CN108446656A - A kind of parser carrying out Selective recognition to kitchen hazardous gas - Google Patents
A kind of parser carrying out Selective recognition to kitchen hazardous gas Download PDFInfo
- Publication number
- CN108446656A CN108446656A CN201810262705.9A CN201810262705A CN108446656A CN 108446656 A CN108446656 A CN 108446656A CN 201810262705 A CN201810262705 A CN 201810262705A CN 108446656 A CN108446656 A CN 108446656A
- Authority
- CN
- China
- Prior art keywords
- sample
- parameter
- kitchen
- carrying
- matrix
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The parser proposed by the present invention that Selective recognition is carried out to kitchen hazardous gas, is related to information technology field, includes the following steps：Step 1, crosscheck, carry out parameter optimization；Step 2 extracts feature using wavelet transform combination principal component analysis；Step 3 preserves eigenmatrix and projection matrix；Step 4 carries out classified calculating by the KNN based on mahalanobis distance and keeps the covariance matrix of sample.The present invention has used for reference crosscheck during classifying, and optimal parameter pair is found in recognizer, to improve recognition accuracy.Since the value range of parameter may be very big, PCA transformation is carried out to being originally inputted again with obtained optimized parameter P after using progress parameter optimization is cross-checked based on the considerations of efficiency, then preserves eigenmatrix and projection matrix.The covariance matrix that mahalanobis distance calculates required each classification sample is also preserved simultaneously, to save the calculating cost of prediction process.
Description
Technical field
The present invention relates to information technology field, especially a kind of analysis carrying out Selective recognition to kitchen hazardous gas is calculated
Method.
Background technology
Kitchen is the place that operation is carried out using open fire, generally there is liquefied petroleum gas, coal gas, natural gas, charcoal using fuel
Deng during operation and use, if cannot be operated by chapter, it is easy to generate the things such as leakage, burning, explosion and sparrow gas poisoning
Therefore this is the one side for causing fire incident.At present to the detection meter selection sensor array of various gases in kitchen and
Recognizer appropriate, various other for detecting, analyzing and differentiate, the conventional method of gas identification is often only conceived to office
Portion such as extracts feature, so according to the various manual features of the signal waveform of gas sample design or by various signal processing methods
After simply enter and be identified in the sorting algorithms such as SVM, KNN；Or simple feature is only used, and in sorting algorithm
It is improved.
Invention content
The present invention provides a kind of parser carrying out Selective recognition to kitchen hazardous gas, improves recognition accuracy,
The calculating cost for saving prediction process, specifically includes following steps：
Step 1, crosscheck, carry out parameter optimization；
Step 2 extracts feature using wavelet transform combination principal component analysis；
Step 3 preserves eigenmatrix and projection matrix；
Step 4 carries out classified calculating by the KNN based on mahalanobis distance and keeps the covariance matrix of sample.
Preferably, the step 1 crosscheck the specific steps are：
Training sample is grouped by step 11, and the principle of grouping is to ensure to include all classes as possible in each group
Other gas, Fig. 3 by 3- roll over cross validation for, illustrate grouping method；
Step 12 is grouped each, other groupings is all used to be used as training sample；
Step 13 predicts this current grouping using KNN, obtains the classification accuracy on entire sample；
The bound and stepping of step 14 and then setup parameter P and K carry out trellis traversal, compare each parameter to obtaining
The classification accuracy arrived, by the highest parameter of classification accuracy to the optimized parameter as prediction unknown sample.
Preferably, feature extraction detailed process is in the step 2：
Each is passed individual each signal waveform progress wavelet transform in each sample by step 21
The response sequence of sensor intercepts identical length；
Step 22, fixed wavelet basis, all implement these sequences same transformation；
This wavelets coefficient is combined into new sample by step 23 by original sequence, and all samples pass through this
The processing of sample forms a sample matrix；
Step 24 implements this matrix of wavelet coefficients PCA transformation, obtains final eigenmatrix.
Preferably, being classified using KNN in the step 4, using mahalanobis distance as the similar of gas sample
Degree is weighed.
A kind of parser carrying out Selective recognition to kitchen hazardous gas provided by the invention, advantage exist
In：Crosscheck has been used for reference during classification, optimal parameter pair is found in recognizer, to improve recognition accuracy.
Since the value range of parameter may be very big, it is based on efficiency, after using crosscheck progress parameter optimization, with obtaining
Optimized parameter P carry out PCA transformation to being originally inputted again, then preserve eigenmatrix and projection matrix.Also preserve horse simultaneously
Family name's distance calculates the covariance matrix of required each classification sample, to save the calculating cost of prediction process.
Description of the drawings
Fig. 1 is the flow chart for the parser that the present invention carries out kitchen hazardous gas Selective recognition；
Fig. 2 is the specific flow chart of crosscheck；
Fig. 3 is 3- folding cross validation schematic diagrames；
Fig. 4 is the flow chart of feature extraction.
Specific implementation mode
To further illustrate that each embodiment, the present invention are provided with attached drawing.These attached drawings are that the invention discloses one of content
Point, mainly to illustrate embodiment, and the associated description of specification can be coordinated to explain the operation principles of embodiment.Cooperation ginseng
These contents are examined, those of ordinary skill in the art will be understood that other possible embodiments and advantages of the present invention.In figure
Component be not necessarily to scale, and similar component symbol is conventionally used to indicate similar component.
In conjunction with the drawings and specific embodiments, the present invention is further described.
As shown in Figure 1, a kind of parser carrying out Selective recognition to kitchen hazardous gas that the present embodiment proposes, main
Include the following steps：
Step 1, crosscheck, carry out parameter optimization；
Step 2 extracts feature using wavelet transform combination principal component analysis；
Step 3 preserves eigenmatrix and projection matrix；
Step 4 carries out classified calculating by the KNN based on mahalanobis distance and keeps the covariance matrix of sample.
This method sets pivot number P and KNN, and (K-Nearest Neighbor algorithm, closest node are calculated
Method) K as parameter, then optimized by cross-checking.It in this way can be improper to classification to avoid single parameter selection
Resulting negative influence.This method is carrying out PCA (Principle Component Analysis, principal component analysis) changes
It changes and does not set energy threshold, but directly using the pivot number of selection as a parameter.Parameter K in KNN be one very
Crucial parameter, is highly desirable to cross-check training sample, to find the value of optimal K.As shown in Fig. 2, step
Rapid 1, which cross-checks specific method, is：
Training sample is grouped by step 11, and the principle of grouping is to ensure to include all classes as possible in each group
Other gas, Fig. 3 by 3- roll over cross validation for, illustrate grouping method；
Step 12 is grouped each, other groupings is all used to be used as training sample；
Step 13 predicts this current grouping using KNN, obtains the classification accuracy on entire sample；
The bound and stepping of step 14 and then setup parameter P and K carry out trellis traversal, compare each parameter to obtaining
The classification accuracy arrived, by the highest parameter of classification accuracy to the optimized parameter as prediction unknown sample.
Wherein, in step 2, feature extraction is carried out to gas sample by wavelet transform combination principal component analysis, such as
Detailed process shown in Fig. 4 is：
Each is passed individual each signal waveform progress wavelet transform in each sample by step 21
The response sequence of sensor intercepts identical length；
Step 22, fixed wavelet basis, all implement these sequences same transformation；
This wavelets coefficient is combined into new sample by step 23 by original sequence, and all samples pass through this
The processing of sample forms a sample matrix；
Step 24 implements this matrix of wavelet coefficients PCA transformation, obtains final eigenmatrix.
Step 2 is amplified using the change commanders difference of sensor of discrete wavelet transformer, the main composition converted using PCA
These are remained by the information highlighted, indicates that the ability of data greatly enhances.
The feature of sample has been subjected to dimension-reduction treatment in step 2, has been classified using KNN in step 4.Mahalanobis distance
It is a kind of method of the effective similarity for calculating two unknown sample collection, uses mahalanobis distance as gas sample in KNN
Measuring similarity.Gas sample similarity calculation process is as follows：
If known gas sampleWith unknown gas sampleTwo samples
This similarity is calculated by mahalanobis distance,
Wherein Σ indicates the covariance matrix of classification described in known sample, if the category is expressed as
Then covariance matrix is calculated by following formula：
Wherein, μiIt indicatesExpectation.For each classification in gas sample, it is required for its covariance of calculated in advance
Matrix is calculated by these covariance matrixes at a distance from known sample then for new samples.
Crosscheck has been used for reference during classification, and optimal parameter pair is found in recognizer, it is accurate to improve identification
True rate.Since the value range of parameter may be very big, based on the considerations of efficiency, after using progress parameter optimization is cross-checked,
PCA transformation is carried out to being originally inputted again with obtained optimized parameter P, then preserves eigenmatrix and projection matrix.Simultaneously also
The covariance matrix that mahalanobis distance calculates required each classification sample is preserved, to save the calculating cost of prediction process.
Although specifically showing and describing the present invention in conjunction with preferred embodiment, those skilled in the art should be bright
In vain, it is not departing from the spirit and scope of the present invention defined by the appended claims, it in the form and details can be right
The present invention makes a variety of changes, and is protection scope of the present invention.
Claims (4)
1. a kind of parser carrying out Selective recognition to kitchen hazardous gas, which is characterized in that mainly include the following steps that：
Step 1, crosscheck, carry out parameter optimization；
Step 2 extracts feature using wavelet transform combination principal component analysis；
Step 3 preserves eigenmatrix and projection matrix；
Step 4 carries out classified calculating by the KNN based on mahalanobis distance and keeps the covariance matrix of sample.
2. the parser according to claim 1 for carrying out Selective recognition to kitchen hazardous gas, it is characterised in that：Institute
State step 1 crosscheck the specific steps are：
Training sample is grouped by step 11, and the principle of grouping is to ensure to include all categories as possible in each group
Gas, Fig. 3 by 3- roll over cross validation for, illustrate grouping method；
Step 12 is grouped each, other groupings is all used to be used as training sample；
Step 13 predicts this current grouping using KNN, obtains the classification accuracy on entire sample；
The bound and stepping of step 14 and then setup parameter P and K carries out trellis traversal, compares each parameter to obtaining
Classification accuracy, by the highest parameter of classification accuracy to the optimized parameter as prediction unknown sample.
3. the parser according to claim 1 for carrying out Selective recognition to kitchen hazardous gas, it is characterised in that：Institute
Stating feature extraction detailed process in step 2 is：
Step 21, for each sample, wavelet transform is carried out to individual each signal waveform, i.e., by each sensor
Response sequence intercept identical length；
Step 22, fixed wavelet basis, all implement these sequences same transformation；
This wavelets coefficient is combined into new sample by step 23 by original sequence, as all samples pass through
Processing forms a sample matrix；
Step 24 implements this matrix of wavelet coefficients PCA transformation, obtains final eigenmatrix.
4. the parser according to claim 1 for carrying out Selective recognition to kitchen hazardous gas, it is characterised in that：Institute
It states in step 4 and is classified using KNN, using mahalanobis distance as the measuring similarity of gas sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810262705.9A CN108446656A (en) | 2018-03-28 | 2018-03-28 | A kind of parser carrying out Selective recognition to kitchen hazardous gas |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810262705.9A CN108446656A (en) | 2018-03-28 | 2018-03-28 | A kind of parser carrying out Selective recognition to kitchen hazardous gas |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108446656A true CN108446656A (en) | 2018-08-24 |
Family
ID=63197559
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810262705.9A Pending CN108446656A (en) | 2018-03-28 | 2018-03-28 | A kind of parser carrying out Selective recognition to kitchen hazardous gas |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108446656A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110175195A (en) * | 2019-04-23 | 2019-08-27 | 哈尔滨工业大学 | Mixed gas detection model construction method based on extreme random tree |
| CN113139963A (en) * | 2021-06-22 | 2021-07-20 | 常州微亿智造科技有限公司 | Defect detection method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106680241A (en) * | 2017-01-13 | 2017-05-17 | 北京化工大学 | Novel spectrum multi-analysis classification and identification method and application thereof |
| CN107292225A (en) * | 2016-08-18 | 2017-10-24 | 北京师范大学珠海分校 | A kind of face identification method |
| CN107506700A (en) * | 2017-08-07 | 2017-12-22 | 苏州经贸职业技术学院 | Pedestrian's recognition methods again based on the study of broad sense similarity measurement |
-
2018
- 2018-03-28 CN CN201810262705.9A patent/CN108446656A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107292225A (en) * | 2016-08-18 | 2017-10-24 | 北京师范大学珠海分校 | A kind of face identification method |
| CN106680241A (en) * | 2017-01-13 | 2017-05-17 | 北京化工大学 | Novel spectrum multi-analysis classification and identification method and application thereof |
| CN107506700A (en) * | 2017-08-07 | 2017-12-22 | 苏州经贸职业技术学院 | Pedestrian's recognition methods again based on the study of broad sense similarity measurement |
Non-Patent Citations (1)
| Title |
|---|
| 胡晓楠: "基于深度学习的气体识别研究", 《CNKI硕士电子期刊》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110175195A (en) * | 2019-04-23 | 2019-08-27 | 哈尔滨工业大学 | Mixed gas detection model construction method based on extreme random tree |
| CN110175195B (en) * | 2019-04-23 | 2022-11-29 | 哈尔滨工业大学 | Mixed gas detection model construction method based on extreme random tree |
| CN113139963A (en) * | 2021-06-22 | 2021-07-20 | 常州微亿智造科技有限公司 | Defect detection method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108362662B (en) | Near infrared spectrum similarity calculation method and device and substance qualitative analysis system | |
| KR101852074B1 (en) | Electronic Nose System and Method for Gas Classification | |
| CN109934269B (en) | Open set identification method and device for electromagnetic signals | |
| WO2019041488A1 (en) | Toxic and harmful gas detection and recognition method based on machine olfaction | |
| CN109002859B (en) | Sensor array feature selection and array optimization method based on principal component analysis | |
| CN107818298A (en) | General Raman spectral characteristics extracting method for machine learning material recognition | |
| KR100795227B1 (en) | Method and apparatus for analyzing signal pattern of sensor array | |
| US20220390374A1 (en) | Method for extracting raman characteristic peaks employing improved principal component analysis | |
| CN108446656A (en) | A kind of parser carrying out Selective recognition to kitchen hazardous gas | |
| CN108573105A (en) | The method for building up of soil heavy metal content detection model based on depth confidence network | |
| CN113516228A (en) | Network anomaly detection method based on deep neural network | |
| CN109840548A (en) | One kind being based on BP neural network Diagnosis Method of Transformer Faults | |
| CN111259929A (en) | Random forest based food-borne pathogenic bacteria classification model training method | |
| CN110349621B (en) | Method, system, storage medium and device for checking reliability of peptide fragment-spectrogram matching | |
| CN114113471A (en) | Method and system for detecting food freshness of artificial nose refrigerator based on machine learning | |
| CN111929285A (en) | Spectral identification method for alcohol doped in laser-induced fluorescent red wine | |
| CN111209373A (en) | Sensitive text recognition method and device based on natural semantics | |
| CN112634945A (en) | Intelligent water leakage sound identification method based on cloud platform | |
| CN115620818A (en) | Protein mass spectrum peptide fragment verification method based on natural language processing | |
| CN114062305B (en) | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network | |
| CN115700557A (en) | Method, device and storage medium for classifying nucleic acid samples | |
| CN114141316A (en) | Method and system for predicting biological toxicity of organic matters based on spectrogram analysis | |
| Zhu et al. | Rapid freshness prediction of crab based on a portable electronic nose system | |
| CN114781484A (en) | Cancer serum SERS spectrum classification method based on convolutional neural network | |
| CN115143128A (en) | Fault diagnosis method and system for small submersible electric pump |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180824 |