CN106971073A - A kind of identification of nonlinearity method at water bursting in mine water source - Google Patents

A kind of identification of nonlinearity method at water bursting in mine water source Download PDF

Info

Publication number
CN106971073A
CN106971073A CN201710192720.6A CN201710192720A CN106971073A CN 106971073 A CN106971073 A CN 106971073A CN 201710192720 A CN201710192720 A CN 201710192720A CN 106971073 A CN106971073 A CN 106971073A
Authority
CN
China
Prior art keywords
water
sample
eml
matrix
water source
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
Application number
CN201710192720.6A
Other languages
Chinese (zh)
Inventor
周孟然
王亚
闫鹏程
何晨阳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Anhui University of Science and Technology
Original Assignee
Anhui University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Anhui University of Science and Technology filed Critical Anhui University of Science and Technology
Priority to CN201710192720.6A priority Critical patent/CN106971073A/en
Publication of CN106971073A publication Critical patent/CN106971073A/en
Pending legal-status Critical Current

Links

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
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
    • 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
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
    • G16Z99/00—Subject matter not provided for in other main groups of this subclass

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computational Linguistics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)

Abstract

The present invention relates to a kind of identification of nonlinearity method at water bursting in mine water source, comprise the following steps, (1) spectrum data gathering；(2) original spectral data of collection is pre-processed using convolution (SG) smoothing technique；(3) Data Dimensionality Reduction is carried out using Non-linear Kernel PCA, and carries out Nonlinear feature extraction；(4) by the test set and training set of the water sample Sample Establishing independence of collection；(5) EML extreme learning machines are set up using training set, testing classification result is then carried out by test set.The present invention carries out feature extraction using nonlinear way, Time ＆ Space Complexity is reduced, while there is provided the recognition performance of sample；Core principle component analysis optimizes the learning parameter of extreme learning machine, and EML disaggregated models have the features such as training parameter is few, pace of learning is fast, and KPCA and EML is combined into the nonlinear discrimination for water bursting in mine water source, is highly suitable for the monitoring of online water bursting sources.

Description

A kind of identification of nonlinearity method at water bursting in mine water source
Technical field
It is a kind of identification of nonlinearity method at water bursting in mine water source, and in particular to a kind of induced with laser based on KPCA-EML The nonlinear method at fluorescence spectrum (LIF) identification water bursting in mine water source.
Background technology
Mine water inrush causes the harm of seriousness to coal production, needs rapidly and accurately to differentiate gushing water when preventing and treating gushing water Water source.Water bursting in mine water source type mainly has：Ordovician karst water, goaf water, alluvial water, Sandstone Water, limestone water.At present, conventional coal The method of discrimination of ore deposit water bursting sources, is mostly based on water chemistry basis, but differentiates that the time is longer, efficiency comparison is low, it is difficult to adapt to The real-time monitoring of mine water inrush.The fluorescence light of water bursting sources is obtained using laser-induced fluorescence spectroscopy (LIF) technology real-time online Modal data, can be achieved fast and accurately to monitor.
Spectroscopic data collection is pre-processed and feature extraction is necessary operation before setting up classification learning model, mainly In order to ensure setting up accurate and stable disaggregated model.The use of extensive linear characteristic extracting method is mainly at present principal component point Analyse (PCA, Principal Component Analysis).Due to influenceing the factor of water bursting in mine numerous, with the non-of complexity Linear relationship, thus design a kind of sorting technique of non-linear water bursting sources be it is feasible, this method combination Non-linear Kernel it is main into Analyze (KPCA) and limits of application learning machine (EML) carries out classification learning.
Core principle component analysis (KPCA, Kernel Principal Component Analysis) is a kind of based on core letter Several principal component analytical methods, is mainly used to solve the problems, such as the feature extraction of large-scale nonlinear data, nonlinear data is reflected High-dimensional feature space is mapped to, and then seeks the linear solution in higher dimensional space.In the sorting technique based on supervised learning, BP god There is convergence rate through network slower, take it is longer, local optimum may be absorbed in, it is necessary to repeatedly training be only possible to reach it is global most Excellent deficiency.Support vector machines (Support Vector Machine) expend a large amount of machines there is also parameter to be optimized is more The defect such as device internal memory and operation time length.Because extreme learning machine (ELM, Extreme Learning Machine) is a kind of single Hidden layer feedforward neural network (SLFNs, Single-hidden Layer Feedforward Neural Networks) study is calculated Method, has the advantages that very strong learning ability and approaches complex nonlinear function.EML is mainly for based on gradient descent algorithm BP network trainings take excessively tediously long deficiency, and with asking, the Moor-Penrose generalized inverses substitution of hidden layer output matrix is traditional Iterative process, shortens the convergence time of network.
The content of the invention
The object of the invention is mainly to provide a kind of identification of nonlinearity method at water bursting in mine water source, in order to solve non-linear number According to feature extraction problem and accurate and Fast Learning classification problem, propose the laser-induced fluorescence spectroscopy skill based on KPCA-EML The nonlinear discrimination method at the water bursting in mine water source of art.
To achieve these goals, the technical solution adopted by the present invention is as follows：
A kind of identification of nonlinearity method at water bursting in mine water source, comprises the following steps：
(1), spectrum data gathering：The water sample spectrum number at water bursting in mine water source is gathered by laser-induced fluorescence spectroscopy instrument According to acquisition water sample primary light spectrogram；
(2), Pretreated spectra：The noise in spectrometer collection water sample fluorescence data is eliminated, it is smooth using SG convolution Method is pre-processed to the spectroscopic data collected；
(3), KPCA feature information extractions：Data are carried out to pretreated spectroscopic data using core principle component analysis KPCA The Nonlinear Principal Component Analysis of dimensionality reduction, obtains principal character value (i.e. principal component)；
(4), classification learning：The water sample sample that will be made up of principal character value, constructs independent training set and test set；
(5) EML extreme learning machines, are set up：EML extreme learning machines are set up on training set, then by test set Classification learning model carries out testing classification result.
In the present invention：In the step (3), using Non-linear Kernel principal component analysis KPCA on influenceing water in spectroscopic data Source is sentenced another characteristic and extracted, and it is comprised the following steps that：
(31) pretreated spectrum data matrix X, is taken, data space sample point (x is realizedi,xj) reflecting to feature space Penetrate, (xi,xj)→K(xi,xj)=φ (xi)·φ(xj), substitute the non-linear of complexity with relatively simple kernel function K () Transforming function transformation function φ ()；
(32), select Gaussian radial basis function (RBF) kernel function and calculate nuclear matrix K, correction matrix simultaneously calculates nuclear matrix Correlation matrix KL；
(33) characteristic value and corresponding characteristic vector of KL matrixes, are calculated with Jacobi alternative manners, by characteristic value by drop Sequence is arranged, and corresponding characteristic vector is also arranged in descending order；
(34), by Schmidt process unit orthogonalized eigenvectors, the accumulation contribution rate of characteristic value is calculated, with Principal component number is determined, EML input layer number is determined by principal component number.
In the present invention：Construction independent training set and test set in the step (4), it is comprised the following steps that：
(41), assume there is a N number of arbitrary sample, give training setD principal component is used as input Layer data, containing L hidden node, g (χ) is that excitation function is g (αi·χ+bi), it is basic function to choose RBF functions, i.e.,
(42) the output matrix H of hidden layer, is calculated, L hidden node, can be unlimited in the presence of excitation function g (χ) Close to N number of arbitrary sample, then H β=Τ,
Wherein H is the output matrix of hidden layer；
(43) output weight vector β, is calculated, EML disaggregated models are trained, the least square solution for calculating H β=Τ is converted into, That is β=H+Τ, wherein, H+For H Moore-Penrose generalized inverse matrix.
In the present invention：Described laser-induced fluorescence spectroscopy instrument for USB2000+ models, laser wavelength of incidence and work( Rate is respectively 405nm and 120mW, detects fluorescence spectra 400-800nm, resolution ratio 0.5nm.
By above-mentioned technical proposal, the beneficial effects of the invention are as follows：Feature extraction is carried out using nonlinear way, reduced Time ＆ Space Complexity, while there is provided the recognition performance of sample；Core principle component analysis optimizes the study of extreme learning machine Parameter, EML disaggregated models have the features such as training parameter is few, pace of learning is fast, and KPCA and EML, which is combined, is used for water bursting in mine water The nonlinear discrimination in source, is highly suitable for the monitoring of online water bursting sources.
Brief description of the drawings
Fig. 1 is flow chart schematic diagram of the invention；
Fig. 2 is the EML classification learning models in the present invention.
Embodiment
As shown in Figure 1-2：The present invention is combined with LIF technologies using KPCA-ELM, can intrusive mood fluorescent probe be placed in In water bursting in mine water source, in real time collection water bursting sources water sample fluorescence spectrum, in order to reduce collection in noise to fluorescence spectrum Influence, is pre-processed to spectrum, is analysed method progress feature extraction using Non-linear Kernel principal component, is set up independent training set and survey Examination collection, after being optimized to extreme learning machine EML learning parameters, by training set training generation classification learning model, by surveying Try set pair classification learning model and carry out test result.
The present invention is proposed to the non-of water bursting in mine water source in the laser-induced fluorescence spectroscopy LIF technologies based on KPCA-EML Linear Discriminant's method, comprises the following steps：
(1), spectrum data gathering：Using USB2000+ model laser-induced fluorescence spectroscopy instrument, laser wavelength of incidence and work( Rate is respectively 405nm and 120mW, detects fluorescence spectra 400-800nm, resolution ratio 0.5nm；Using ability of swimming laser can be invaded Fluorescent probe FPB-405-V3 is excited, is gathered using Spectrum software Spectra Suite and records the fluorescence spectrum of water sample, to 5 Multiple samples of type gushing water water sample (Ordovician karst water, goaf water, alluvial water, Sandstone Water, limestone water) carry out fluorescence spectrum Sampling, obtains water sample primary light spectrogram.
(2), Pretreated spectra：Spectrometer can also be collected into some noises while water sample fluorescence data is gathered, In order that disaggregated model has stronger predictive ability and preferable robustness, spectrometer collection water sample fluorescence data is eliminated In noise, the spectroscopic data collected is pre-processed using SG convolution smoothing method.
(3), KPCA feature information extractions：Data are carried out to pretreated spectroscopic data using core principle component analysis KPCA The Nonlinear Principal Component Analysis of dimensionality reduction, obtains principal character value (i.e. principal component), comprises the following steps：
(31) pretreated spectrum data matrix X, is taken, data space sample point (x is realizedi,xj) reflecting to feature space Penetrate, (xi,xj)→K(xi,xj)=φ (xi)·φ(xj), substitute the non-linear of complexity with relatively simple kernel function K () Transforming function transformation function φ ()；
Assuming that there is a N number of index, each index has M sample, and the input space is mapped to certain by kernel function by mapping phi Data meet centralization condition in the feature space of individual higher-dimension, feature space, i.e.,The then association in feature space Variance matrix isC eigenvalue λ >=0 and characteristic vector μ is sought, i.e. (C μ=λ μ-＞ φ (χμ) C μ= λφ(χμ) μ) then：Define M × M dimension matrixes
Kμν=K (χμ,χν)=(φ (χμ)·φ(χν)) (formula 1)
(32) Gaussian radial basis function (RBF) kernel function, is selectedBy formula (1) nuclear matrix K is calculated, is passed throughCorrection matrix, and calculate nuclear matrix Correlation matrix
(33) eigenvalue λ of KL matrixes, is calculated with Jacobi alternative manners1,…,λnWith corresponding characteristic vector μ1,…, μn, characteristic value is arranged in descending order, corresponding characteristic vector is also arranged in descending order；
(34), by Schmidt process unit orthogonalized eigenvectors, α is obtained1,…,αn, calculate characteristic value Accumulation contribution rate B1,…,Bn, according to given extraction efficiency p, if Bd>=p, then extract d principal component α1,…,αd；By principal component Number determines that EML input layer number is d.
(4), classification learning：The water sample sample that will be made up of principal character value, constructs independent training set and test set, bag Include following steps：
(41), assume there is a N number of arbitrary sample, give training setD principal component is used as the input number of plies According to can be expressed as containing L hidden node point SLFNs Wherein, αiIt is to connect i-th of hidden layer node and the weights of input node, biIt is the deviation of i-th of hidden layer node, βiIt is connection i-th The weights of individual hidden layer node and output node；Random generation hidden layer node parameter (αi,bi), i=1 ..., Lg (χ) is excitation function For g (αi·χ+bi), it is basic function to choose RBF functions, i.e.,
(42) the output matrix H of hidden layer, is calculated, L hidden node, can be unlimited in the presence of excitation function g (χ) Close to N number of arbitrary sample, then H β=Τ,
Wherein H is the output matrix of hidden layer；
(43) output weight vector β, is calculated, EML disaggregated models are trained, the least square solution for calculating H β=Τ is converted into, That is β=H+Τ, wherein, H+For H Moore-Penrose generalized inverse matrix
(5) EML extreme learning machines, are set up：EML extreme learning machines are set up on training set, then by test set Classification learning model carries out testing classification result.
The embodiment to the present invention is described above, but the present invention is not limited to above description.For this For the technical staff in field, any equal modifications and substitutions to the technical program are all within the scope of the invention.Cause This, the impartial conversion and modification made without departing from the spirit and scope of the invention all should be contained within the scope of the invention.

Claims (4)

1. a kind of identification of nonlinearity method at water bursting in mine water source, it is characterised in that：Comprise the following steps：
(1), spectrum data gathering：The water sample spectroscopic data at water bursting in mine water source is gathered by laser-induced fluorescence spectroscopy instrument, is obtained Obtain water sample primary light spectrogram；
(2), Pretreated spectra：The noise in spectrometer collection water sample fluorescence data is eliminated, using SG convolution smoothing methods The spectroscopic data collected is pre-processed；
(3), KPCA feature information extractions：Data Dimensionality Reduction is carried out to pretreated spectroscopic data using core principle component analysis KPCA Nonlinear Principal Component Analysis, obtain principal character value (i.e. principal component)；
(4), classification learning：The water sample sample that will be made up of principal character value, constructs independent training set and test set；
(5) EML extreme learning machines, are set up：EML extreme learning machines are set up on training set, then pass through the classification on test set Learning model carries out testing classification result.
2. a kind of identification of nonlinearity method at water bursting in mine water source according to claim 1, its feature is：The step (3) in, extracted using Non-linear Kernel principal component analysis KPCA on influenceing water source to sentence another characteristic in spectroscopic data, its is specific Step is as follows：
(31) pretreated spectrum data matrix X, is taken, data space sample point (x is realizedi,xj) to the mapping of feature space, (xi,xj)→K(xi,xj)=φ (xi)·φ(xj), substitute complicated non-linear change with relatively simple kernel function K () Change function phi ()；
(32), select Gaussian radial basis function (RBF) kernel function and calculate nuclear matrix K, correction matrix and the correlation for calculating nuclear matrix Matrix K L；
(33) characteristic value and corresponding characteristic vector of KL matrixes, are calculated with Jacobi alternative manners, characteristic value is arranged in descending order Row, corresponding characteristic vector is also arranged in descending order；
(34), by Schmidt process unit orthogonalized eigenvectors, the accumulation contribution rate of characteristic value is calculated, to determine Principal component number, EML input layer number is determined by principal component number.
3. a kind of identification of nonlinearity method at water bursting in mine water source according to claim 1, its feature is：The step (4) construction independent training set and test set in, it is comprised the following steps that：
(41), assume there is a N number of arbitrary sample, give training setD principal component is used as the input number of plies According to containing L hidden node, g (χ) is that excitation function is g (αi·χ+bi), it is basic function to choose RBF functions, i.e.,
(42) the output matrix H of hidden layer, is calculated, L hidden node, can be with infinite approach in the presence of excitation function g (χ) N number of arbitrary sample, then H β=Τ,
H = g ( α 1 · x 1 + b 1 ) ... g ( α L · x 1 + b L ) ... g ( α 1 · x N + b 1 ) ... g ( α L · x N + b L ) N × L β = β 1 T ... β L T L × M , T = t 1 T ... t N T N × M ,
Wherein H is the output matrix of hidden layer；
(43), calculate output weight vector β, train EML disaggregated models, be converted into the least square solution for calculating H β=Τ, i.e. β= H+Τ, wherein, H+For H Moore-Penrose generalized inverse matrix.
4. a kind of identification of nonlinearity method at water bursting in mine water source according to claim 1, its feature is：Described swashs Light induced fluorescence spectrometer for USB2000+ models, laser wavelength of incidence and power are respectively 405nm and 120mW, are detected glimmering Light spectral region 400-800nm, resolution ratio 0.5nm.
CN201710192720.6A 2017-03-28 2017-03-28 A kind of identification of nonlinearity method at water bursting in mine water source Pending CN106971073A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710192720.6A CN106971073A (en) 2017-03-28 2017-03-28 A kind of identification of nonlinearity method at water bursting in mine water source

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710192720.6A CN106971073A (en) 2017-03-28 2017-03-28 A kind of identification of nonlinearity method at water bursting in mine water source

Publications (1)

Publication Number Publication Date
CN106971073A true CN106971073A (en) 2017-07-21

Family

ID=59335701

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710192720.6A Pending CN106971073A (en) 2017-03-28 2017-03-28 A kind of identification of nonlinearity method at water bursting in mine water source

Country Status (1)

Country Link
CN (1) CN106971073A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108509993A (en) * 2018-04-02 2018-09-07 安徽理工大学 A kind of water bursting in mine laser-induced fluorescence spectroscopy image-recognizing method
CN109308498A (en) * 2018-11-28 2019-02-05 安徽理工大学 A kind of small thin rice gruel discrimination method of laser induced fluorescence vegetable oil doping
CN109388944A (en) * 2018-11-06 2019-02-26 吉林大学 A kind of intrusion detection method based on KPCA and ELM
CN111562285A (en) * 2020-06-03 2020-08-21 安徽大学 Mine water inrush source identification method and system based on big data and deep learning
CN117367751A (en) * 2023-10-19 2024-01-09 中聚科技股份有限公司 Performance detection method and device for ultra-pulse thulium-doped laser

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103211590A (en) * 2013-04-24 2013-07-24 浙江理工大学 Heart electrical function quick imaging method of accelerated extreme learning machine based on GPU
CN103745093A (en) * 2013-12-25 2014-04-23 中国矿业大学 Principal component analysis-extreme learning machine (PCA-ELM) based coal mine water inrush prediction method
CN104156560A (en) * 2014-07-12 2014-11-19 中国矿业大学 Multi-level coal mine water inrush prediction method based on SaE-ELM (self-adaptive evolutionary extreme learning machine)
CN104749134A (en) * 2015-03-31 2015-07-01 江苏大学 Method for detecting canopy moisture content of leaf vegetable crops
CN108509993A (en) * 2018-04-02 2018-09-07 安徽理工大学 A kind of water bursting in mine laser-induced fluorescence spectroscopy image-recognizing method

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103211590A (en) * 2013-04-24 2013-07-24 浙江理工大学 Heart electrical function quick imaging method of accelerated extreme learning machine based on GPU
CN103745093A (en) * 2013-12-25 2014-04-23 中国矿业大学 Principal component analysis-extreme learning machine (PCA-ELM) based coal mine water inrush prediction method
CN104156560A (en) * 2014-07-12 2014-11-19 中国矿业大学 Multi-level coal mine water inrush prediction method based on SaE-ELM (self-adaptive evolutionary extreme learning machine)
CN104749134A (en) * 2015-03-31 2015-07-01 江苏大学 Method for detecting canopy moisture content of leaf vegetable crops
CN108509993A (en) * 2018-04-02 2018-09-07 安徽理工大学 A kind of water bursting in mine laser-induced fluorescence spectroscopy image-recognizing method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
李培 等: "基于PCA-ELM的煤矿突水预测方法研究", 《工矿自动化》 *
李怀俊: "《基于核主元模糊聚类的旋转机械故障诊断技术研究》", 31 July 2016, 西南交通大学出版社 *
王亚 等: "PCA-BP模型在判别基于LIF技术煤矿突水水源的应用", 《光谱学与光谱分析》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108509993A (en) * 2018-04-02 2018-09-07 安徽理工大学 A kind of water bursting in mine laser-induced fluorescence spectroscopy image-recognizing method
CN109388944A (en) * 2018-11-06 2019-02-26 吉林大学 A kind of intrusion detection method based on KPCA and ELM
CN109308498A (en) * 2018-11-28 2019-02-05 安徽理工大学 A kind of small thin rice gruel discrimination method of laser induced fluorescence vegetable oil doping
CN109308498B (en) * 2018-11-28 2021-01-01 安徽理工大学 Identification method of laser-induced fluorescent vegetable oil-doped millet soup
CN111562285A (en) * 2020-06-03 2020-08-21 安徽大学 Mine water inrush source identification method and system based on big data and deep learning
CN117367751A (en) * 2023-10-19 2024-01-09 中聚科技股份有限公司 Performance detection method and device for ultra-pulse thulium-doped laser
CN117367751B (en) * 2023-10-19 2024-05-10 中聚科技股份有限公司 Performance detection method and device for ultra-pulse thulium-doped laser

Similar Documents

Publication Publication Date Title
CN106971073A (en) A kind of identification of nonlinearity method at water bursting in mine water source
CN112101381B (en) Tensor collaborative drawing discriminant analysis remote sensing image feature extraction method
Casas et al. Linear and non-linear Modified Gravity forecasts with future surveys
CN109145992A (en) Cooperation generates confrontation network and sky composes united hyperspectral image classification method
Zhang et al. Classification modeling method for near‐infrared spectroscopy of tobacco based on multimodal convolution neural networks
CN109493287A (en) A kind of quantitative spectra data analysis processing method based on deep learning
CN109000930A (en) A kind of turbogenerator performance degradation assessment method based on stacking denoising self-encoding encoder
CN107044976A (en) Heavy metal content in soil analyzing and predicting method based on LIBS Yu stack RBM depth learning technologies
CN111754028A (en) Hyperspectrum-based coal ash content and moisture detection system and method
CN112712108B (en) Raman spectrum multivariate data analysis method
CN108985335A (en) The integrated study prediction technique of nuclear reactor cladding materials void swelling
CN107451760B (en) Rolling bearing fault diagnosis method based on time window slip limited Boltzmann machine
CN114399685B (en) Remote sensing monitoring and evaluating method and device for forest pest and disease damage
CN109669023A (en) A kind of soil attribute prediction technique based on Multi-sensor Fusion
CN104483292B (en) A kind of method that use multiline ratio method improves laser microprobe analysis accuracy
CN107679569A (en) Raman spectrum substance automatic identifying method based on adaptive hypergraph algorithm
CN104965973B (en) A kind of Apple Mould Core multiple-factor Non-Destructive Testing discrimination model and method for building up thereof
CN112948932A (en) Surrounding rock grade prediction method based on TSP forecast data and XGboost algorithm
CN106841053A (en) A kind of jujube kind rapid identification method
CN110111885A (en) Attribute forecast method, apparatus, computer equipment and computer readable storage medium
CN105334186A (en) Infrared spectral analysis method
CN108573105A (en) The method for building up of soil heavy metal content detection model based on depth confidence network
CN103278467A (en) Rapid nondestructive high-accuracy method with for identifying abundance degree of nitrogen element in plant leaf
CN109145685A (en) Fruits and vegetables EO-1 hyperion quality detecting method based on integrated study
CN112232343B (en) Grain mildew grain identification neural network and identification method

Legal Events

Date Code Title Description
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: 20170721