CN101344576A - Judging functional failure of electromechanical by speech recognition technology - Google Patents
Judging functional failure of electromechanical by speech recognition technology Download PDFInfo
- Publication number
- CN101344576A CN101344576A CNA2007101306175A CN200710130617A CN101344576A CN 101344576 A CN101344576 A CN 101344576A CN A2007101306175 A CNA2007101306175 A CN A2007101306175A CN 200710130617 A CN200710130617 A CN 200710130617A CN 101344576 A CN101344576 A CN 101344576A
- Authority
- CN
- China
- Prior art keywords
- language
- machine
- voice
- speech recognition
- mechanical
- 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
Images
Landscapes
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention develops the steamer and mechanical-electrical failures detecting method which exploits the application of voice identification technique into a novel field developed 'from machine identifying human language to machine identifying the mechanical-electrical language'. On theoretic application, a voiceprint identification method of voice identification technique is applied to the mechanical-electrical failure identification. The mechanical sound waves and ultrasonic are collected by a microphone and an ultrasonic sensor for the system and then are recorded into a voice identification SBC, and then the characteristics extraction, matching computation and sample model comparison processing are carried out for the machine 'voice' signals in virtue of functions of software in the SBC. For a sample model extraction process, the data storing capability of a computer is adopted to compare a plurality of well 'trained' language samples with the machine 'language' signals which pass the characteristic extraction and are just now recorded into an SCM one by one. Finally the optimal result and conclusion can be obtained.
Description
Technical field:
This invention is to utilize speech recognition technology to judge the intelligent automatic control detection system of the electrical category apparatus failure in the industrial circle.Range of application mainly comprises the fault diagnosis of turbodynamo, the fault diagnosis of mechanical electric apparatus classes such as the generally fault diagnosis of motor, and automobile etc.
Background technology:
Present equipment fault diagnosis for general electrical category, the method that adopts is the electronic product according to different electrical equipment types, adopt sensors of various types, the information such as at work electric current of electromechanical equipment, voltage, temperature, humidity of gathering are imported corresponding analysis equipment into, as computing machine or single-chip microcomputer etc., carry out quantitatively, analyze qualitatively, detect.All right for simple equipment and simple fault, deal with just very difficult for complicated situation.Most of sensor is a contact, and is also very inconvenient on using.Although modern age is for some complex apparatus, adopted means such as UT (Ultrasonic Testing), laser detection, but also there are many problems in cost and effect aspect in some fields, for example aspect the steam turbine fault detect, just can not accomplish the clearly discovery and the effectively detection of the initial stage of fault.What this invention was adopted is a kind of brand-new contactless measurement monitoring method and Theoretical Calculation means.Be to utilize speech recognition technology to judge and detect, monitor a kind of completely new approach of the fault of electrical category.
Summary of the invention:
So-called speech recognition is meant the automatic identification that the utilization computer system is carried out content that voice carried and speaker's pronunciation character etc.The computer speech identifying is consistent with the people to the voice recognition processing process basically.The speech recognition technology of main flow is based on the basic theories of statistical model identification at present.A complete speech recognition system can roughly be divided into three parts:
(1) phonetic feature extracts: purpose is to extract time dependent phonetic feature sequence from speech waveform.
(2) acoustic model and pattern match (recognizer): acoustic model is the bottom model of recognition system, and is the part of most critical in the speech recognition system.Acoustic model is produced by training by the phonetic feature that obtains usually, and purpose is to set up the pronunciation template for each pronunciation.Phonetic feature with the unknown when identification mates and compares with acoustics model (pattern), calculates the feature vector sequence of unknown voice and the distance between each pronunciation template.The design of acoustic model is closely related with the language pronouncing characteristics.Acoustic model cell size (word pronunciation model, semitone joint model or phoneme model) is to voice training data volume size, system recognition rate, and dirigibility has considerable influence.
(3) semantic understanding: computing machine carries out grammer, semantic analysis to recognition result.Understand meaning of language so that make corresponding reaction.
At present: speech recognition technology is comparative maturity, and this invention is to utilize the specific identification method of this technology then the fault diagnosis that is used for electrical category.The mechanical sound wave that we can send general electrical category at work, ultrasound wave is regarded a kind of special machinery " language " as, and this " language " can express the various information of machinery, normally with abnormal job information or the like.These " language " information are put into different categories also as speech recognition technology, adopt phonetic feature to extract, acoustic model and pattern match (recognizer), technology such as semantic understanding, thus finish diagnosis of technique for the electrical category fault.
For example to be applied to the fault diagnosis effect of steam turbine be very significant to this technology.In the past, human fault diagnosis for steam turbine and general electrical category was to adopt artificial diagnosis, mainly adopted ear to listen mode.From twentieth century seventies, along with developing rapidly of artificial intelligence theory, signal processing technology, electronic technology, singlechip technology, make and use that the intelligent diagnostics technology is carried out real-time status monitoring to the complicated like this system of genset and fault diagnosis becomes possibility.But present intelligent diagnostics technology is to rest on ultrasound wave substantially, the carrying out on the physics monitoring method of laser etc.The category of on a certain frequency that also only rests on a certain frequency domain or time domain in the application of mathematical theory or a certain waveform, studying.
And this invention is expanded and enlarged method to the electrical category fault detect.The reality of speech recognition technology and theoretical application extension have been gone out " from the machine recognition human language to a machine recognition electrical category language " new field.On mathematical theory was used, it was the method that time-domain and frequency-domain is superimposed simultaneously and studies that the vocal print in the speech recognition technology is differentiated, and is a kind of brand-new technology.With it fault " language " research that is applied to electrical category is the principal character of this invention.This invention is with three parts of speech recognition system, being that phonetic feature extraction, acoustic model and pattern match (recognizer), semantic understanding are diverted to phonetic feature extraction, acoustic model and the pattern match (recognizer) of machine " language ", the semantic understanding of machine " language ", is the taproot content of this technology.On this basis, again according to different machine " language " feature, revise relevant mathematical computations parameter, reach and to detect and the very fine variation of prosecution electromechanical equipment in line process, play the unreachable effect of similar detection means for the initial stage diagnosis of fault.
Description of drawings:
Fig. 1 is the schematic diagram of total system; Electromechanical equipment sends mechanical sound wave and ultrasound wave, system's process microphone and ultrasonic sensor are to mechanical sound wave and hyperacoustic collection, be entered into speech recognition tailored version single card microcomputer, and handle with feature extraction, coupling calculating, sample pattern comparison that the software function in the single card microcomputer is carried out machine " voice " signal.In extracting the process of sample pattern is to utilize data storage capacities powerful in the computing machine, with hundreds ofly proposing one by one to thousands of language sample of being " trained ", with machine " language " signal fusing of passing through feature extraction of single-chip microcomputer typing just.Obtain best comparison result.And, from known database, obtain which kind of Fault Diagnosis answer at this comparison result.
Fig. 2 is the system unit synoptic diagram
Fig. 3 is system's external form synoptic diagram
Specific implementation method:
The computer processing system of this invention is made up of following major part.Hardware components is single card microcomputer, desk-top computer or the notebook computer of wide band microphone or pole type microphone and ultrasonic sensor, voice collecting and processing usefulness, RS485 or 422 ports and the corresponding transmission cable that telecommunication is used.
Software section is machine " voice " sample database after the input of machine " voice " sensing, feature extraction, coupling calculating, sample pattern, the training.Detailed process is, the mechanical sound wave that just sending at active machine and ultrasound wave is input to voice collecting with wide band microphone or pole type microphone and ultrasonic sensor collection and handles the single card microcomputer of usefulness.Utilize the interior software of single card microcomputer to carry out feature extraction, coupling calculating.The machine " language " of this time period is compared one by one with various machines " language " sample in the computing machine sample database, find out best matched sample, reach a conclusion.The mode of this conclusion with figure and data representation is presented on the screen, carries out craft or fault distinguishing and technology control automatically for the technician.Various machines " language " sample in the computing machine sample database before this, through repeatedly, machine " language " under the various different condition is entered in the Computer Database.This process is also referred to as " training " process in speech recognition technology.For new sample signal, computing machine is realized " benefit goes into to increase " automatically by software, and " study automatically " reaches intelligentized purpose.
Claims (3)
1 utilizes speech recognition technology to judge steam turbine and electrical category fault, realizes the technique extension of " from the machine recognition human language to machine recognition electrical category language ".
2 use the vocal print method of discrimination in the speech recognition technologies, i.e. the time-domain and frequency-domain method that is superimposed simultaneously and studies, and the fault " language " of differentiating the steam turbine electrical category is the principal character of this invention.
3 system works principles are that electromechanical equipment is sent mechanical sound wave and ultrasound wave, to mechanical sound wave and ultrasound wave collection and be entered into single card microcomputer, and carry out with the software function in the voice identification form plate machine that feature extraction, coupling to machine " voice " signal calculated, the sample pattern comparison is handled by microphone and ultrasonic sensor.In extracting the process of sample pattern is to utilize data storage capacities powerful in the computing machine, with hundreds ofly proposing one by one to thousands of language sample of being " trained ", with machine " language " signal fusing of passing through feature extraction of single-chip microcomputer typing just.Obtain best comparison result.And, from known database, obtain which kind of Fault Diagnosis answer at this comparison result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101306175A CN101344576A (en) | 2007-07-11 | 2007-07-11 | Judging functional failure of electromechanical by speech recognition technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007101306175A CN101344576A (en) | 2007-07-11 | 2007-07-11 | Judging functional failure of electromechanical by speech recognition technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101344576A true CN101344576A (en) | 2009-01-14 |
Family
ID=40246611
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2007101306175A Pending CN101344576A (en) | 2007-07-11 | 2007-07-11 | Judging functional failure of electromechanical by speech recognition technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101344576A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102650960A (en) * | 2012-03-31 | 2012-08-29 | 奇智软件（北京）有限公司 | Method and device for eliminating faults of terminal equipment |
| CN103591975A (en) * | 2013-11-20 | 2014-02-19 | 深圳市航盛电子股份有限公司 | Ultrasonic wave sensor index detection method and device |
| CN104139731A (en) * | 2014-07-17 | 2014-11-12 | 广汽吉奥汽车有限公司 | Automobile structure detection method and system |
| CN109132758A (en) * | 2018-08-17 | 2019-01-04 | 寿县理康信息技术服务有限公司 | A kind of elevator operation monitoring system |
| CN109495817A (en) * | 2017-09-11 | 2019-03-19 | 星电株式会社 | Sound processing apparatus |
| CN111017670A (en) * | 2019-12-23 | 2020-04-17 | 江苏省特种设备安全监督检验研究院 | Voiceprint recognition and fault diagnosis monitoring and alarming method for elevator abnormity |
| CN111247443A (en) * | 2018-09-29 | 2020-06-05 | 深圳市大疆创新科技有限公司 | Motor state monitoring device and motor state monitoring method |
-
2007
- 2007-07-11 CN CNA2007101306175A patent/CN101344576A/en active Pending
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102650960A (en) * | 2012-03-31 | 2012-08-29 | 奇智软件（北京）有限公司 | Method and device for eliminating faults of terminal equipment |
| CN103591975A (en) * | 2013-11-20 | 2014-02-19 | 深圳市航盛电子股份有限公司 | Ultrasonic wave sensor index detection method and device |
| CN103591975B (en) * | 2013-11-20 | 2016-05-11 | 深圳市航盛电子股份有限公司 | A kind of ultrasonic sensor index detection method and device |
| CN104139731A (en) * | 2014-07-17 | 2014-11-12 | 广汽吉奥汽车有限公司 | Automobile structure detection method and system |
| CN104139731B (en) * | 2014-07-17 | 2016-07-13 | 广汽吉奥汽车有限公司 | A kind of vehicle structure detection method and system |
| CN109495817A (en) * | 2017-09-11 | 2019-03-19 | 星电株式会社 | Sound processing apparatus |
| CN109495817B (en) * | 2017-09-11 | 2022-02-25 | 星电株式会社 | Sound processing device |
| CN109132758A (en) * | 2018-08-17 | 2019-01-04 | 寿县理康信息技术服务有限公司 | A kind of elevator operation monitoring system |
| CN111247443A (en) * | 2018-09-29 | 2020-06-05 | 深圳市大疆创新科技有限公司 | Motor state monitoring device and motor state monitoring method |
| CN111017670A (en) * | 2019-12-23 | 2020-04-17 | 江苏省特种设备安全监督检验研究院 | Voiceprint recognition and fault diagnosis monitoring and alarming method for elevator abnormity |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dahake et al. | Speaker dependent speech emotion recognition using MFCC and Support Vector Machine | |
| CN101344576A (en) | Judging functional failure of electromechanical by speech recognition technology | |
| Rosales-Pérez et al. | Classifying infant cry patterns by the genetic selection of a fuzzy model | |
| CN105841797A (en) | Window motor abnormal noise detection method and apparatus based on MFCC and SVM | |
| CN102005070A (en) | Voice identification gate control system | |
| CN105895078A (en) | Speech recognition method used for dynamically selecting speech model and device | |
| CN110444202B (en) | Composite voice recognition method, device, equipment and computer readable storage medium | |
| CN102592593A (en) | Emotional-characteristic extraction method implemented through considering sparsity of multilinear group in speech | |
| Özseven et al. | SPeech ACoustic (SPAC): A novel tool for speech feature extraction and classification | |
| CN113314144A (en) | Voice recognition and power equipment fault early warning method, system, terminal and medium | |
| Zhang et al. | A supervised machine learning-based sound identification for construction activity monitoring and performance evaluation | |
| CN112985574B (en) | High-precision classification identification method for optical fiber distributed acoustic sensing signals based on model fusion | |
| CN115932659A (en) | Transformer fault detection method based on voiceprint characteristics | |
| Rahman et al. | Dynamic time warping assisted svm classifier for bangla speech recognition | |
| CN114373452A (en) | Voice abnormity identification and evaluation method and system based on deep learning | |
| CN117169639B (en) | Product detection method and system for power adapter production | |
| CN113571095B (en) | Speech emotion recognition method and system based on nested deep neural network | |
| CN117437916A (en) | Navigation system and method for inspection robot | |
| Maniak et al. | Automated sound signalling device quality assurance tool for embedded industrial control applications | |
| Prakash et al. | Analysis of emotion recognition system through speech signal using KNN & GMM classifier | |
| CN111862991A (en) | Method and system for identifying baby crying | |
| Shome et al. | Speaker Recognition through Deep Learning Techniques: A Comprehensive Review and Research Challenges | |
| CN114839960A (en) | Method and system for detecting vehicle fault based on artificial intelligence algorithm | |
| CN108074585A (en) | A kind of voice method for detecting abnormality based on sound source characteristics | |
| CN114550711A (en) | Cable surrounding environment sound identification method based on time-frequency attention network model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090114 |