CN116933643A - Intelligent data monitoring method based on partial robust M regression and multiple interpolation - Google Patents
Intelligent data monitoring method based on partial robust M regression and multiple interpolation Download PDFInfo
- Publication number
- CN116933643A CN116933643A CN202310925203.0A CN202310925203A CN116933643A CN 116933643 A CN116933643 A CN 116933643A CN 202310925203 A CN202310925203 A CN 202310925203A CN 116933643 A CN116933643 A CN 116933643A
- Authority
- CN
- China
- Prior art keywords
- data
- obs
- method based
- quality
- support vector
- 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
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000012544 monitoring process Methods 0.000 title claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 11
- 230000002159 abnormal effect Effects 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 238000007405 data analysis Methods 0.000 claims abstract description 3
- 230000008569 process Effects 0.000 claims description 30
- 238000001514 detection method Methods 0.000 claims description 9
- 238000004519 manufacturing process Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000000342 Monte Carlo simulation Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 230000008030 elimination Effects 0.000 abstract 1
- 238000003379 elimination reaction Methods 0.000 abstract 1
- 230000004913 activation Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- 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)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Geometry (AREA)
- Databases & Information Systems (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses an intelligent data monitoring method based on partial robust M regression and multiple interpolation, which comprises an online monitoring method based on information multiple interpolation and variation information bottlenecks and an offline modeling method based on variation information bottlenecks and depth support vector data description, wherein the online monitoring method based on information multiple interpolation and variation information bottlenecks comprises data acquisition and preprocessing, data analysis and abnormal value missing data elimination filling, and the offline modeling method based on variation information bottlenecks and depth support vector data description comprises data acquisition and preprocessing, variation bottleneck processing and depth support vector processing. The invention overcomes the influence of abnormal values on the model by using a partial robust M regression method, and realizes effective extraction of output related features by parameterizing the model by using a deep learning and variation theory based on Bayesian reasoning.
Description
Technical Field
The invention relates to the field of process monitoring and fault detection, in particular to an intelligent data monitoring method based on partial robust M regression and multiple interpolation.
Background
Process monitoring (Statistical Process Monitoring, PM) is a quality control method that enables real-time monitoring and improvement of industrial processes by analyzing and interpreting data collected during the production process. Generally, PM consists of three phases: 1) Offline modeling, namely establishing a multivariate statistical analysis model by utilizing historical data under normal process conditions; 2) And monitoring the process on line, and detecting faults by using the model. The operation working condition and the working environment become more complex, and the failure probability is gradually increased
The operation working condition and the working environment become more complex, and the failure probability is gradually increased. In addition, the existence of the missing value in the normal training data can lead to inaccurate monitoring models, reduce the fault detection rate and increase the false alarm rate. Model-based process monitoring methods appear to be ineffective in the face of faults in modern industrial processes. Compared with the PM method based on the model, the PM method based on data driving does not depend on process mechanism knowledge, and has wide application prospect in actual PM production.
Therefore, the invention provides a process monitoring and fault detection method for filling missing data based on partial robust M regression and multiple interpolation, which aims to effectively solve the problem of data missing in an online monitoring stage.
Disclosure of Invention
The invention aims to provide an intelligent data monitoring method based on partial robust M regression and multiple interpolation so as to solve the problems in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions: the intelligent data monitoring method based on the partial robust M regression and the multiple interpolation comprises an online monitoring method based on the information multiple interpolation and variation information bottleneck and an offline modeling method based on the variation information bottleneck and depth support vector data description, wherein the online monitoring method based on the information multiple interpolation and variation information bottleneck comprises the following steps:
data acquisition and preprocessing, namely arranging sensors at reasonable positions of equipment to be detected aiming at application scenes of process monitoring and fault detection under different data loss conditions so as to collect data; dividing the length of a data sample according to the sampling frequency of the sensor and the collected data quantity; performing fault type labeling on each original data sample to construct a training data set;
data analysis, namely dividing the acquired data into process data X= [ X ] according to specific application scenes 1 ,x 2 ,…,x n ] T And quality data y= [ Y ] 1 ,y 2 ,…,y n ] T If Y is a single quality variable, abnormal values in X and Y can be removed directly by using partial robust M regression; for a plurality of quality variables, converting the quality variables into single quality variables by a weighting method, and then processing the single quality variables by using a partial robust M regression;
reject outliers: normalizing the process data X and the quality data Y; by calculating bad lever weightsHigh residual weight->And general weight w i Initializing a weight matrix W; calculating weighted process data X W =wx and weighted quality data Y W WY, where w=diag (W);
filling in missing data: based on bayesian reasoning, a number of possible interpolations are generated taking into account the uncertainty of the missing data, thus forming a number of complete data sets. Then, carrying out the same processing on each data set, obtaining an estimated value through comprehensive analysis, and completing statistical inference;
the offline modeling method based on the variation information bottleneck and the depth support vector data description comprises the following steps:
acquiring and preprocessing data, acquiring sensor historical data, and dividing preprocessed data X' into: x' = [ X 1 ',X 2 ',…,X b ',…,X B ']Where B is the number of process variables and B is the number of blocks, for each block, the corresponding process data X is used b 'and quality data Y' to train a variation information bottleneck model;
variable bottleneck processing, useExtracting quality related characteristics from the variation information bottleneck: on the basis of an information bottleneck criterion, parameterizing the model by utilizing a deep learning and variation theory; in order to realize effective output related feature extraction under the condition of data missing, an encoder q is designed in a variation information bottleneck model φ (v|x) and decoder p θ (y|x), wherein phi and theta are the variation parameter and the generation parameter, respectively; obtaining quality related characteristics v by solving a mean square error loss function b ;
Depth support vector processing, constructing depth support vector data description for block level monitoring: by using nuclear skills v b Training depth support vector data description model And minimizing the supersphere volume; phi (g) is a deep neural network with L hidden layers and weights M. The depth support vector data description model penalizes the distance from phi (g) to the hyper-sphere center c for the first term, and the last term is a network weight attenuation regular term with the hyper-parameter lambda; the neural network of the depth support vector data description model adopts a two-layer hidden layer design, and the dimension and super parameters of proper hidden features are set, and the confidence level is set to alpha=0.99; optimizing weight matrix M by random gradient descent algorithm b So that v b Hyper-sphere center c capable of being clustered in feature space b Around.
Preferably, the step of filling missing data in the online monitoring method specifically includes:
a: defining observed data Z in industrial process after eliminating abnormal value obs And missing data Z obs Order-making Select reasonable initial value to replace Z mis ;
b: constructing posterior distribution P (theta|Z) of variation information bottleneck model parameter theta obs )=∫P(θ|Z obs ,Z mis )P(Z obs ,Z mis )dZ mis And given Z obs Z of (2) mis The conditional density P (Z) mis |Z obs )=∫P(Z mis |Z obs ,θ)P(θ|Z obs )dθ;
c: from the conditional density P of θ given Z (θ|z obs ,Z mis ) The initial parameter θ, the posterior distribution P (θ|z obs );
d: based on the current parameter theta i Based on the second step, missing data is calculated(i+1) th Iterating for the second time;
e: from the density of conditions(i+1) th Calculating estimation parameter of variation information bottleneck model in iteration>
f: repeating the step c and the step d to generate a Markov chainThe distribution of the missing data is given by a Monte Carlo method; thereby obtaining a complete data set Z ' = [ X ', Y ] ']Wherein X 'and Y' are complete process data and outlier-free quality data, respectively.
Compared with the prior art, the invention has the beneficial effects that:
the invention overcomes the influence of abnormal values on the model by using a partial robust M regression method, and constructs a plurality of complete data sets by taking uncertainty of missing data into consideration based on Bayesian reasoning; based on the information bottleneck criterion, the model is parameterized by utilizing the deep learning and variation theory, so that effective extraction of output related features is realized.
In summary, the method of the invention can be realized:
1) The invention can effectively solve the problem of data missing in the online monitoring stage of the nonlinear process;
2) The invention can reduce false alarm rate, improve fault detection rate and detect faults in early stage with shorter detection time delay;
3) The method can also be used for solving various types of missing data scenes such as classification, process control and the like, and has wide application prospect.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2 is a schematic diagram of a network representation of deep support vector data description in embodiment 2 of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1: referring to fig. 1, the present invention provides a technical solution: an on-line monitoring method based on information multiple interpolation and variation information bottleneck comprises the following steps:
and a data acquisition step: aiming at application scenes of process monitoring and fault detection under different data loss conditions, arranging sensors at reasonable positions in an industrial process to collect data; performing fault type labeling on each original data sample to construct a training data set;
data primary analysis: dividing the acquired data into process data X= [ X ] 1 ,x 2 ,…,x n ] T And quality data y= [ Y ] 1 ,y 2 ,…,y n ] T . If Y is a single quality variable, the outliers in X and Y can be directly eliminated by using partial robust M regressionThe method comprises the steps of carrying out a first treatment on the surface of the For a plurality of quality variables, the quality variables can be converted into single quality variables by a weighting method, and then the single quality variables are processed by using partial robust M regression;
and (3) eliminating abnormal values: normalizing the process data X and the quality data Y; poor lever weight by giving different weights to samples using a partial robust M regression methodHigh residual weight->And general weight w i The weight matrix W is initialized, and the influence of an abnormal value on the model is overcome; calculating weighted process data X W =wx and weighted quality data Y W WY, where w=diag (W).
The missing data filling steps are as follows: based on bayesian reasoning, a number of possible interpolations are generated taking into account the uncertainty of the missing data, thus forming a number of complete data sets. Then, the same processing is carried out on each data set, an estimated amount is obtained through comprehensive analysis, and statistical inference is completed.
The method comprises the following specific steps:
a: defining observed data Z in industrial process after eliminating abnormal value obs And missing data Z obs Order-makingSelect reasonable initial value to replace Z mis ;
b: constructing posterior distribution P (theta|Z) of variation information bottleneck model parameter theta obs )=∫P(θ|Z obs ,Z mis )P(Z obs ,Z mis )dZ mis And given Z obs Z of (2) mis The conditional density P (Z) mis |Z obs )=∫P(Z mis |Z obs ,θ)P(θ|Z obs )dθ;
c: from the conditional density P of θ given Z (θ|z obs ,Z mis ) The initial parameter theta is used for calculating posterior distribution in an iterative modeP(θ|Z obs );
d: based on the current parameter theta i Based on the second step, missing data is calculated(i+1) th Iterating for the second time;
e: from the density of conditions(i+1) th Calculating estimation parameter of variation information bottleneck model in iteration>
f: repeating the step c and the step d to generate a Markov chainThe distribution of missing data is given by the monte carlo method. Thereby obtaining a complete data set Z ' = [ X ', Y ] ']Wherein X 'and Y' are complete process data and outlier-free quality data, respectively.
Example 2: referring to fig. 1-2, the present invention provides a technical solution: an offline modeling method based on variation information bottleneck and depth support vector data description comprises the following steps:
data preprocessing: based on prior knowledge and expert experience, the preprocessed data X' is divided into: x' = [ X 1 ',X 2 ',…,X b ',…,X B ']B is the number of process variables and B is the number of blocks. For each block, corresponding process data X is used b 'and quality data Y' to train a variation information bottleneck model;
extracting quality related features by using variation information bottlenecks: on the basis of an information bottleneck criterion, parameterizing the model by utilizing a deep learning and variation theory; in order to realize effective output related feature extraction under the condition of data missing, an encoder q is designed in a variation information bottleneck model φ (v|x) and decoder p θ (y|x) wherein phi and theta are variation parameters, respectivelyAnd generating parameters; obtaining quality related characteristics v by solving a mean square error loss function b ;
Constructing a depth support vector data description for block level monitoring: by using nuclear skills v b Training depth support vector data description modelAnd minimizing the supersphere volume; phi (g) is a deep neural network with L hidden layers and weights M. The depth support vector data description model penalizes the distance from phi (g) to the hyper-sphere center c for the first term, and the last term is a network weight attenuation regular term with the hyper-parameter lambda; the neural network of the depth support vector data description model adopts a two-layer hidden layer design, and the dimension and super parameters of proper hidden features are set, and the confidence level is set to alpha=0.99; optimizing weight matrix M by random gradient descent algorithm b So that v b Hyper-sphere center c capable of being clustered in feature space b Around.
In this embodiment, in order to avoid collapse of the hypersphere (optimal radius R * =0), the present invention fixes the hypersphere center c+.0 as the mean of the initial network representation and uses an unbounded activation function in the network.
In this embodiment, for a new sample x new A monitoring statistic of block b is designed to Wherein v is b,new Is the new sample x in block b new Quality related features of (a).
In the present embodiment, construct S b (x) Probability density function of (c): probability density function of Wherein K (g) is a kernel function and H is a bandwidth; by solving integral formulasCalculate S b (x) Threshold of>Is S b (x) Is set to a threshold value of (2). If S b (x) Beyond the limit ofThe fault occurs in the corresponding zone b.
And constructing a global level monitoring model by fusing the block level monitoring results.
The monitoring performance of different faults in each mode was evaluated using the fault detection rate FDR, which was defined as:wherein S is a statistic and f represents a fault.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (2)
1. The intelligent data monitoring method based on the partial robust M regression and the multiple interpolation is characterized by comprising an online monitoring method based on the information multiple interpolation and the variation information bottleneck and an offline modeling method based on the variation information bottleneck and the depth support vector data description, wherein the online monitoring method based on the information multiple interpolation and the variation information bottleneck comprises the following steps:
data acquisition and preprocessing, namely arranging sensors at reasonable positions of equipment to be detected aiming at application scenes of process monitoring and fault detection under different data loss conditions so as to collect data; dividing the length of a data sample according to the sampling frequency of the sensor and the collected data quantity; performing fault type labeling on each original data sample to construct a training data set;
data analysis, namely dividing the acquired data into process data X= [ X ] according to specific application scenes 1 ,x 2 ,...,x n ] T And quality data y= [ Y ] 1 ,y 2 ,...,y n ] T If Y is a single quality variable, abnormal values in X and Y can be removed directly by using partial robust M regression; for a plurality of quality variables, converting the quality variables into single quality variables by a weighting method, and then processing the single quality variables by using a partial robust M regression;
reject outliers: normalizing the process data X and the quality data Y; by calculating bad lever weightsHigh residual weight->And general weight wi, initializing a weight matrix W; calculating weighted process data xw=wx and weighted quality data Y W WY, where w=diag (W);
filling in missing data: based on bayesian reasoning, a number of possible interpolations are generated taking into account the uncertainty of the missing data, thus forming a number of complete data sets. Then, carrying out the same processing on each data set, obtaining an estimated value through comprehensive analysis, and completing statistical inference;
the offline modeling method based on the variation information bottleneck and the depth support vector data description comprises the following steps:
acquiring and preprocessing data, acquiring sensor historical data, and dividing preprocessed data X' into: x '= [ X ]' 1 ,X′ 2 ,...,X′ b ,...,X′ B ]Where B is the number of process variables and B is the number of blocks, for each block, the corresponding process data X 'is used' b And quality data Y' to train a variational information bottleneck modelA shape;
and (3) processing the variation bottleneck, and extracting quality related characteristics by using the variation information bottleneck: on the basis of an information bottleneck criterion, parameterizing the model by utilizing a deep learning and variation theory; in order to realize effective output related feature extraction under the condition of data missing, an encoder q is designed in a variation information bottleneck model φ (v|x) and decoder p θ (y|x), wherein phi and theta are the variation parameter and the generation parameter, respectively; obtaining quality related characteristics v by solving a mean square error loss function b ;
Depth support vector processing, constructing depth support vector data description for block level monitoring: by using nuclear skills v b Training depth support vector data description model And minimizing the supersphere volume; phi (g) is a deep neural network with L hidden layers and weights M. The depth support vector data description model penalizes the distance from phi (g) to the hyper-sphere center c for the first term, and the last term is a network weight attenuation regular term with the hyper-parameter lambda; the neural network of the depth support vector data description model adopts a two-layer hidden layer design, and the dimension and super parameters of proper hidden features are set, and the confidence level is set to alpha=0.99; optimizing weight matrix M by random gradient descent algorithm b So that v b Hyper-sphere center c capable of being clustered in feature space b Around.
2. The intelligent data monitoring method based on partial robust M regression and multiple interpolation according to claim 2, wherein: the step of filling the missing data in the online monitoring method specifically comprises the following steps:
a: defining observed data Z in industrial process after eliminating abnormal value obs And missing data Z obs Order-making Select reasonable initial value to replace Z mis ;
b: constructing posterior distribution P (theta|Z) of variation information bottleneck model parameter theta obs )=∫P(θ|Z obs ,Z mis )P(Z obs ,Z mis )dZ mis And given Z obs Z of (2) mis The conditional density P (Z) mis |Z obs )=∫P(Z mis |Z obs ,θ)P(θ|Z obs )dθ;
c: from the conditional density P of θ given Z (θ|z obs ,Z mis ) The initial parameter θ, the posterior distribution P (θ|z obs );
d: based on the current parameter theta i Based on the second step, missing data is calculated(i+1) th Iterating for the second time;
e: from the density of conditions(i+1) th Calculating estimation parameter of variation information bottleneck model in iteration>
f: repeating the step c and the step d to generate a Markov chainThe distribution of the missing data is given by a Monte Carlo method; thereby obtaining a complete data set Z ' = [ X ', Y ] ']Wherein X 'and Y' are complete process data and outlier-free quality data, respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310925203.0A CN116933643A (en) | 2023-07-25 | 2023-07-25 | Intelligent data monitoring method based on partial robust M regression and multiple interpolation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310925203.0A CN116933643A (en) | 2023-07-25 | 2023-07-25 | Intelligent data monitoring method based on partial robust M regression and multiple interpolation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116933643A true CN116933643A (en) | 2023-10-24 |
Family
ID=88392159
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310925203.0A Pending CN116933643A (en) | 2023-07-25 | 2023-07-25 | Intelligent data monitoring method based on partial robust M regression and multiple interpolation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116933643A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117609737A (en) * | 2024-01-18 | 2024-02-27 | 中国人民解放军火箭军工程大学 | Method, system, equipment and medium for predicting health state of inertial navigation system |
-
2023
- 2023-07-25 CN CN202310925203.0A patent/CN116933643A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117609737A (en) * | 2024-01-18 | 2024-02-27 | 中国人民解放军火箭军工程大学 | Method, system, equipment and medium for predicting health state of inertial navigation system |
CN117609737B (en) * | 2024-01-18 | 2024-03-19 | 中国人民解放军火箭军工程大学 | Method, system, equipment and medium for predicting health state of inertial navigation system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109492193B (en) | Abnormal network data generation and prediction method based on deep machine learning model | |
CN113255848B (en) | Water turbine cavitation sound signal identification method based on big data learning | |
CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
CN117290800B (en) | Timing sequence anomaly detection method and system based on hypergraph attention network | |
CN112000084B (en) | Intelligent BIT design method of controller module based on 1D-CNN and GRU-SVM | |
CN112668775A (en) | Air quality prediction method based on time sequence convolution network algorithm | |
CN115484102A (en) | Industrial control system-oriented anomaly detection system and method | |
CN116933643A (en) | Intelligent data monitoring method based on partial robust M regression and multiple interpolation | |
CN114282443A (en) | Residual service life prediction method based on MLP-LSTM supervised joint model | |
CN109298633A (en) | Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization | |
CN113780420B (en) | GRU-GCN-based method for predicting concentration of dissolved gas in transformer oil | |
CN111191823A (en) | Production logistics prediction method based on deep learning | |
CN114500004A (en) | Anomaly detection method based on conditional diffusion probability generation model | |
CN111709577B (en) | RUL prediction method based on long-range correlation GAN-LSTM | |
CN118155662B (en) | Transformer voiceprint fault identification method based on artificial intelligence | |
CN116415505A (en) | System fault diagnosis and state prediction method based on SBR-DBN model | |
CN117408422A (en) | Natural gas flowmeter performance evaluation method and system based on deep learning | |
CN113283546B (en) | Furnace condition abnormity alarm method and system of heating furnace integrity management centralized control device | |
CN113151842B (en) | Method and device for determining conversion efficiency of wind-solar complementary water electrolysis hydrogen production | |
CN110779988A (en) | Bolt life prediction method based on deep learning | |
CN117633690A (en) | Rotary machine health state monitoring method and equipment based on data driving | |
CN113159088A (en) | Fault monitoring and diagnosis method based on multi-feature fusion and width learning | |
CN116821610A (en) | Method for optimizing wind power generation efficiency by utilizing big data | |
CN116561691A (en) | Power plant auxiliary equipment abnormal condition detection method based on unsupervised learning mechanism | |
CN115081483A (en) | Hydro-generator rotor fault diagnosis method based on feature selection and GWO-BP |
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 |