CN115096627A - Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment - Google Patents

Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment Download PDF

Info

Publication number
CN115096627A
CN115096627A CN202210679014.5A CN202210679014A CN115096627A CN 115096627 A CN115096627 A CN 115096627A CN 202210679014 A CN202210679014 A CN 202210679014A CN 115096627 A CN115096627 A CN 115096627A
Authority
CN
China
Prior art keywords
equipment
maintenance
state
data
model
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.)
Granted
Application number
CN202210679014.5A
Other languages
Chinese (zh)
Other versions
CN115096627B (en
Inventor
钱康安
潘晴
黄明辉
陈琬淇
李毅波
余艺
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.)
Central South University
Original Assignee
Central South University
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 Central South University filed Critical Central South University
Priority to CN202210679014.5A priority Critical patent/CN115096627B/en
Publication of CN115096627A publication Critical patent/CN115096627A/en
Application granted granted Critical
Publication of CN115096627B publication Critical patent/CN115096627B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M99/00—Subject matter not provided for in other groups of this subclass
    • G01M99/005—Testing of complete machines, e.g. washing-machines or mobile phones
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/80—Management or planning

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention provides a method and a system for fault diagnosis and operation and maintenance in a manufacturing process of hydraulic forming intelligent equipment, wherein the method comprises the following steps: acquiring and predicting the state of hydraulic forming intelligent equipment; hydraulic forming intelligent equipment state classification, wherein the state is subjected to multi-classification to form an equipment health state learning label; diagnosing the fault of the hydraulic forming intelligent equipment, and evaluating the health state of the current equipment; performing fault early warning maintenance on the hydroforming intelligent equipment, analyzing the current state of the equipment, and stopping the equipment for maintenance when a fault is predicted to occur; the intelligent operation and maintenance of the intelligent hydroforming equipment is realized, an expert system and a knowledge graph of the intelligent equipment are constructed, and the quick and accurate matching of the operation and maintenance scheme is realized. The scheme can improve the precision and efficiency of operation and maintenance, reduce the time and energy cost of manual operation and maintenance, and improve the accuracy and effectiveness of fault diagnosis of the hydraulic forming equipment.

Description

Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment
Technical Field
The invention relates to the technical field of state monitoring, fault diagnosis and intelligent operation and maintenance of mechanical equipment, in particular to a fault diagnosis and operation and maintenance method and system in a manufacturing process of hydraulic forming intelligent equipment.
Background
With the development of intelligent factories and industrial technologies, the mechanism and structure of intelligent manufacturing equipment are gradually perfected and complicated, the requirements of factory production on the intelligent manufacturing equipment are high, and high-performance and high-precision equipment can ensure better processing quality and productivity. It is important for the regular management and maintenance of intelligent manufacturing equipment. However, the mechanism and structure of the intelligent manufacturing equipment are complex, and the manual operation and maintenance inspection causes huge waste of time and energy. And for some specific intelligent manufacturing equipment, the operation and maintenance of the equipment are limited by the professional knowledge of people.
Therefore, with the development of monitoring technology and the coming of big data era, how to extract high-quality information from massive data and solve the problem that the explosion of the state space of the data and the information isolated island are present. The operation and maintenance of the traditional intelligent manufacturing equipment judges the operation health state of the equipment through an empirical method and mechanism analysis, when the manufacturing equipment breaks down, the traditional method has the defects of long fault finding time delay, large repair risk and the like, and the traditional operation and maintenance method cannot meet the requirements of the current intelligent factory production development; and further failures cannot be diagnosed, i.e. the flexibility of failure operation and maintenance is lacked.
In recent years, big data, machine learning, and deep learning have become effective methods for condition monitoring and fault diagnosis. The big data technology provides data layer support for machine learning and deep learning, basic features of data can be mined and extracted through the machine learning and the deep learning, the defects of a traditional recommendation method are overcome to a certain extent, and high-quality operation and maintenance can be achieved. With the emergence of a multi-model integrated learning mechanism, a model integrated learning mechanism is merged into machine learning and deep learning applications in the prior art. The model integrated learning mechanism references the thinking, decision and analysis processes of human brain, and improves the learning effect and efficiency through the complementation, competition and fusion of the model.
Disclosure of Invention
The purpose of the invention is: in view of the defects in the background art, a fault diagnosis and operation and maintenance scheme for a manufacturing process of a hydraulic forming intelligent device based on depth feature self-learning is provided, and the scheme relates to, but is not limited to, the hydraulic forming intelligent manufacturing device, and can also be generalized to other kinds of intelligent manufacturing devices. According to the scheme, the accuracy and the efficiency of intelligent manufacturing equipment fault diagnosis can be improved to a certain extent, and the operation and maintenance capacity of intelligent factory production equipment is further improved.
In order to achieve the purpose, the invention provides a fault diagnosis and operation and maintenance method for a hydraulic forming intelligent equipment manufacturing process, which comprises the following steps:
s1, acquiring and predicting the state of the intelligent hydraulic forming equipment, acquiring and fusing multivariate data by means of a sensing technology, preprocessing the data, including missing value supplement, abnormal value removal and standardization processing, and performing multi-model integrated prediction on various states;
s2, classifying the states of the hydroforming intelligent equipment, and performing multi-classification on the states by an unsupervised learning method;
s3, diagnosing the fault of the hydraulic forming intelligent equipment, performing multi-model integrated learning through a residual error model, a Gaussian mixture clustering model and a deep learning method, evaluating the health state of the current equipment, and introducing a competition fusion mechanism;
s4, carrying out early warning maintenance on the fault of the hydraulic forming intelligent equipment, analyzing the current state of the equipment, and stopping the equipment for maintenance when the fault is predicted to occur;
s5, carrying out intelligent operation and maintenance on the hydroforming intelligent equipment, constructing an expert system and a knowledge graph of the intelligent equipment, and realizing quick and accurate matching of an operation and maintenance scheme.
Further, S1 specifically includes the following sub-steps:
s11, based on hydraulic pressureThe entity structure of intelligent equipment is characterized by numbering the components to be monitored, and designing data acquisition devices comprising temperature sensors, pressure sensors, industrial cameras and the like, wherein different acquisition devices can acquire different characteristics of the equipment, and the acquired state characteristic data set is
Figure BDA0003697584460000021
S12, representing different characteristics by using category fields, and preprocessing data in the data set;
s13, for data sets with different characteristics, a prediction model with different characteristics is constructed, and integrated learning is carried out on the historical state based on the ARIMA-LSTM combination.
Further, in S12, missing values of data acquisition are supplemented based on the KNN algorithm, and a "complex decision coefficient" is used based on the KMEANS clustering method
Figure BDA0003697584460000031
The criterion is that the optimal subset regression method of the selection principle calculates and deletes the chemical and biochemical residual errors to remove the abnormal data;
the specific calculation process of the clustering method for detecting the abnormal value is as follows:
||X j -μ j ||>θ
wherein, | | X j -μ j The | | is the distance between the sample data and the center, and the theta is a set distance threshold; by "determining coefficients repeatedly
Figure BDA0003697584460000032
The specific calculation process of the optimal subset regression detection abnormal value for the selection principle is as follows:
y＝β 0 +β 1 x 1 +β 2 x 2 +...+β p x p +ε
Figure BDA0003697584460000033
wherein beta is a regression coefficient not equal to zero, xSelecting p variables from m variables to maximize the complex coefficient, wherein epsilon is an error term; r 2 In order to determine the coefficients for a complex number,
Figure BDA0003697584460000034
for the adjusted complex decision coefficient, n is the sample size;
Figure BDA0003697584460000035
Figure BDA0003697584460000036
wherein e i Is a common residual error, h ii Being diagonal elements on the hat matrix, SRE i For student formation of residual errors, SRE (i) To delete the biochemical residual; if SRE (i) |>3, judging as an abnormal value; then based on KMEANS method and "coefficient of complex decision
Figure BDA0003697584460000037
The criterion is that the optimal subset regression method of the selection principle is comprehensively judged, and if the data is judged to be an abnormal value, the data is judged to be abnormal and is removed;
finally, Z-score was used to de-dimensionalize the data.
Further, model C for state prediction based on ARIMA in S13 (ARIMA) (t)：
Figure BDA0003697584460000038
Wherein the content of the first and second substances,
Figure BDA0003697584460000039
theta is a regression coefficient not equal to zero, C is a stable, normal, zero-mean time sequence, and epsilon is white noise;
model C for predicting state based on LSTM (LSTM) (t): relative to oneIn a general RNN structure, a forgetting gate, an input gate and an output gate and a cell state vector are newly added to an LSTM, the cell state is used for storing important time series memory information, wherein the forgetting gate is the content for determining the removal or retention of the cell state of the previous layer and is completed through a Sigmoid function, and the updating formula is as follows:
f t ＝σ(W f ·[h t-1 ,x t ]+b f )
the input gate determines which new information needs to be updated to the cell state according to the current state information, and the updating formula is as follows, wherein
Figure BDA0003697584460000041
Candidate vectors for newly added information:
i t ＝σ(W i ·[h t-1 ，x t ]+b f )
Figure BDA0003697584460000042
the output gate selectively outputs the information of the cell state, and the final model output is h t ：
o t ＝σ(W o [h t-1 ,x t ]+b 0 )
h t ＝o t tanh(C t )
Through the linear weighted combination of the two models, the state prediction precision of different characteristics is improved: let y t (t 1,2,3.., n) is actual time-series data,
Figure BDA0003697584460000043
is the result of prediction of two methods, w i Is the weight coefficient of the ith prediction method, the predicted value Y of the combined model t Can be expressed as:
Figure BDA0003697584460000044
Figure BDA0003697584460000045
Figure BDA0003697584460000046
further, S2 specifically includes the following sub-steps:
s21, dividing the data collected by the hydroforming equipment, taking the characteristic data of different time intervals as different data samples, classifying the data samples of the time intervals by a KMEANS clustering method, primarily evaluating the health status, and selecting an initial clustering center mu 1 ,μ 2 ,…,μ k Are described as
Figure BDA0003697584460000047
Classifying each sample according to a nearest distance principle, recalculating a clustering center, wherein a center iteration formula is as follows:
Figure BDA0003697584460000048
and S22, introducing a clustering initial center self-adaptive adjustment mechanism, optimizing the clustering initial center through a self-adaptive genetic algorithm, selecting a fitness function as an intra-cluster error variance SSE, optimizing the clustering number through an Elbow method, and optimizing the clustering number k by adopting a DBI index.
Further, S3 specifically includes the following sub-steps:
s31, based on the residual error fault diagnosis model, according to the state prediction model of S13, counting the predicted value of the state monitoring data of the hydraulic forming equipment in the healthy running through the state parameter prediction model
Figure BDA0003697584460000051
With the true value y t The residual error of (2) is obtained by counting the mean value μ, the root mean square error RMSE and the information entropy h (x) ═ p (x) logp (x) dx and root of the residual error by using a sliding window algorithmDetermining a data degradation metric threshold according to the data change trend, thereby judging the health state of the hydraulic forming equipment:
Figure BDA0003697584460000052
s32, based on GMM fault diagnosis model, according to S11 data samples with different characteristics at a single moment, fully performing statistical analysis on the distribution conditions of a plurality of state characteristics at different moments, establishing a probability distribution model, and realizing identification of the health state of the hydroforming equipment and diagnosis of faults, wherein the calculation formula of the health decline index is shown as follows, x is a state parameter sequence, and w is a state parameter sequence k PHM as a weight of the GMM model * The greater (t) indicates a worse state of health of the hydroforming equipment:
Figure BDA0003697584460000053
Figure BDA0003697584460000054
s33, obtaining a data set (X) according to the preliminary diagnosis result of the health state of the hydraulic forming equipment of S1 and S2 based on the fault diagnosis model of the deep belief network 1 ,y 1 ),(X 2 ,y 2 ),...,(X i ,y i ),X i Is a sequence of states, y i Is a health index label; constructing a deep belief network, setting learning parameters and a network topology structure, preprocessing different characteristic category fields of S1 to be used as network input, training through unsupervised learning and semi-supervised fine tuning, and defining f as mapping of the deep belief network, wherein a fault diagnosis formula is as follows:
y(t)＝f(X)
s34, introducing a competitive fusion cooperation mechanism into the fault diagnosis model, inputting each data set into a corresponding prediction model of S31-S33 to obtain the prediction value of the model and the confidence interval of the model under the selected confidence degree(ii) a If the models are compatible, the health state of the current hydroforming equipment is synthesized through a hard voting method, and the health state is passed; otherwise, requesting the direct monitoring method to acquire the data set again for training; note N i ＝[N(1),N(2),N(3)]The epsilon {1,2,3} is a prediction result of the ith fault diagnosis model on the health state of the hydraulic forming equipment, and the comprehensive formula of the hard voting method is as follows:
Figure BDA0003697584460000061
further, S4 specifically includes the following sub-steps:
s41, performing preview and judgment on possible faults according to offline learning of historical data;
and S42, when the fault is judged to occur, checking and repairing the corresponding part according to the extracted features.
Further, S5 specifically includes the following sub-steps:
s51, matching the operation and maintenance requirements of the intelligent manufacturing equipment with a scheme library in the form of texts, pictures and the like through a professional field manual acquired by a web text crawler;
s52, mapping the fault description to the knowledge graph of the intelligent manufacturing equipment to form a fault description subgraph of the scheme library;
s53, when the user inputs the operation and maintenance demand, the operation and maintenance demand is mapped to the knowledge graph to form an operation and maintenance demand sub-graph;
and S54, searching the graph by using the knowledge graph, matching the operation and maintenance appeal subgraph of the user with the fault description subgraph of the scheme library, and finally obtaining the operation and maintenance scheme.
The invention also provides a hydraulic forming intelligent equipment manufacturing process fault diagnosis and operation and maintenance system based on depth characteristic self-learning, which comprises the following steps:
the hydraulic forming intelligent equipment state acquisition and prediction module acquires and fuses multivariate data based on a sensing technology, preprocesses the data, comprises missing value supplement, abnormal value elimination and standardization processing, constructs an ARIMA-LSTM combined model, and performs multi-model integrated prediction on various states;
the hydraulic forming intelligent equipment state classification module is used for performing multi-classification on states through an unsupervised learning method;
the hydraulic forming intelligent equipment fault diagnosis module carries out multi-model integrated learning through a residual error model, a Gaussian mixture clustering model and a deep learning method, evaluates the health state of the current equipment and introduces a competition fusion mechanism;
the hydraulic forming intelligent equipment fault early warning maintenance module analyzes the current state of the equipment and carries out shutdown maintenance on the equipment when a fault is predicted to occur;
the intelligent operation and maintenance module of the hydraulic forming intelligent equipment constructs an expert system and a knowledge graph of the intelligent equipment, and realizes quick and accurate matching of an operation and maintenance scheme.
The system comprises a data information layer, an information analysis processing layer and an application implementation layer, wherein a sensing part of a hydraulic forming intelligent equipment state acquisition and prediction module is arranged on the data information layer, the hydraulic forming intelligent equipment state acquisition and prediction module, the hydraulic forming intelligent equipment state classification module and the hydraulic forming intelligent equipment fault diagnosis module are arranged on the information analysis processing layer, and the hydraulic forming intelligent equipment fault early warning maintenance module and the hydraulic forming intelligent equipment intelligent operation and maintenance module are arranged on the application implementation layer.
The scheme of the invention has the following beneficial effects:
compared with the traditional operation and maintenance scheme which mainly adopts an empirical method and mechanism analysis to judge the operation health state of the equipment, the fault diagnosis and operation and maintenance scheme provided by the invention has the defects of long time delay for fault discovery, large repair risk and the like, the traditional operation and maintenance method cannot meet the requirement of the current intelligent factory production development, cannot diagnose more faults, namely the problems of lack of flexibility of fault operation and maintenance and the like, and the method combining big data, machine learning and deep learning can improve the precision and efficiency of operation and maintenance to a certain extent and reduce the time and energy cost of manual operation and maintenance; according to the scheme, multi-model integrated learning is introduced into machine learning and deep learning, an intelligent fault diagnosis system of the hydraulic forming equipment is realized, a competitive cooperation fusion mechanism is introduced into different prediction models, a prediction result and a confidence interval are determined by a single prediction model, single models are mutually supplemented, verified and supervised, if the single models can be fused with one another, a fault diagnosis result is comprehensively obtained, otherwise, data acquisition training is required again, and therefore the accuracy and the effectiveness of fault diagnosis of the hydraulic forming equipment are improved;
other advantages of the present invention will be described in detail in the detailed description that follows.
Drawings
FIG. 1 is a flow chart of method steps and corresponding system of the present invention;
FIG. 2 is a block diagram of the system of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments. It should be apparent that the described embodiments are only some of the embodiments of the present invention, and not all of them. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention. In addition, the technical features involved in the different embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
In describing the present invention, for the sake of simplicity of explanation, the method or rule is depicted and described as a series of acts that are not intended to be exhaustive or to limit the order of the acts. For example, the experimental procedures can be performed in various orders and/or simultaneously, and include other experimental procedures not described again. Moreover, not all illustrated steps may be required to implement a methodology or algorithm described herein. Those skilled in the art will recognize and appreciate that the methodologies and algorithms may be represented as a series of interrelated states via a state diagram or items.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on those shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element referred to must have a specific orientation, be constructed in a specific orientation, and be operated, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
Referring to fig. 1, an embodiment 1 of the present invention provides a method for diagnosing, operating and maintaining a fault in a manufacturing process of a hydraulic forming intelligent device, including the following steps:
s1, acquiring and predicting the state of the intelligent hydraulic forming equipment, acquiring and fusing multivariate data by means of a sensing technology, and preprocessing the data, wherein the preprocessing comprises missing value supplement, abnormal value elimination and standardization processing; and constructing an ARIMA-LSTM combined model, and performing multi-model integrated prediction on various states.
The S1 specifically includes the following sub-steps:
s11, numbering the parts to be monitored based on the entity structure of the intelligent hydroforming equipment, designing data acquisition devices including temperature sensors, pressure sensors, industrial cameras and the like, wherein different acquisition devices can acquire different characteristics of the equipment, and acquiring a state characteristic data set
Figure BDA0003697584460000091
S12, representing different characteristics by using category fields, and preprocessing data in the data set; missing values of data acquisition are supplemented based on a KNN algorithm, and a 'complex decision coefficient' is used for performing clustering based on KMEANS
Figure BDA0003697584460000092
The criterion is that the optimal subset regression method of the selection principle calculates and deletes the chemical and biochemical residual errors to remove the abnormal data.
The specific calculation process of the clustering method for detecting the abnormal value is as follows:
||X j -μ j ||>θ
wherein, | | X j -μ j And | | is the distance between the sample data and the center, and theta is a set distance threshold.
By "determining coefficients repeatedly
Figure BDA0003697584460000093
The criterion "the specific calculation process for regression detection of outliers for the optimal subset of the selection principle is as follows:
y＝β 0 +β 1 x 1 +β 2 x 2 +...+β p x p +ε
Figure BDA0003697584460000094
where β is a regression coefficient not zero, x is p variables selected from m variables such that the complex coefficient reaches a maximum, and ε is an error term. R 2 In order to determine the coefficients for a complex decision,
Figure BDA0003697584460000095
for the adjusted complex decision coefficient, n is the sample size.
Figure BDA0003697584460000096
Figure BDA0003697584460000097
Wherein e i Is a common residual error, h ii Being diagonal elements on the hat matrix, SRE i For student formation of residual errors, SRE (i) To delete the biochemical residuals. If SRE (i) |>And 3, judging as an abnormal value. Then based on KMEANS method and "coefficient of complex decision
Figure BDA0003697584460000098
The criterion is that the optimal subset regression method of the selection principle is comprehensively judged, and if the data is judged to be an abnormal value, the data is judged to be abnormal and is removed.
Finally, Z-score was used to de-dimensionalize the data.
S13, for data sets with different characteristics, a prediction model with different characteristics is constructed, and integrated learning is carried out on the historical state based on the ARIMA-LSTM combination.
In particular, model C for state prediction based on ARIMA (ARIMA) (t)：
Figure BDA0003697584460000101
Wherein the content of the first and second substances,
Figure BDA0003697584460000102
theta is a regression coefficient not equal to zero, C is a stationary, normal, zero-mean time series, and epsilon is white noise.
Model C for predicting state based on LSTM (LSTM) (t): compared with a general RNN structure, the LSTM is newly added with three control gates, namely a forgetting gate, an input gate and an output gate and a cell state vector, the cell state is used for storing more important time series memory information, the forgetting gate is the content for determining the removal or retention of the cell state of the previous layer and is completed through a Sigmoid function, and the updating formula is as follows:
f t ＝σ(W f ·[h t-1 ,x t ]+b f )
the input gate determines which new information needs to be updated to the cell state according to the current state information, and the updating formula is as follows, wherein
Figure BDA0003697584460000103
Candidate vectors for newly added information:
i t ＝σ(W i ·[h t-1 ，x t ]+b f )
Figure BDA0003697584460000104
the output gate selectively outputs the information of the cell state, and the final model output is h t ：
o t ＝σ(W o [h t-1, x t ]+b 0 )
h t ＝o t tanh(C t )
Through the linear weighted combination of the two models, the state prediction precision of different characteristics is improved: let y t (t 1,2,3.., n) is actual time-series data,
Figure BDA0003697584460000105
is the result of prediction of two methods, w i Is the weight coefficient of the ith prediction method, the predicted value Y of the combined model t Can be expressed as:
Figure BDA0003697584460000106
Figure BDA0003697584460000107
Figure BDA0003697584460000111
and S2, classifying the states of the hydroforming intelligent equipment, and performing multi-classification on the states by an unsupervised learning method.
The S2 specifically includes the following sub-steps:
s21, dividing the data collected by the hydroforming equipment, taking the characteristic data of different time intervals as different data samples, classifying the data samples of the time intervals by a KMEANS clustering method, primarily evaluating the health status, and selecting an initial clustering center mu 1 ,μ 2 ,…,μ k Are described as
Figure BDA0003697584460000112
Classifying each sample according to a nearest distance principle, recalculating a clustering center, wherein a center iteration formula is as follows:
Figure BDA0003697584460000113
and S22, introducing a clustering initial center self-adaptive adjustment mechanism, optimizing the clustering initial center through a self-adaptive genetic algorithm, selecting a fitness function as an intra-cluster error variance SSE, optimizing the clustering number through an Elbow method, and optimizing the clustering number k by adopting a DBI index.
And S3, diagnosing the fault of the hydraulic forming intelligent equipment, performing multi-model ensemble learning through a residual error model, a Gaussian mixture clustering model and a deep learning method on the basis of S1 and S2, evaluating the health state of the current equipment, and introducing a competition fusion mechanism.
The S3 specifically includes the following sub-steps:
s31, based on the residual error fault diagnosis model, according to the state prediction model of S13, counting the predicted value of the state monitoring data of the hydraulic forming equipment in the healthy running through the state parameter prediction model
Figure BDA0003697584460000114
With the true value y t The residual error of (2) is counted by using a sliding window algorithm, and the average value mu, the root mean square error RMSE and the information entropy h (x) ═ p (x) logp (x) dx of the residual error are calculated, and according to the variation trend of the data, a data degradation metric threshold value is determined, so that the health state of the hydroforming equipment is judged:
Figure BDA0003697584460000115
s32, based on the fault diagnosis model of the GMM, according to the data samples of different characteristics at a single moment of S11, fully performing statistical analysis on the distribution conditions of a plurality of state characteristics at different moments, establishing a probability distribution model, and realizing the hydraulic forming equipmentThe health state identification and fault diagnosis are carried out by calculating the health decline index according to the formula shown below, wherein x is the state parameter sequence, and w k PHM as a weight of the GMM model * The greater (t) indicates a worse state of health of the hydroforming equipment.
Figure BDA0003697584460000121
Figure BDA0003697584460000122
S33, obtaining a data set (X) according to the preliminary diagnosis result of the health state of the hydraulic forming equipment of S1 and S2 based on the fault diagnosis model of the deep belief network 1 ,y 1 ),(X 2 ,y 2 ),...,(X i ,y i ),X i Is a sequence of states, y i Is a health index label. Constructing a deep belief network, setting learning parameters and a network topology structure, preprocessing different characteristic category fields of S1 to be used as network input, training through unsupervised learning and semi-supervised fine tuning, and defining f as mapping of the deep belief network, wherein a fault diagnosis formula is as follows:
y(t)＝f(X)
s34, introducing a competitive fusion cooperation mechanism into the fault diagnosis model, inputting each data set into a corresponding prediction model of S31-S33, and obtaining a prediction value of the model and a confidence interval of the model under the selected confidence coefficient; if the models are compatible, synthesizing the health state of the current hydroforming equipment through a hard voting method, and passing; otherwise, requesting the direct monitoring method to acquire the data set again for training. Note N i ＝[N(1),N(2),N(3)]The element {1,2,3} belongs to the prediction result of the ith fault diagnosis model on the health state of the hydroforming equipment, and the comprehensive formula of the hard voting method is as follows:
Figure BDA0003697584460000123
and S4, carrying out early warning maintenance on the hydraulic forming intelligent equipment fault, analyzing the current state of the equipment on the basis of S1 and S3, and stopping the equipment for maintenance by operation and maintenance personnel when the fault is predicted to occur.
The S4 specifically includes the following sub-steps:
s41, previewing and judging the possible faults according to the off-line learning of the historical data of the steps S1-S3;
and S42, when the fault is judged to occur, checking and repairing the corresponding part according to the extracted features.
S5, carrying out intelligent operation and maintenance on the intelligent hydraulic forming equipment, constructing an expert system and a knowledge graph of the intelligent equipment, and realizing quick and accurate matching of the operation and maintenance scheme.
The S5 specifically includes the following sub-steps:
s51, matching the operation and maintenance requirements of the intelligent manufacturing equipment with a scheme library in the form of texts, pictures and the like through a professional field manual acquired by a web text crawler;
s52, mapping the fault description to the knowledge graph of the intelligent manufacturing equipment to form a fault description subgraph of the scheme library;
s53, when the user inputs the operation and maintenance demand, the operation and maintenance demand is mapped to the knowledge graph to form an operation and maintenance demand sub-graph;
and S54, searching the graph by using the knowledge graph, matching the operation and maintenance appeal subgraph of the user with the fault description subgraph of the scheme library, and finally obtaining the operation and maintenance scheme.
In a word, the method firstly designs a data acquisition system for the hydraulic forming intelligent equipment, numbers the position and the characteristics of the acquired data, and obtains a required data set through data migration calculation and pretreatment on the basis; secondly, building an ARIMA-LSTM combined state prediction model, wherein the model is combined by adopting a weighted combination mode on the basis of an ARIMA and LSTM single model; further performing unsupervised classification learning on sample data in different time periods, and further adjusting the initial clustering center and the number of clustering clusters by using a self-adaptive genetic algorithm and an Elbow method on the basis of a KEMANS method; then, on the basis of off-line learning of state prediction, faults are diagnosed through a residual error model, a Gaussian mixture model and a deep belief network, and a competitive fusion mechanism is further introduced into the off-line learning, so that mutual verification and supervision of learning results are realized, and the fault diagnosis results are more credible; and finally, constructing a knowledge graph to realize quick and accurate matching of the operation and maintenance method, so that the operation and maintenance result is more flexible and has higher reliability, the time and energy cost of manual maintenance is reduced, and the operation and maintenance efficiency of equipment is improved.
Therefore, compared with the traditional operation and maintenance scheme which mainly judges the operation health state of equipment by an empirical method and mechanism analysis, the method has the defects of long time delay for fault discovery, high repair risk and the like, the traditional operation and maintenance method cannot meet the requirements of the production and development of the current intelligent factory, and more faults cannot be diagnosed, namely the problems of lack of flexibility of fault operation and maintenance and the like are solved; according to the method, multi-model integrated learning is introduced into machine learning and deep learning, an intelligent fault diagnosis system of the hydroforming equipment is realized, a competitive cooperation fusion mechanism is introduced into different prediction models, a prediction result and a confidence interval are determined by the single prediction model, single models are mutually supplemented, verified and supervised, if the single models can be fused with one another, a fault diagnosis result is comprehensively obtained, otherwise, data re-acquisition training is requested, and therefore the accuracy and the effectiveness of the fault diagnosis of the hydroforming equipment are improved.
Example 2:
the embodiment 2 of the invention provides a hydraulic forming intelligent equipment manufacturing process fault diagnosis and operation and maintenance system, which comprises:
the hydraulic forming intelligent equipment state acquisition and prediction module acquires and fuses multivariate data based on a sensing technology, preprocesses the data, comprises missing value supplement, abnormal value elimination and standardization processing, constructs an ARIMA-LSTM combined model, and performs multi-model integrated prediction on various states;
the hydraulic forming intelligent equipment state classification module classifies states in a multi-classification mode through an unsupervised learning method;
the hydraulic forming intelligent equipment fault diagnosis module carries out multi-model integrated learning through a residual error model, a Gaussian mixture clustering model and a deep learning method, evaluates the health state of the current equipment and introduces a competition fusion mechanism;
the hydraulic forming intelligent equipment fault early warning and maintaining module analyzes the current state of the equipment and stops the equipment for maintenance by operation and maintenance personnel when a fault is predicted to occur;
the intelligent operation and maintenance module of the hydroforming intelligent equipment constructs an expert system and a knowledge graph of the intelligent equipment, and realizes quick and accurate matching of an operation and maintenance scheme.
Meanwhile, as shown in fig. 2, the system includes a data information layer, an information analysis processing layer, and an application implementation layer. For a data information layer, data are firstly acquired through a multi-sensor system, and then data sets with different characteristics are constructed through a data migration calculation and preprocessing method. For the information analysis processing layer, the layer mainly calculates and analyzes the data set of the data information layer and carries out fault diagnosis results. And for the application implementation layer, the health state evaluation and intelligent operation and maintenance are carried out on the running state of the hydroforming intelligent equipment.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (10)

1. A hydraulic forming intelligent equipment manufacturing process fault diagnosis and operation and maintenance method is characterized by comprising the following steps:
s1, acquiring and predicting the state of the intelligent hydraulic forming equipment, acquiring and fusing multivariate data by means of a sensing technology, preprocessing the data, including missing value supplement, abnormal value removal and standardization processing, and performing multi-model integrated prediction on various states;
s2, classifying the states of the hydroforming intelligent equipment, and performing multi-classification on the states by an unsupervised learning method;
s3, diagnosing the fault of the hydraulic forming intelligent equipment, performing multi-model integrated learning through a residual error model, a Gaussian mixture clustering model and a deep learning method, evaluating the health state of the current equipment, and introducing a competition fusion mechanism;
s4, carrying out early warning maintenance on the fault of the hydraulic forming intelligent equipment, analyzing the current state of the equipment, and stopping the equipment for maintenance when the fault is predicted to occur;
s5, carrying out intelligent operation and maintenance on the intelligent hydraulic forming equipment, constructing an expert system and a knowledge graph of the intelligent equipment, and realizing quick and accurate matching of the operation and maintenance scheme.
2. The method for fault diagnosis and operation and maintenance in the manufacturing process of the hydroformed intelligent equipment according to claim 1, wherein S1 specifically comprises the following substeps:
s11, numbering the parts to be monitored based on the entity structure of the intelligent hydroforming equipment, designing data acquisition devices including temperature sensors, pressure sensors, industrial cameras and the like, wherein different acquisition devices can acquire different characteristics of the equipment, and acquiring a state characteristic data set
Figure FDA0003697584450000011
S12, representing different characteristics by using category fields, and preprocessing data in the data set;
s13, for data sets with different characteristics, a prediction model with different characteristics is constructed, and integrated learning is carried out on the historical state based on the ARIMA-LSTM combination.
3. The method for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment according to claim 2, wherein missing values of data acquisition in S12 are supplemented based on KNN algorithm, and based on KMEANS clustering method, a 'double decision coefficient' is adopted
Figure FDA0003697584450000012
The criterion is that the optimal subset regression method of the selection principle calculates and deletes the chemical and biochemical residual errors to remove the abnormal data;
the specific calculation process of the clustering method for detecting the abnormal value is as follows:
||X j -μ j ||＞θ
wherein, | | X j -μ j The | | is the distance between the sample data and the center, and the theta is a set distance threshold; by "determining coefficients repeatedly
Figure FDA0003697584450000021
The criterion "the specific calculation process for regression detection of outliers for the optimal subset of the selection principle is as follows:
y＝β 0 +β 1 x 1 +β 2 x 2 +...+β p x p +ε
Figure FDA0003697584450000022
wherein, beta is a regression coefficient which is not zero, x is p variables which are selected from m variables and enable the complex decision coefficient to reach the maximum, and epsilon is an error term; r is 2 In order to determine the coefficients for a complex decision,
Figure FDA0003697584450000023
for the adjusted complex decision coefficient, n is the sample size;
Figure FDA0003697584450000024
Figure FDA0003697584450000025
wherein e i Is a common residual error, h ii Being diagonal elements on the hat matrix, SRE i To studyBiochemical residual error, SRE (i) To delete the biochemical residual; if SRE (i) If the value is greater than 3, judging the value to be an abnormal value; then based on KMEANS method and "coefficient of complex decision
Figure FDA0003697584450000026
The criterion is that the optimal subset regression method of the selection principle is comprehensively judged, and if the data is judged to be an abnormal value, the data is judged to be abnormal and is removed;
finally, Z-score was used to de-dimensionalize the data.
4. The method for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment as claimed in claim 3, wherein the model C for state prediction based on ARIMA in S13 (ARIMA) (t)：
Figure FDA0003697584450000027
Wherein the content of the first and second substances,
Figure FDA0003697584450000028
theta is a non-zero regression coefficient, C is a stable, normal, zero-mean time sequence, and epsilon is white noise;
model C for predicting state based on LSTM (LSTM) (t): compared with a general RNN structure, the LSTM is newly added with three control gates, namely a forgetting gate, an input gate and an output gate and a cell state vector, the cell state is used for storing more important time series memory information, the forgetting gate is the content for determining the removal or retention of the cell state of the previous layer and is completed through a Sigmoid function, and the updating formula is as follows:
f t ＝σ(W f ·[h t-1 ，x t ]+b f )
the input gate determines which new information needs to be updated to the cell state according to the current state information, and the update formula is as follows, wherein
Figure FDA0003697584450000031
Candidate vectors for newly added information:
i t ＝σ(W i ·[h t-1 ，x t ]+b f )
Figure FDA0003697584450000032
the output gate selectively outputs the information of the cell state, and the final model output is h t ：
o t ＝σ(W o [h t-1 ，x t ]+b 0 )
h t ＝o t tanh(C t )
Through the linear weighted combination of the two models, the state prediction precision of different characteristics is improved: let y t (t 1,2,3.., n) is actual time-series data,
Figure FDA0003697584450000033
is the result of prediction of two methods, w i Is the weight coefficient of the ith prediction method, the predicted value Y of the combined model t Can be expressed as:
Figure FDA0003697584450000034
Figure FDA0003697584450000035
Figure FDA0003697584450000036
5. the method for fault diagnosis and operation and maintenance in the manufacturing process of the hydroformed intelligent equipment according to claim 4, wherein S2 specifically comprises the following substeps:
s21, dividing the data collected by the hydroforming equipment, taking the characteristic data of different time intervals as different data samples, classifying the data samples of the time intervals by a KMEANS clustering method, primarily evaluating the health status, and selecting an initial clustering center mu 1 ，μ 2 ，...，μ k Are described as
Figure FDA0003697584450000037
Classifying each sample according to a nearest distance principle, recalculating a clustering center, wherein a center iteration formula is as follows:
Figure FDA0003697584450000038
and S22, introducing a clustering initial center self-adaptive adjustment mechanism, optimizing the clustering initial center through a self-adaptive genetic algorithm, selecting a fitness function as an intra-cluster error variance SSE, optimizing the clustering number through an Elbow method, and optimizing the clustering number k by adopting a DBI index.
6. The method for fault diagnosis and operation and maintenance in the manufacturing process of the hydroformed intelligent equipment according to claim 5, wherein S3 specifically comprises the following substeps:
s31, based on the residual error fault diagnosis model, according to the state prediction model of S13, counting the predicted value of the state monitoring data of the hydraulic forming equipment in the healthy running through the state parameter prediction model
Figure FDA0003697584450000044
With the true value y t The residual error of (2) is counted by using a sliding window algorithm, and the average value mu, the root mean square error RMSE and the information entropy h (x) ═ p (x) logp (x) dx of the residual error are calculated, and according to the variation trend of the data, a data degradation metric threshold value is determined, so that the health state of the hydroforming equipment is judged:
Figure FDA0003697584450000041
s32, based on GMM fault diagnosis model, according to S11 data samples with different characteristics at a single moment, fully performing statistical analysis on the distribution conditions of a plurality of state characteristics at different moments, establishing a probability distribution model, and realizing identification of the health state of the hydroforming equipment and diagnosis of faults, wherein the calculation formula of the health decline index is shown as follows, x is a state parameter sequence, and w is a state parameter sequence k PHM as a weight of the GMM model * The greater (t) indicates a poorer state of health of the hydroforming equipment:
Figure FDA0003697584450000042
Figure FDA0003697584450000043
s33, based on the fault diagnosis model of the deep belief network, obtaining a data set (X) according to the initial diagnosis results of the hydraulic forming equipment health states of S1 and S2 1 ，y 1 )，(X 2 ，y 2 )，...，(X i ，y i )，X i As a sequence of states, y i Is a health index label; constructing a deep belief network, setting learning parameters and a network topology structure, preprocessing different characteristic category fields of S1 to be used as network input, training through unsupervised learning and semi-supervised fine tuning, and defining f as mapping of the deep belief network, wherein a fault diagnosis formula is as follows:
y(t)＝f(X)
s34, introducing a competitive fusion cooperation mechanism into a fault diagnosis model, inputting each data set into a corresponding prediction model of S31-S33, and obtaining a prediction value of the model and a confidence interval of the model under the selected confidence; if the models are compatible, the health state of the current hydroforming equipment is synthesized through a hard voting method, and the health state is passed; otherwise, requesting the direct monitoring method to acquire the data set again for training; and recording Ni ═ N (1), N (2), N (3) ], (1, 2 and 3) belongs to the health state prediction result of the ith fault diagnosis model on the hydroforming equipment, wherein the comprehensive formula of the hard voting method is as follows:
Figure FDA0003697584450000051
7. the method for fault diagnosis and operation and maintenance in the manufacturing process of the hydroformed intelligent equipment according to claim 6, wherein S4 specifically comprises the following substeps:
s41, performing preview and judgment on possible faults according to offline learning of historical data;
and S42, when the fault is judged to occur, checking and repairing the corresponding part according to the extracted features.
8. The method for fault diagnosis and operation and maintenance in the manufacturing process of the hydroformed intelligent equipment according to claim 7, wherein S5 specifically comprises the following substeps:
s51, matching the operation and maintenance requirements of the intelligent manufacturing equipment with a scheme library in the form of texts, pictures and the like through a professional field manual acquired by a web text crawler;
s52, mapping the fault description to the knowledge graph of the intelligent manufacturing equipment to form a fault description subgraph of the scheme library;
s53, when the user inputs the operation and maintenance demand, the operation and maintenance demand is mapped to the knowledge graph to form an operation and maintenance demand sub-graph;
and S54, searching the graph by using the knowledge graph, matching the operation and maintenance appeal subgraph of the user with the fault description subgraph of the scheme library, and finally obtaining the operation and maintenance scheme.
9. The utility model provides a hydroform intelligence is equipped manufacturing process fault diagnosis and operation and maintenance system which characterized in that includes:
the hydraulic forming intelligent equipment state acquisition and prediction module acquires and fuses multivariate data based on a sensing technology, preprocesses the data, comprises missing value supplement, abnormal value elimination and standardization processing, constructs an ARIMA-LSTM combined model, and performs multi-model integrated prediction on various states;
the hydraulic forming intelligent equipment state classification module is used for performing multi-classification on states through an unsupervised learning method;
the hydraulic forming intelligent equipment fault diagnosis module carries out multi-model integrated learning through a residual error model, a Gaussian mixture clustering model and a deep learning method, evaluates the health state of the current equipment and introduces a competition fusion mechanism;
the hydraulic forming intelligent equipment fault early warning maintenance module analyzes the current state of the equipment and carries out shutdown maintenance on the equipment when a fault is predicted to occur;
the intelligent operation and maintenance module of the hydroforming intelligent equipment constructs an expert system and a knowledge graph of the intelligent equipment, and realizes quick and accurate matching of an operation and maintenance scheme.
10. The system of claim 9, comprising a data information layer, an information analysis processing layer and an application implementation layer, wherein the sensing part of the hydroforming intelligent equipment state collection and prediction module is disposed on the data information layer, the hydroforming intelligent equipment state collection and prediction module, the hydroforming intelligent equipment state classification module and the hydroforming intelligent equipment fault diagnosis module are disposed on the information analysis processing layer, and the hydroforming intelligent equipment fault early warning maintenance module and the hydroforming intelligent equipment intelligent operation and maintenance module are disposed on the application implementation layer.
CN202210679014.5A 2022-06-16 2022-06-16 Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment Active CN115096627B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210679014.5A CN115096627B (en) 2022-06-16 2022-06-16 Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210679014.5A CN115096627B (en) 2022-06-16 2022-06-16 Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment

Publications (2)

Publication Number Publication Date
CN115096627A true CN115096627A (en) 2022-09-23
CN115096627B CN115096627B (en) 2023-04-07

Family

ID=83290373

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210679014.5A Active CN115096627B (en) 2022-06-16 2022-06-16 Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment

Country Status (1)

Country Link
CN (1) CN115096627B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116484268A (en) * 2023-06-21 2023-07-25 西安黑石智能科技有限公司 Intelligent industrial equipment fault diagnosis system based on machine learning
CN116862081A (en) * 2023-09-05 2023-10-10 北京建工环境修复股份有限公司 Operation and maintenance method and system for pollution treatment equipment
CN117273553A (en) * 2023-11-22 2023-12-22 青州市华康生物科技有限公司 Production anomaly monitoring system based on syrup concentration detection

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107728602A (en) * 2017-09-28 2018-02-23 合肥工业大学 A kind of Personalized service method of hydroforming equipment failure
CN108875281A (en) * 2018-08-08 2018-11-23 佛山科学技术学院 A kind of hybrid intelligent health status diagnostic method and device based on deep learning
CN109186973A (en) * 2018-08-20 2019-01-11 东南大学 A kind of mechanical failure diagnostic method of unsupervised deep learning network
CN109800127A (en) * 2019-01-03 2019-05-24 众安信息技术服务有限公司 A kind of system fault diagnosis intelligence O&M method and system based on machine learning
CN110059601A (en) * 2019-04-10 2019-07-26 西安交通大学 A kind of multi-feature extraction and the intelligent failure diagnosis method merged
CN110488150A (en) * 2019-08-09 2019-11-22 国网河北省电力有限公司沧州供电分公司 A kind of intelligent fault diagnosis method based on more algorithm fusions
CN110701137A (en) * 2019-10-16 2020-01-17 无锡华瀚能源装备科技有限公司 Intelligent online detection and diagnosis device and method for hydraulic system of heading machine
CN111178553A (en) * 2019-12-16 2020-05-19 北京航天智造科技发展有限公司 Industrial equipment health trend analysis method and system based on ARIMA and LSTM algorithms
CN112559870A (en) * 2020-12-18 2021-03-26 北京百度网讯科技有限公司 Multi-model fusion method and device, electronic equipment and storage medium
CN112859822A (en) * 2021-04-25 2021-05-28 北京邮电大学 Equipment health analysis and fault diagnosis method and system based on artificial intelligence
US20220172076A1 (en) * 2020-11-27 2022-06-02 At&T Intellectual Property I, L.P. Prediction of network events via rule set representations of machine learning models
CN114579875A (en) * 2022-03-03 2022-06-03 中国科学院大学 Equipment fault diagnosis and maintenance knowledge recommendation system based on knowledge graph

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107728602A (en) * 2017-09-28 2018-02-23 合肥工业大学 A kind of Personalized service method of hydroforming equipment failure
CN108875281A (en) * 2018-08-08 2018-11-23 佛山科学技术学院 A kind of hybrid intelligent health status diagnostic method and device based on deep learning
CN109186973A (en) * 2018-08-20 2019-01-11 东南大学 A kind of mechanical failure diagnostic method of unsupervised deep learning network
CN109800127A (en) * 2019-01-03 2019-05-24 众安信息技术服务有限公司 A kind of system fault diagnosis intelligence O&M method and system based on machine learning
CN110059601A (en) * 2019-04-10 2019-07-26 西安交通大学 A kind of multi-feature extraction and the intelligent failure diagnosis method merged
CN110488150A (en) * 2019-08-09 2019-11-22 国网河北省电力有限公司沧州供电分公司 A kind of intelligent fault diagnosis method based on more algorithm fusions
CN110701137A (en) * 2019-10-16 2020-01-17 无锡华瀚能源装备科技有限公司 Intelligent online detection and diagnosis device and method for hydraulic system of heading machine
CN111178553A (en) * 2019-12-16 2020-05-19 北京航天智造科技发展有限公司 Industrial equipment health trend analysis method and system based on ARIMA and LSTM algorithms
US20220172076A1 (en) * 2020-11-27 2022-06-02 At&T Intellectual Property I, L.P. Prediction of network events via rule set representations of machine learning models
CN112559870A (en) * 2020-12-18 2021-03-26 北京百度网讯科技有限公司 Multi-model fusion method and device, electronic equipment and storage medium
CN112859822A (en) * 2021-04-25 2021-05-28 北京邮电大学 Equipment health analysis and fault diagnosis method and system based on artificial intelligence
CN114579875A (en) * 2022-03-03 2022-06-03 中国科学院大学 Equipment fault diagnosis and maintenance knowledge recommendation system based on knowledge graph

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
LANJUN WAN: "A novel deep convolution multi-adversarial domain adaptation model for rolling bearing fault diagnosis" *
杨忠炯;徐康;周立强;姜东身;: "压缩式垃圾车液压系统故障诊断方法研究" *
王焱玉,杨胜: "液压故障诊断专家系统的开发" *
高帆;王玉军;杨露霞;: "基于物联网和运行大数据的设备状态监测诊断" *
黄忠山;田凌;: "智能装备故障预测与健康管理系统研究" *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116484268A (en) * 2023-06-21 2023-07-25 西安黑石智能科技有限公司 Intelligent industrial equipment fault diagnosis system based on machine learning
CN116484268B (en) * 2023-06-21 2023-09-05 西安黑石智能科技有限公司 Intelligent industrial equipment fault diagnosis system based on machine learning
CN116862081A (en) * 2023-09-05 2023-10-10 北京建工环境修复股份有限公司 Operation and maintenance method and system for pollution treatment equipment
CN116862081B (en) * 2023-09-05 2023-11-21 北京建工环境修复股份有限公司 Operation and maintenance method and system for pollution treatment equipment
CN117273553A (en) * 2023-11-22 2023-12-22 青州市华康生物科技有限公司 Production anomaly monitoring system based on syrup concentration detection
CN117273553B (en) * 2023-11-22 2024-02-02 青州市华康生物科技有限公司 Production anomaly monitoring system based on syrup concentration detection

Also Published As

Publication number Publication date
CN115096627B (en) 2023-04-07

Similar Documents

Publication Publication Date Title
CN115096627B (en) Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment
JP7240691B1 (en) Data drive active power distribution network abnormal state detection method and system
CN108762228B (en) Distributed PCA-based multi-working-condition fault monitoring method
Zhou et al. Recognition of control chart patterns using fuzzy SVM with a hybrid kernel function
CN106355030B (en) A kind of fault detection method based on analytic hierarchy process (AHP) and Nearest Neighbor with Weighted Voting Decision fusion
CN106371427B (en) Industrial process Fault Classification based on analytic hierarchy process (AHP) and fuzzy Fusion
Wang et al. Data-driven mode identification and unsupervised fault detection for nonlinear multimode processes
CN109612513B (en) Online anomaly detection method for large-scale high-dimensional sensor data
CN110990461A (en) Big data analysis model algorithm model selection method and device, electronic equipment and medium
Yu et al. Identifying source (s) of out-of-control signals in multivariate manufacturing processes using selective neural network ensemble
CN113762329A (en) Method and system for constructing state prediction model of large rolling mill
CN112085252B (en) Anti-fact prediction method for set type decision effect
CN110571792A (en) Analysis and evaluation method and system for operation state of power grid regulation and control system
Shi In-process quality improvement: Concepts, methodologies, and applications
CN117393076B (en) Intelligent monitoring method and system for heat-resistant epoxy resin production process
CN110757510A (en) Method and system for predicting remaining life of robot
CN111340110A (en) Fault early warning method based on industrial process running state trend analysis
CN114015825A (en) Method for monitoring abnormal state of blast furnace heat load based on attention mechanism
CN117349583A (en) Intelligent detection method and system for low-temperature liquid storage tank
CN108470699B (en) intelligent control system of semiconductor manufacturing equipment and process
CN117688458A (en) Method and system for automatically optimizing technological parameters of injection molding die
CN116861232A (en) Air quality data anomaly detection model based on DBN-OCSVM
CN108537249A (en) A kind of industrial process data clustering method of density peaks cluster
Yu et al. A hybrid learning-based model for on-line monitoring and diagnosis of out-of-control signals in multivariate manufacturing processes
CN113551904B (en) Gear box multi-type concurrent fault diagnosis method based on hierarchical machine 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
GR01 Patent grant
GR01 Patent grant