CN113537334A - Gas-liquid multi-component real-time intelligent detection method - Google Patents
Gas-liquid multi-component real-time intelligent detection method Download PDFInfo
- Publication number
- CN113537334A CN113537334A CN202110780829.8A CN202110780829A CN113537334A CN 113537334 A CN113537334 A CN 113537334A CN 202110780829 A CN202110780829 A CN 202110780829A CN 113537334 A CN113537334 A CN 113537334A
- Authority
- CN
- China
- Prior art keywords
- neural network
- gas
- sample
- matrix
- intelligent detection
- 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
- 239000007788 liquid Substances 0.000 title claims abstract description 32
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 53
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims abstract description 20
- 230000009467 reduction Effects 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000000034 method Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 230000008878 coupling Effects 0.000 abstract description 3
- 238000010168 coupling process Methods 0.000 abstract description 3
- 238000005859 coupling reaction Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 17
- 238000000513 principal component analysis Methods 0.000 description 12
- 210000002569 neuron Anatomy 0.000 description 11
- 238000004458 analytical method Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Chemical & Material Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
Abstract
The invention provides a gas-liquid multi-component real-time intelligent detection method, which comprises the following steps: carrying out dimensionality reduction on a sample to be identified to obtain k-dimensional data; processing the k-dimensional data by using a preset artificial neural network to finally obtain the composition components in the sample to be identified and the corresponding concentrations of the composition components; the invention combines the PCA recognition algorithm and the learning algorithm of the artificial neural network to solve the problems of weak signals, mutual coupling and strong interference of various detected components in the prior art.
Description
Technical Field
The invention belongs to the technical field of component detection, and particularly relates to a gas-liquid multi-component real-time intelligent detection method.
Background
The existing gas-liquid multi-component detection technology can realize 3-5 real-time detections only under an off-line condition, and the difficulty is mainly that signals among various detected components are weak, coupled and interfered with each other strongly, so that indexes such as acquisition, processing, control, comparison and classification of various signals hardly meet practical requirements, and the difficulty is a technical bottleneck in the world detection field at present. Corresponding on-line detection products are available in Japan, Germany and Israel, the number of detection components is not more than 5, the price is high, and the method is only used in experimental research industry and is difficult to popularize and apply.
Disclosure of Invention
The invention aims to provide a gas-liquid multi-component real-time intelligent detection method, which solves the defects of low efficiency and high cost of the existing gas-liquid multi-component detection technology.
In order to achieve the purpose, the invention adopts the technical scheme that:
the invention provides a gas-liquid multi-component real-time intelligent detection method, which comprises the following steps:
carrying out dimensionality reduction on a sample to be identified to obtain k-dimensional data;
and processing the k-dimensional data by using a preset artificial neural network to finally obtain the composition components in the sample to be identified and the corresponding concentrations of the composition components.
Preferably, the PCA recognition algorithm is used for carrying out dimensionality reduction on the sample to be recognized to obtain k-dimensional data.
Preferably, the dimension reduction processing is performed on the sample to be identified to obtain k-dimension data, and the specific method is as follows:
obtaining a sample matrix according to a sample structure to be identified, and carrying out zero-mean processing on each row of data of the sample matrix to obtain a standardized matrix;
obtaining a covariance matrix according to the standardized matrix;
respectively calculating an eigenvalue and an eigenvector of the covariance matrix;
and arranging the eigenvectors according to the size of the eigenvalue from top to bottom according to rows, and selecting front k rows of data to form a matrix to obtain k-dimensional data.
Preferably, the covariance matrix is obtained from the normalized matrix by the following specific method:
the covariance matrix is obtained using the following equation:wherein C is a covariance matrix; x is a standardized matrix; xTIs the transpose of the normalized matrix X.
Preferably, the preset artificial neural network is constructed by the following method:
and constructing an artificial neural network, and optimizing the constructed artificial neural network by using a back propagation algorithm to obtain the optimized artificial neural network.
Preferably, the constructed artificial neural network is optimized by using a back propagation algorithm, and the specific method comprises the following steps:
selecting a square sum error function, and initializing weight parameters of each layer in the neural network to obtain an error function of the neural network;
using a back propagation algorithm, calculating in a forward direction to obtain an error function, and reducing a back derivative gradient to minimize the error function;
and solving partial derivatives of the error function about the self variables of the neural network by using a gradient descent method, updating weight parameters of each layer in the neural network until the error function is reduced to a preset range, and finishing the optimization of the artificial neural network.
A gas-liquid multi-component real-time intelligent detection system can operate the gas-liquid multi-component real-time intelligent detection method, and comprises a data processing module and a component identification module, wherein:
the data processing module is used for carrying out dimensionality reduction on the sample to be identified to obtain k-dimensional data;
and the component identification module is used for processing the k-dimensional data by utilizing a preset artificial neural network to finally obtain the component in the sample to be identified and the corresponding concentration of the component.
A gas-liquid multi-component real-time intelligent detection device comprising a processor and a computer program capable of running on said processor, characterized in that said processor implements the steps of said method when executing said computer program.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a gas-liquid multi-component real-time intelligent detection method, which utilizes PCA to convert original data into a group of linearly independent expressions with each dimension through linear transformation, thereby extracting main characteristic components of the data, inputs different material templates and corresponding identification results into an artificial neural network through the self-learning function of the artificial neural network, and identifies similar materials through self-learning by an algorithm; meanwhile, the artificial neural network is utilized to exert the high-speed computing capability of the computer and quickly obtain an optimized solution; the invention processes the detected data by applying the PCA algorithm, provides the main characteristics for analysis, removes interference, solves the problems of weak signals of all components and mutual coupling, improves the prediction precision and enhances the robustness and fault tolerance by combining with the neural network.
Drawings
Fig. 1 is a diagram of an artificial neural network.
Detailed Description
The present invention is described in further detail below.
The invention provides a gas-liquid multi-component real-time intelligent detection method, which comprises the following steps:
carrying out dimensionality reduction on a sample to be identified to obtain k-dimensional data;
extracting characteristic components from the obtained k-dimensional data;
and processing the extracted characteristic components by using the constructed artificial neural network to finally obtain the composition components and the concentration thereof in the collected sample for recognition and test.
Specifically, the method comprises the following steps:
carrying out dimensionality reduction on a sample to be identified to obtain k-dimensional data, wherein the method specifically comprises the following steps:
carrying out dimensionality reduction on the n-dimensional original data by utilizing a Principal Component Analysis (PCA) recognition algorithm to obtain k-dimensional data; wherein:
wherein C is a covariance matrix; x is a standardized matrix; xTIs the transpose of the normalized matrix X.
step 4, arranging the eigenvectors according to the size of the eigenvalue from top to bottom according to rows, and selecting the first k rows of data to form a matrix P;
and 5, reducing the dimension of the obtained matrix P according to the following formula to obtain a matrix after dimension reduction:
Y=PX
wherein, Y is a matrix after dimensionality reduction, namely a matrix formed by characteristic components after dimensionality reduction to k dimensions.
The k-dimension data is processed by utilizing the constructed artificial neural network, and the specific method comprises the following steps:
the artificial neural network is divided into three layers, namely an input layer, a hidden layer and an output layer, wherein each neuron of the Nth layer is connected with all neurons of the N-1 th layer, and the output of the neuron of the N-1 th layer is the input of the neuron of the Nth layer.
The neural network algorithm can provide a complex and nonlinear hypothesis model hW,b(x) The data can be fitted with this, where if a neuron is one with x1,x2,x3An input arithmetic unit having outputs of:
where f (x) is the activation function.
When the neural network is trained, a back propagation algorithm is used, an error function is obtained through forward calculation, and the reverse derivative gradient is reduced, so that the defined error function is minimized.
Selecting a sum of squares error function to initialize weight parameters (w, b) of each layer in the neural network0,θ,b1) The error function of the neural network is obtained as shown in formula (4):
the self-variables of the neural network are then updated such that the error function of the neural network is reduced under the updated self-variables. Using a gradient descent method, the error function is determined with respect to the neural network variables (w, b)0,θ,b1) Then the updating of the variables is performed by equation (5).
Where γ is called a learning rate, and 0.1 is taken as a representative of the learning rate and quality of the neural network.
And continuously repeating the steps to approximate the relation between the data acquired by the real detector unit and the category and the concentration of the components in the sample to be detected until the error function is reduced to an acceptable range, so that the accuracy of the method is improved, and the learning of the training data by the neural network is completed.
The neural network structure is shown in the following diagram, wherein,
pi(i ═ 1,2,1.., R) is the i-th neuron of the input layer after PCA algorithm processing;
lwi,j(i ═ 1,2, 3; j ═ 1,2,3) is the weight of the ith neuron in the input layer to the jth neuron in the hidden layer;
fian activation function for the ith layer;
a1=f1(LW1,1p+b1);a2=f2(LW2,1a1+b2);a3=f3(LW3,2a2+b3);
a3=f3(LW3,2f2(LW2,1f1(LW1,1p+b1)+b2)+b3)。
the method comprises the steps of processing gas-liquid sample data with known components through PCA, inputting the processed gas-liquid component data into an artificial neural network for training to obtain a first round of analysis results, calculating the error between the obtained result and known information, and judging whether the obtained error is within a preset range; if so, taking the obtained first round analysis result as a final result; otherwise, adjusting the number of layers of the hidden layer in the artificial neural network model, and training again to obtain a final result; and accumulating the norm of the final result, squaring to obtain a weighted value of the known gas-liquid component, adding the weighted value into the training process of the artificial neural network model, retraining the gas-liquid component to obtain the characteristic parameter of the known gas-liquid component, and storing the characteristic parameter.
The gas-liquid multi-component real-time intelligent detection device can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The gas-liquid multi-component real-time intelligent detection device may include, but is not limited to, a processor, a memory … ….
The processor may be a Central Processing Unit (CPU), other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. … … are provided.
The working principle of the invention is as follows:
and after the data of the sample to be detected is subjected to characteristic processing through PCA, inputting the processed data into a final gas-liquid component prediction model to obtain the content of each component of the sample to be detected. According to the method, data received by a sensor is calculated into a covariance matrix through a PCA technology, a characteristic vector matrix corresponding to k characteristics with the largest characteristic value is selected, characteristic dimension reduction of gas-liquid components is achieved, related influences among data characteristics are eliminated, workload of index selection is reduced, most information is kept, meanwhile, a small number of comprehensive indexes are used for replacing original indexes for analysis, an artificial neural network algorithm is combined, and universality and accuracy of gas-liquid component measurement results are improved.
The invention has the following beneficial effects:
the invention uses the recognition algorithm of Principal Component Analysis (PCA) and the learning algorithm of Artificial Neural Network (ANN) to solve the problems of weak signals, mutual coupling and strong interference of various detected components in the prior art. PCA transforms raw data into a set of representations linearly independent of each dimension through linear transformation, thereby extracting principal feature components of the data. Different material templates and corresponding recognition results are input into the artificial neural network through the self-learning function of the artificial neural network, and similar materials are recognized through self-learning by the algorithm. Meanwhile, the artificial neural network is utilized to exert the high-speed computing capability of the computer and quickly obtain an optimized solution.
Claims (8)
1. A gas-liquid multi-component real-time intelligent detection method is characterized by comprising the following steps:
carrying out dimensionality reduction on a sample to be identified to obtain k-dimensional data;
and processing the k-dimensional data by using a preset artificial neural network to finally obtain the composition components in the sample to be identified and the corresponding concentrations of the composition components.
2. The gas-liquid multi-component real-time intelligent detection method according to claim 1, characterized in that a PCA recognition algorithm is used to perform dimensionality reduction on a sample to be recognized to obtain k-dimensional data.
3. The gas-liquid multi-component real-time intelligent detection method according to claim 2, wherein the dimension reduction processing is performed on the sample to be identified to obtain k-dimensional data, and the specific method comprises the following steps:
obtaining a sample matrix according to a sample structure to be identified, and carrying out zero-mean processing on each row of data of the sample matrix to obtain a standardized matrix;
obtaining a covariance matrix according to the standardized matrix;
respectively calculating an eigenvalue and an eigenvector of the covariance matrix;
and arranging the eigenvectors according to the size of the eigenvalue from top to bottom according to rows, and selecting front k rows of data to form a matrix to obtain k-dimensional data.
4. The gas-liquid multicomponent real-time intelligent detection method according to claim 3, wherein the covariance matrix is obtained from the normalized matrix by:
5. The gas-liquid multi-component real-time intelligent detection method according to claim 1, wherein the preset artificial neural network is constructed by the following steps:
and constructing an artificial neural network, and optimizing the constructed artificial neural network by using a back propagation algorithm to obtain the optimized artificial neural network.
6. The gas-liquid multi-component real-time intelligent detection method according to claim 5, wherein the constructed artificial neural network is optimized by using a back propagation algorithm, and the specific method comprises the following steps:
selecting a square sum error function, and initializing weight parameters of each layer in the neural network to obtain an error function of the neural network;
using a back propagation algorithm, calculating in a forward direction to obtain an error function, and reducing a back derivative gradient to minimize the error function;
and solving partial derivatives of the error function about the self variables of the neural network by using a gradient descent method, updating weight parameters of each layer in the neural network until the error function is reduced to a preset range, and finishing the optimization of the artificial neural network.
7. A gas-liquid multi-component real-time intelligent detection system, which is capable of operating a gas-liquid multi-component real-time intelligent detection method according to any one of claims 1 to 6, and comprises a data processing module and a component identification module, wherein:
the data processing module is used for carrying out dimensionality reduction on the sample to be identified to obtain k-dimensional data;
and the component identification module is used for processing the k-dimensional data by utilizing a preset artificial neural network to finally obtain the component in the sample to be identified and the corresponding concentration of the component.
8. A gas-liquid multi-component real-time intelligent detection device comprising a processor and a computer program capable of running on said processor, characterized in that said processor, when executing said computer program, carries out the steps of the method according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110780829.8A CN113537334A (en) | 2021-07-09 | 2021-07-09 | Gas-liquid multi-component real-time intelligent detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110780829.8A CN113537334A (en) | 2021-07-09 | 2021-07-09 | Gas-liquid multi-component real-time intelligent detection method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113537334A true CN113537334A (en) | 2021-10-22 |
Family
ID=78098345
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110780829.8A Pending CN113537334A (en) | 2021-07-09 | 2021-07-09 | Gas-liquid multi-component real-time intelligent detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113537334A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101933809A (en) * | 2010-08-31 | 2011-01-05 | 天津大学 | Multiband reflection spectrum noninvasive blood component measuring device and method |
CN110414169A (en) * | 2019-08-05 | 2019-11-05 | 上海神开石油科技有限公司 | A kind of fourier infrared gas detection logging method and device thereof |
CN110426421A (en) * | 2019-09-09 | 2019-11-08 | 浙江大学 | Multi-component harmful gas detection device and detection method in a kind of kitchen environment |
CN111754028A (en) * | 2020-06-08 | 2020-10-09 | 吉林大学 | Hyperspectrum-based coal ash content and moisture detection system and method |
-
2021
- 2021-07-09 CN CN202110780829.8A patent/CN113537334A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101933809A (en) * | 2010-08-31 | 2011-01-05 | 天津大学 | Multiband reflection spectrum noninvasive blood component measuring device and method |
CN110414169A (en) * | 2019-08-05 | 2019-11-05 | 上海神开石油科技有限公司 | A kind of fourier infrared gas detection logging method and device thereof |
CN110426421A (en) * | 2019-09-09 | 2019-11-08 | 浙江大学 | Multi-component harmful gas detection device and detection method in a kind of kitchen environment |
CN111754028A (en) * | 2020-06-08 | 2020-10-09 | 吉林大学 | Hyperspectrum-based coal ash content and moisture detection system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109060001B (en) | Multi-working-condition process soft measurement modeling method based on feature transfer learning | |
CN110738984A (en) | Artificial intelligence CNN, LSTM neural network speech recognition system | |
CN111126575A (en) | Gas sensor array mixed gas detection method and device based on machine learning | |
CN110580268A (en) | Credit scoring integrated classification system and method based on deep learning | |
CN110334881A (en) | A kind of Financial Time Series Forecasting method based on length memory network and depth data cleaning, device and server | |
CN113033309A (en) | Fault diagnosis method based on signal downsampling and one-dimensional convolution neural network | |
CN112732921B (en) | False user comment detection method and system | |
Guo et al. | Research on optimization of static gesture recognition based on convolution neural network | |
CN113705715B (en) | Time sequence classification method based on LSTM and multi-scale FCN | |
Pang et al. | Convolutional neural network-based sub-pixel line-edged angle detection with applications in measurement | |
Liu et al. | High Similarity Image Recognition and Classification Algorithm Based on Convolutional Neural Network | |
CN108460412B (en) | Image classification method based on subspace joint sparse low-rank structure learning | |
CN108388918B (en) | Data feature selection method with structure retention characteristics | |
CN106599391B (en) | Association vector machine soft measurement modeling method based on dynamic weighting of triangle angle values | |
CN113537334A (en) | Gas-liquid multi-component real-time intelligent detection method | |
CN115206455B (en) | Deep neural network-based rare earth element component content prediction method and system | |
Sharma et al. | An Integrated Approach towards Efficient Image Classification Using Deep CNN with Transfer Learning and PCA | |
Li et al. | A staked discriminative auto-encoder based on center loss for radar target HRRP recognition | |
Huang et al. | Deep learning with tactile sequences enables fruit recognition and force prediction for damage-free grasping | |
CN112801172A (en) | Chinese cabbage pesticide residue qualitative analysis method based on fuzzy pattern recognition | |
Liang et al. | Facial expression recognition based on gabor multi-orientation feature fusion | |
Wang | An Expression Recognition Method based on Improved Convolutional Network | |
Wang et al. | Compare CNN and LSTM Model to Forecast the Stock Price | |
Yang et al. | Identifying cherry maturity and disease using different fusions of deep features and classifiers | |
Sapkal et al. | Analysis of classification by supervised and unsupervised learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211022 |