CN101226162A - Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity - Google Patents
Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity Download PDFInfo
- Publication number
- CN101226162A CN101226162A CNA2008100693688A CN200810069368A CN101226162A CN 101226162 A CN101226162 A CN 101226162A CN A2008100693688 A CNA2008100693688 A CN A2008100693688A CN 200810069368 A CN200810069368 A CN 200810069368A CN 101226162 A CN101226162 A CN 101226162A
- Authority
- CN
- China
- Prior art keywords
- gas
- neural network
- sensitivity
- artificial neural
- characteristic
- 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
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
- Investigating Or Analyzing Materials By The Use Of Electric Means (AREA)
Abstract
The invention discloses an intelligent method for inhibiting cross sensitivity of gas sensor, which uses intrinsic gas characteristic spectrum amplitude as the input vector of neural network, calculates via an artificial neural network based on a neural network structure marked and trained via standard gas to further eliminate the cross sensitivity of gas sensor caused by laminated frequency to obtain single gas content and improve inhibit effect, thereby improving gas detection precision and accuracy. The invention uses wavelet analysis to input characteristic vector to the artificial neutral network and uses the localization ability of wavelet analysis in time and frequency domains, to analyze characteristic gas response waveform and cross sensitive gas response waveform, to further improving the inhibition effect and improving gas detection precision and accuracy.
Description
Technical field
The present invention relates to a kind of data analysing method, particularly a kind of intelligent method that suppresses gas-sensitive sensor decussation sensitivity.
Background technology
Transformer oil dissolved gas content is the important parameter of reflection transformer built-in electrical insulation situation and operation conditions, and the oil dissolved gas on-line monitoring of transformer is the important means that reflects gas content in the oil in real time, continuously.Semiconductor gas sensor can detected gas content, it is converted into specific voltage signal, it is the important channel of detected gas content.At present, the technology of gas content is very extensive in the application semiconductor gas sensor monitoring oil.Though specific semiconductor gas sensor mainly detects specific pure gas, sensor has response to the poor selectivity of gas to multiple gases, and this non-single selective is determined by its sensitive mechanism.Adopt certain method (as adding an amount of precious metal Pt, Pd adds catalyzer, manufacturing process and work temperature etc.) to improve sensitivity, but still can certain sensitivity be arranged, " cross sensitivity " of Here it is gas sensor other gas to a certain gas." cross sensitivity " characteristic will influence specific gas content monitoring accurately, therefore, the mode of eliminating the gas sensor cross sensitivity also occurs in succession.In the prior art, using more is inhibition method at the cross sensitivity of fiber-optic grating sensor, and it is a kind of non-pneumatic sensor.
The method that the cross sensitivity influence that utilizes the neural network method to eliminate gas sensor is arranged in the prior art, but related associated gas sensor decussation sensitivity characteristic aspect content does not have the cross sensitivity inhibition method under many gas sensors condition, research be from suppressing and aspect content such as cross sensitivity dynamic perfromance, do not relate to for the content of the multisensor cross sensitivity of transformer oil dissolved gas on-line monitoring.And neural network is used for the diagnosis and the Fault Pattern Recognition of insulation fault more in transformer fault diagnosis, does not suppress cross sensitivity.
Therefore, the inhibition method that needs a kind of gas-sensitive sensor decussation sensitivity, at the cross sensitivity problem of existing many gas sensors, can suppress many gas sensors cross sensitivity influences the pure gas accuracy of detection, thereby improves dissolved gas accuracy of detection and accuracy in the electric power transformer oil.
Summary of the invention
The object of the present invention is to provide a kind of intelligent method that suppresses gas-sensitive sensor decussation sensitivity, can suppress the cross sensitivity of gas sensor, improve the monitoring accuracy of pure gas, improve gas detection precision and accuracy.
The intelligent method of inhibition gas-sensitive sensor decussation sensitivity of the present invention may further comprise the steps:
A. gather at least two gas sensor response wave shapes;
B. obtain the feature spectrogram of at least two kinds of gas responses, draw the characteristic gas response amplitude of particular sensor, eliminate the response spectra figure amplitude of cross sensitivity gas;
C. utilize the BP artificial neural network further to suppress gas-sensitive sensor decussation sensitivity:
I. make and contain characteristic gas and the dense mixed gas of known gas is imported by BP artificial neural network input layer, adjust the input layer and the hidden layer weight matrix of BP artificial neural network, make output and the input gas concentrated phase of artificial neural network etc., thereby eliminate the influence of cross sensitivity;
The characteristic gas amplitude of step b is imported by BP artificial neural network input layer, the weight matrix that utilizes step I to determine is handled the input data, data behind the elimination cross sensitivity as output, are responded thereby obtain pure gas, thereby obtain gas content in the mixed gas.
Further, obtain the mode of gas response characteristic spectrogram among the step b for obtaining by Wave data being carried out the wavelet analysis processing.
The invention has the beneficial effects as follows: the intelligent method of inhibition gas-sensitive sensor decussation sensitivity of the present invention, the present invention is with the input vector of intrinsic gas characteristic spectral magnitude as neural network, on the basis of the neural network structure of demarcating and training with calibrating gas, calculate by artificial neural network, further eliminated because the overlapping gas-sensitive sensor decussation sensitivity degree that causes of portions of the spectrum, obtain pure gas content, increase inhibition effect of the present invention, thereby improve gas detection precision and accuracy; The characteristic gas spectral magnitude that adopts method of wavelet analysis to obtain provides input feature value as artificial neural network, utilize wavelet analysis to have good localization ability in time domain and frequency domain, can better analyze its characteristic gas response wave shape and cross sensitivity gas response wave shape, further increase inhibition effect of the present invention, thereby improve gas detection precision and accuracy.
Description of drawings
Below in conjunction with drawings and Examples the present invention is further described.
Fig. 1 is a schematic diagram of the present invention;
Fig. 2 detects comparison diagram for the present invention is intelligent with MQ type gas sensor.
Embodiment
Present embodiment is to utilize the present invention to carry out the content analysis of dissolved gas in the electric power transformer oil.
Fig. 1 is a schematic diagram of the present invention, and Fig. 2 is intelligent and MQ type sensor comparison diagram for the present invention, and as shown in the figure: mixed gas 1 comprises H
2, C
2H
4, CO, C
2H
2, CH
4, C
2H
6After gas sensor 2 identifications, produce response wave shape separately, obtain and the pick-up transducers response wave shape, carry out small echo and handle 3, obtain the feature spectrogram of above six kinds of gases response, draw the characteristic gas response amplitude of particular sensor, the laggard input layer of going into BP artificial neural network 4 of the preliminary response spectra figure amplitude that suppresses cross sensitivity gas obtains the pure gas content by BP artificial neural network 4; BP artificial neural network 4 is for utilizing the input of concentration known mixed gas as artificial neural network, adjust the input layer and the hidden layer weight matrix of artificial neural network, make the output of artificial neural network equate, eliminate the influence of cross sensitivity with the input gas concentration.
The intelligent method of inhibition gas-sensitive sensor decussation sensitivity of the present invention may further comprise the steps:
A. gather and comprise H
2, C
2H
4, CO, C
2H
2, CH
4, C
2H
6The mixed gas of six kinds of gases with respect to gas sensor response wave shape separately; As shown in Figure 2, there is the cross sensitivity that to eliminate in the MQ type gas sensor;
B. handle by Wave data being carried out wavelet analysis, obtain the feature spectrogram of above six kinds of gases response, draw the characteristic gas response amplitude of particular sensor, tentatively suppress the response spectra figure amplitude of cross sensitivity gas;
C. utilize the BP artificial neural network further to suppress gas-sensitive sensor decussation sensitivity:
The mixed gas that contains characteristic gas and known gas concentration is imported by BP artificial neural network input layer, adjust the input layer and the hidden layer weight matrix of BP artificial neural network, make output and the input gas concentrated phase of artificial neural network etc., thereby eliminate the influence of cross sensitivity;
The characteristic gas amplitude of step b is imported by BP artificial neural network input layer, the weight matrix that utilizes step I to determine is handled the input data, data behind the elimination cross sensitivity as output, are responded thereby obtain pure gas, thereby obtain gas content in the mixed gas.
As shown in Figure 2, by method of the present invention mixed gas is detected, apparent in view suppresses cross sensitivity.
Certainly, among the step b, the mode of obtaining characteristic frequency spectrum from transducing signal is not limited to method of wavelet analysis, also can be other mathematical model modes, can both reach purpose of the present invention.
Present embodiment is adopting wavelet analysis method, extract the characteristic frequency spectrum and the cross-inductive gas characteristic spectral magnitude of gas sensor response intrinsic gas, remove cross-inductive gas characteristic frequency spectrum, obtained the pure gas response amplitude, realized preliminary inhibition cross sensitivity effect.Then with the input vector of intrinsic gas characteristic spectral magnitude as neural network, on the basis of the neural network structure of having demarcated and having trained with calibrating gas, calculate by artificial neural network, further eliminated because the overlapping gas-sensitive sensor decussation sensitivity degree that causes of portions of the spectrum, increase inhibition effect of the present invention, thereby improve gas detection precision and accuracy.
Explanation is at last, above embodiment is only unrestricted in order to technical scheme of the present invention to be described, although the present invention is had been described in detail with reference to preferred embodiment, those of ordinary skill in the art is to be understood that, can make amendment or be equal to replacement technical scheme of the present invention, and not breaking away from the aim and the scope of technical solution of the present invention, it all should be encompassed in the middle of the claim scope of the present invention.
Claims (2)
1. intelligent method that suppresses gas-sensitive sensor decussation sensitivity is characterized in that: may further comprise the steps:
A. gather at least two gas sensor response wave shapes;
B. obtain the feature spectrogram of at least two kinds of gas responses, draw the characteristic gas response amplitude of particular sensor, eliminate the response spectra figure amplitude of cross sensitivity gas;
C. utilize the BP artificial neural network further to suppress gas-sensitive sensor decussation sensitivity:
I. make and contain characteristic gas and the dense mixed gas of known gas is imported by BP artificial neural network input layer, adjust the input layer and the hidden layer weight matrix of BP artificial neural network, make output and the input gas concentrated phase of artificial neural network etc., thereby eliminate the influence of cross sensitivity;
The characteristic gas amplitude of step b is imported by BP artificial neural network input layer, the weight matrix that utilizes step I to determine is handled the input data, data behind the elimination cross sensitivity as output, are responded thereby obtain pure gas, thereby obtain gas content in the mixed gas.
2. the intelligent method of inhibition gas-sensitive sensor decussation sensitivity according to claim 1 is characterized in that: the mode that obtains the characteristic spectrum figure amplitude of gas response among the step b obtains for carry out the wavelet analysis processing by Wave data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2008100693688A CN101226162A (en) | 2008-02-18 | 2008-02-18 | Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2008100693688A CN101226162A (en) | 2008-02-18 | 2008-02-18 | Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101226162A true CN101226162A (en) | 2008-07-23 |
Family
ID=39858281
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNA2008100693688A Pending CN101226162A (en) | 2008-02-18 | 2008-02-18 | Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101226162A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102087311A (en) * | 2010-12-21 | 2011-06-08 | 彭浩明 | Method for improving measurement accuracy of power mutual inductor |
CN108287183A (en) * | 2017-12-28 | 2018-07-17 | 东华大学 | A method of reducing cross sensitivity of the semiconductor hydrogen gas sensor to carbon monoxide |
-
2008
- 2008-02-18 CN CNA2008100693688A patent/CN101226162A/en active Pending
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102087311A (en) * | 2010-12-21 | 2011-06-08 | 彭浩明 | Method for improving measurement accuracy of power mutual inductor |
CN102087311B (en) * | 2010-12-21 | 2013-03-06 | 彭浩明 | Method for improving measurement accuracy of power mutual inductor |
CN108287183A (en) * | 2017-12-28 | 2018-07-17 | 东华大学 | A method of reducing cross sensitivity of the semiconductor hydrogen gas sensor to carbon monoxide |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101701982B (en) | Method for detecting harmonic waves of electric system based on window and interpolated FFT | |
CN108169639B (en) | Method for identifying switch cabinet fault based on parallel long-time and short-time memory neural network | |
CN101701984B (en) | Fundamental wave and harmonic wave detecting method based on three-coefficient Nuttall windowed interpolation FFT | |
EP3736578A3 (en) | Current measurement | |
AU2009331596A8 (en) | Method for detecting an electric arc in photovoltaic equipment | |
WO2003098769A3 (en) | Methods and apparatus for indicating a fault condition in fuel cells and fuel cell components | |
CN104483598A (en) | Cable fault detection and analysis method | |
CN102798798A (en) | Power transformer winding deformation detection method based on vibration analysis | |
WO2003073089A8 (en) | Method and device for performing impedance spectroscopy | |
CN105223444A (en) | The inner local overheating of oil-filled transformer judges and heat spot localization method | |
CN105301099A (en) | Food crispness detection method | |
CN102749495B (en) | Quantitative assessment method for stability of double-wire arc welding electrical signal | |
CN107064753A (en) | Bow net arc-plasma Multi-parameter Data Acquisition method and apparatus | |
CN109932053A (en) | A kind of state monitoring apparatus and method for high-voltage shunt reactor | |
CN112858845A (en) | Partial discharge diagnosis method for gas insulated switchgear | |
CN105527391A (en) | Electric-nose-analysis-based determination method of tuna oil corruption degree in storage process | |
CN204525045U (en) | A kind of Tool Wear Monitoring system based on electric current and sound emission composite signal | |
CN101226162A (en) | Intelligent method for inhibiting gas-sensitive sensor decussation sensitivity | |
CN109870614B (en) | Method for rapidly detecting early-stage faults of power equipment through half-cycle waves | |
CN101718816B (en) | Fundamental wave and harmonic wave detection method based on four-item coefficient Nuttall window interpolation FFT | |
WO2010068653A3 (en) | Combined optical and resistance measuring apparatus and method | |
CN109164427A (en) | A kind of detection method of noise of radar receiver power | |
CN106771928A (en) | A kind of online pick-up method of partial discharge pulse's initial time | |
CN109884483A (en) | Insulating tube type busbar shelf depreciation acoustics on-line monitoring method and device | |
CN114201991A (en) | Partial discharge signal detection method and system based on ultrasonic sensor array |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C12 | Rejection of a patent application after its publication | ||
RJ01 | Rejection of invention patent application after publication |
Open date: 20080723 |