as follows:
Step 1, it is f that reference voltage source, which generates frequency,2, virtual value U2Reference voltage signal, two times transfer device from voltage pass
The received reference voltage signal u of sensor calibration voltage output end2After demodulated processing, high-temperature stability, high precision are obtained
The voltage signal of degree, is expressed as
N is the counting of data sample in formula;tnFor the sampling time of nth data;For the benchmark of distal end acquisition module acquisition
Voltage signal u2, initial phase;
Step 2, two times transfer device receive optical voltage sensing unit sensitivity from the inductive signal output end of voltage sensor and obtain
Induction measured voltage signal u '1With induction reference voltage signal u '2, and data processing is carried out to the signal, it obtains vulnerable to environment
The induction measured voltage and induction reference voltage that temperature influences, are expressed as
Δ k is the output factor variable quantity that the external influence factors such as environment temperature cause optical voltage sensing unit in formula, and quick
It is unrelated to feel voltage signal frequency;For the induction measured voltage signal u ' of optical voltage sensing unit sensitivity1Initial phase;U1
For the virtual value of measured voltage source output voltage signal;For the induction reference voltage signal of optical voltage sensing unit sensitivity
u′2Initial phase;
Step 3, using triangle window weighting algorithm and discrete fourier algorithm, two times transfer device realizes the induction benchmark to acquisition
Voltage signal u '2With reference voltage signal u2Multicycle data virtual value calculate;
The external influence factors such as environment temperature cause the output factor variable quantity k of optical voltage sensing unit to be calculate by the following formula
It arrives;
U ' in formula2The induction reference voltage signal u ' obtained for optical voltage sensing unit sensitivity2Virtual value;
Step 4 utilizes the above-mentioned coefficient induction measured voltage signal u ' sensitive to optical voltage sensing unit1It is modified, obtains
It is to output voltage signal hardly influenced by ambient temperature
1+ Δ k is the self calibration coefficient of voltage sensor output signal in formula;
Measured voltage source output voltage signal u1In addition to comprising fundamental frequency, it is also possible to include other order harmonic frequencies, above-mentioned meter
Calculation method is equally applicable.
5. the chemical spent material processing that a kind of perform claim requires the multi-state failure prediction method of the 1 chemical spent material processing equipment
The multi-state failure prediction system of equipment, which is characterized in that the multi-state failure prediction system of the chemical spent material processing equipment
Include:
Current detection module is connect with main control module, for detecting chemical spent material processing equipment work electricity by current sensor
Flow data;
Voltage detection module is connect with main control module, for detecting chemical spent material processing equipment work electricity by voltage sensor
Press data;
Noise detection module is connect with main control module, is made an uproar for detecting the work of chemical spent material processing equipment by noise transducer
Sound data;
Main control module constructs module, fault diagnosis with current detection module, voltage detection module, noise detection module, fault database
Module, warning module, detection data memory module, display module connection, for controlling the normal work of modules by single-chip microcontroller
Make;
Fault database constructs module, connect with main control module, for constructing fault case library by data construction procedures;
Fault diagnosis module is connect with main control module, for diagnosing chemical spent material according to noise data by fault diagnostic program
Processing equipment fault type;
Warning module is connect with main control module, for carrying out pre-alert notification according to diagnostic result by alarm;
Detection data memory module, connect with main control module, for electric current, the voltage, noise, event by memory storage detection
Hinder case library, fault diagnosis result;
Display module is connect with main control module, for showing electricity when detection chemical spent material processing equipment work by display
Stream, voltage, noise, fault case library, fault diagnosis result.
6. a kind of multi-state failure prediction method using chemical spent material processing equipment described in Claims 1 to 5 any one
Chemical industry equipment failure predication platform.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910036826.6A CN109739197A (en) | 2019-01-15 | 2019-01-15 | A kind of multi-state failure prediction method of chemical spent material processing equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910036826.6A CN109739197A (en) | 2019-01-15 | 2019-01-15 | A kind of multi-state failure prediction method of chemical spent material processing equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109739197A true CN109739197A (en) | 2019-05-10 |
Family
ID=66364800
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910036826.6A Pending CN109739197A (en) | 2019-01-15 | 2019-01-15 | A kind of multi-state failure prediction method of chemical spent material processing equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109739197A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178378A (en) * | 2019-11-07 | 2020-05-19 | 腾讯科技(深圳)有限公司 | Equipment fault prediction method and device, electronic equipment and storage medium |
WO2021031091A1 (en) * | 2019-08-19 | 2021-02-25 | Siemens Aktiengesellschaft | Method and apparatus for detecting operating status of device, and computer readable medium |
CN115007624A (en) * | 2022-06-07 | 2022-09-06 | 浙江师范大学 | Bioengineering waste material environmental protection type discharging method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102980610A (en) * | 2012-11-16 | 2013-03-20 | 智谱特环境科技(北京)有限公司 | Intelligent fault detection device of continuous emission monitoring system (CEMS) |
CN104502742A (en) * | 2014-11-26 | 2015-04-08 | 华中科技大学 | Voltage fault monitoring system and method for electron irradiation accelerator |
CN107862052A (en) * | 2017-11-08 | 2018-03-30 | 国网电力科学研究院武汉南瑞有限责任公司 | A kind of fault case storehouse, fault tree and fault spectrum construction method |
US20180284737A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for detection in an industrial internet of things data collection environment with large data sets |
CN108827671A (en) * | 2018-03-21 | 2018-11-16 | 南京航空航天大学 | A kind of Trouble Diagnostic Method of Machinery Equipment |
CN108871761A (en) * | 2018-06-07 | 2018-11-23 | 广东石油化工学院 | A kind of initial failure of gear feature extracting method |
-
2019
- 2019-01-15 CN CN201910036826.6A patent/CN109739197A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102980610A (en) * | 2012-11-16 | 2013-03-20 | 智谱特环境科技(北京)有限公司 | Intelligent fault detection device of continuous emission monitoring system (CEMS) |
CN104502742A (en) * | 2014-11-26 | 2015-04-08 | 华中科技大学 | Voltage fault monitoring system and method for electron irradiation accelerator |
US20180284737A1 (en) * | 2016-05-09 | 2018-10-04 | StrongForce IoT Portfolio 2016, LLC | Methods and systems for detection in an industrial internet of things data collection environment with large data sets |
CN107862052A (en) * | 2017-11-08 | 2018-03-30 | 国网电力科学研究院武汉南瑞有限责任公司 | A kind of fault case storehouse, fault tree and fault spectrum construction method |
CN108827671A (en) * | 2018-03-21 | 2018-11-16 | 南京航空航天大学 | A kind of Trouble Diagnostic Method of Machinery Equipment |
CN108871761A (en) * | 2018-06-07 | 2018-11-23 | 广东石油化工学院 | A kind of initial failure of gear feature extracting method |
Non-Patent Citations (3)
Title |
---|
樊鹏: "一种噪声环境下的复杂声音识别方法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
肖智宏: "一种提高光学电压传感器温度稳定性的方法", 《电工技术学报》 * |
薛永端: "小电流接地故障暂态分析及区段定位新方法", 《电力***自动化》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021031091A1 (en) * | 2019-08-19 | 2021-02-25 | Siemens Aktiengesellschaft | Method and apparatus for detecting operating status of device, and computer readable medium |
CN111178378A (en) * | 2019-11-07 | 2020-05-19 | 腾讯科技(深圳)有限公司 | Equipment fault prediction method and device, electronic equipment and storage medium |
CN111178378B (en) * | 2019-11-07 | 2023-05-16 | 腾讯科技(深圳)有限公司 | Equipment fault prediction method and device, electronic equipment and storage medium |
CN115007624A (en) * | 2022-06-07 | 2022-09-06 | 浙江师范大学 | Bioengineering waste material environmental protection type discharging method |
CN115007624B (en) * | 2022-06-07 | 2023-05-23 | 浙江师范大学 | Environment-friendly discharging method for bioengineering waste |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110223196B (en) | Anti-electricity-stealing analysis method based on typical industry feature library and anti-electricity-stealing sample library | |
CN105653444B (en) | Software defect fault recognition method and system based on internet daily record data | |
CN110097297A (en) | A kind of various dimensions stealing situation Intellisense method, system, equipment and medium | |
CN110320892A (en) | The sewage disposal device fault diagnosis system and method returned based on Lasso | |
CN109416531A (en) | The different degree decision maker of abnormal data and the different degree determination method of abnormal data | |
CN109739197A (en) | A kind of multi-state failure prediction method of chemical spent material processing equipment | |
CN106338706A (en) | Electric energy metering device overall error detecting method, device and system | |
JP2012137934A (en) | Abnormality detection/diagnostic method, abnormality detection/diagnostic system, abnormality detection/diagnostic program and company asset management/facility asset management system | |
JP2013041448A (en) | Method of abnormality detection/diagnosis and system of abnormality detection/diagnosis | |
CN104677997B (en) | A kind of transformer oil chromatographic on-line monitoring differentiation method for early warning | |
CN103103570B (en) | Based on the aluminium cell condition diagnostic method of pivot similarity measure | |
CN108287327A (en) | Metering automation terminal fault diagnostic method based on Bayes's classification | |
Eickmeyer et al. | Data Driven Modeling for System-Level Condition Monitoring on Wind Power Plants. | |
CN115544140A (en) | Photovoltaic power station performance detection system and method based on artificial intelligence | |
CN110968703B (en) | Method and system for constructing abnormal metering point knowledge base based on LSTM end-to-end extraction algorithm | |
Jiao et al. | Data-driven evaluation for error states of standard electricity meters on automatic verification assembly line | |
CN111784537A (en) | Power distribution network state parameter monitoring method and device and electronic equipment | |
CN117554821A (en) | Energy storage battery on-line monitoring management method and system based on deep learning network | |
CN115908082A (en) | Enterprise pollution discharge monitoring method and device based on electricity utilization characteristic indexes | |
CN115310693A (en) | Comprehensive energy data cleaning method, system, equipment and medium | |
CN112256735B (en) | Power consumption monitoring method and device, computer equipment and storage medium | |
CN115127605B (en) | Remote intelligent diagnosis system and method for water quality automatic monitoring system | |
Traxler | Fault detection of large amounts of photovoltaic systems | |
CN116502156B (en) | Intelligent identification method and system for light CT abnormal state of converter station | |
Dhingra et al. | A machine learning based fault identification framework for smart grid automation |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |
|
RJ01 | Rejection of invention patent application after publication |