CN107862375A - A kind of two stage equipment fault diagnosis method - Google Patents
A kind of two stage equipment fault diagnosis method Download PDFInfo
- Publication number
- CN107862375A CN107862375A CN201711033505.8A CN201711033505A CN107862375A CN 107862375 A CN107862375 A CN 107862375A CN 201711033505 A CN201711033505 A CN 201711033505A CN 107862375 A CN107862375 A CN 107862375A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- model
- case
- equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of two stage equipment fault diagnosis method, it is related to Diagnosis Technique field.The present invention can be according to different response times, the requirement of computing capability, fault diagnosis type details, the step of adjusting forecast model, the dependence to maintenance personal's human factor can be reduced, improve the efficiency of equipment fault diagnosis, important reference is provided to the judgement of equipment fault for maintenance personal, can be played a significant role in the equipment fault type diagnostic under adapting to different condition.
Description
Technical Field
The invention relates to the technical field of equipment fault diagnosis, in particular to a two-stage equipment fault diagnosis method.
Background
In the whole process of equipment maintenance and guarantee, the relationship between the fault phenomenon (or fault symptom) and the fault reason (or fault unit) of the equipment is complex, and the equipment maintenance and guarantee method has the characteristics of randomness, uncertainty and the like.
The traditional fault diagnosis method has the defects of low fault diagnosis efficiency, high cost and the like due to the lack of reasoning ability and autonomous learning ability under inaccurate conditions.
Early fault diagnosis methods rely heavily on the judgment of field experts or experienced maintenance personnel, resulting in a very subjective and costly process of diagnosing faults.
In a general prediction method, all the acquisition time sequence factors of the feature sequences are ignored, so that time information is lost, and the accuracy and the rationality of prediction are greatly reduced.
Most fault diagnosis methods treat the fault type and the no fault as data of the same layer, however, in the real situation, the normal situation is far more than the fault situation. Processing unbalanced data together will cause an over-fitting phenomenon to occur, making the failure prediction effect unreliable.
Disclosure of Invention
Technical problem to be solved
The technical problem to be solved by the invention is as follows: how to design a fault diagnosis method aiming at equipment maintenance guarantee, and the accuracy and the interpretability of prediction are improved.
(II) technical scheme
In order to solve the technical problem, the invention provides a two-stage equipment fault diagnosis method, which comprises the following steps:
step 1, data processing: according to the data collected by the sensor, cleaning and unifying the data;
step 2, fault diagnosis based on case search: searching the acquired data in a case base established in advance through a K neighbor method, returning corresponding fault type information to the searched case based on a preset threshold value, ending, and continuing to execute the step 3 for the case which is not searched;
step 3, fault judgment based on time characteristics: calculating the obtained equipment state characteristic data according to the generalized autoregressive conditional variance model, and judging whether the equipment fails or not according to the value; if judging that the fault exists, continuing to execute the step 4, otherwise, returning to the information that the fault does not exist and ending;
step 4, a deep neural network fault prediction step based on the hierarchy: according to the characteristics of the diagnosis object, decomposing the system, establishing a proper hierarchical classification model, and calculating by using the hierarchical classification model to obtain the corresponding fault type.
Preferably, step 1 specifically comprises:
step 11: and (3) processing missing data by adopting a mean value-based data completion method: storing 5 pieces of data before and after the current time, and taking the average value as a filling basis of the data;
step 12: for redundant data, a sequence similarity method is adopted to eliminate the redundant data: the sequence similarity adopts the sum of Euclidean distances between the characteristic vectors as a measurement standard, a threshold value is set, data larger than the threshold value is judged to be redundant data to be discarded, and a sequence similarity calculation formula is as follows:
wherein, a f Representing the f-th received feature vector, a e Denotes a f The e-th, sim (a) of the previously received 10 feature vectors f ,a e ) Denotes a f And a e The Euclidean distance between;
step 13: discretizing the data by a 0-1 value method: an average value is calculated for each dimension of all the collected data as a threshold, data greater than or equal to the threshold is set as 1, and data smaller than the threshold is set as 0.
Preferably, step 2 specifically comprises:
step 21: and (3) reasoning steps: inputting basic information of current equipment, searching similar cases in a case library, giving fault diagnosis results and solving measures if the similar cases exist, switching to a deep neural network model based on a time sequence if the similar cases do not exist, searching the cases by adopting a searching mode based on K-neighbor matching, and searching the cases by adopting a searching mode based on the K-neighbor matching specifically comprises the following steps:
each case containing m kinds of features, fault case C i (i =1,2.., n) may be represented by an m-dimensional vector: a. The i ＝(a i1 ，a i2 ，...，a im )，a ij (j =1,2,.., m) is fault case C i The value of the jth feature of (1);
the similarity between cases is defined as:
wherein 0 is less than or equal to Sim (C) i ，C j )≤1；ω k Represents the weight of the kth feature in the case feature vector, an
Preferably, step 3 specifically comprises:
step 31: aiming at the characteristic of time serialization of the acquired equipment state information sequence, modeling the data characteristic vector fed back by the equipment according to the time sequence by adopting a generalized autoregressive conditional variance model GARCH, and judging whether the equipment fails according to the obtained calculation result;
the modeling of the data characteristic vector fed back by the equipment according to the time sequence by adopting the generalized autoregressive conditional covariance model GARCH specifically comprises the following steps:
time series X t ：
X t ＝E{X t |Ψ t-1 }+ε t (3)
Wherein psi t-1 Represents all time series X obtained at time t-1 1 ,…,X t-1 ，ε t Representing the residual error, establishing a description equation for the residual error:
in the formula: z t Is a random variable with a mean value of zero and a variance of 1; p and q are respectively the orders of the model; alpha is alpha i And beta j As a parameter to be estimated of the model, for making the conditional variance h t >, 0 requires alpha i And beta j Are all greater than 0, and alpha is used to make the model wide and smooth i And beta j The conditions also need to be satisfied:
∑ i α i +∑ j β j <1 (6)
in the GARCH model, the parameter alpha of the conditional variance is determined by adopting the maximum likelihood principle 0 ，α 1 ，β 1 Make an estimate if { X } 1 ,X 2 ,…X T Is the signal produced by the GARCH model, then the likelihood function is defined by the equation:
where h is t The logarithm of the above formula is taken to obtain a log-likelihood function which is obtained by a recursion method as follows:
wherein: x = (X) 1 ,…,X j ) T ，h＝(h 1 ,…,h j ) T The limiting condition is formula (6), and the parameter alpha of the model 0 ，α 1 ，β 1 Solving by a maximization formula (9);
judging whether the equipment fails or not according to the result obtained by the model; if no fault is diagnosed, the information of no fault is directly returned, and if the fault is diagnosed, the step 4 is carried out.
Preferably, in step 4, a neural network is used to construct a classification model corresponding to the fault classification level, each artificial neural network uses a three-layer BP network, and the result obtained by calculation using the hierarchical neural network matches the specific type of the fault.
(III) advantageous effects
The invention provides a two-stage intelligent fault diagnosis method for equipment maintenance support. On the basis of case-based search, a time-series-based fault judgment method is introduced, and fault type analysis of a multi-level neural network is combined. The invention fully utilizes the time characteristics of the equipment state information and separately judges whether the equipment fails or not and predicts the specific types of the faults, so that the data can effectively avoid the condition that the fault data and the non-fault data are unbalanced, and the accuracy and the interpretability of the prediction are improved. The prediction method based on the time sequence is mainly used in the financial field, and is applied to the field of equipment fault diagnosis for the first time. In addition, the multi-step self-adaptive prediction method can also effectively make adaptation according to the program running environment, and is well suitable for various hardware environments.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of the method of the present invention for building a hierarchical fault diagnosis model.
Detailed Description
In order to make the objects, contents, and advantages of the present invention clearer, the following detailed description of the embodiments of the present invention will be made in conjunction with the accompanying drawings and examples.
As shown in fig. 1, the two-stage device failure diagnosis method of the present invention includes the steps of:
step 1, data processing: according to data collected by the sensor, cleaning and unifying the data;
step 2, fault diagnosis based on case search: and searching the acquired data in a fault case library established in advance by a K neighbor method. Based on a preset threshold value, for the searched cases, returning corresponding fault type information and then ending, and for the cases which are not searched, continuing to execute the step 3;
step 3, fault judgment based on time characteristics: calculating the obtained equipment state characteristic data according to a generalized autoregressive conditional variance model, and judging whether the equipment fails or not according to the value; if judging that the fault exists, continuing to execute the step 4, otherwise, returning the information that the fault does not exist, and ending;
step 4, neural network fault prediction based on hierarchy: according to the characteristics of the diagnosis object, decomposing the system, establishing a proper comprehensive hierarchical classification model, and calculating by using the model to obtain the corresponding fault type.
The data acquisition step 1 includes:
step 11: and (3) processing missing data by adopting a mean value-based data completion method: and storing 5 pieces of data before and after the current time, and taking the average value as a filling basis of the data. The step makes up for the data loss on the basis of saving space and calculating time.
Step 12: for redundant data, a sequence similarity method is implemented to eliminate the redundant data: the sequence similarity adopts the sum of Euclidean distances between the characteristic vectors as a measurement standard, a threshold value is set according to experience, and data larger than the threshold value is judged to be discarded. The sequence similarity calculation formula is as follows:
wherein, a f Representing the f-th received feature vector, a e Denotes a f The e-th of the previously received 10 feature vectors. Sim (a) f ,a e ) Denotes a f And a e The euclidean distance between them.
Step 13: discretizing the data by a 0-1 value method: and calculating an average value for each dimension of all the collected data to serve as a threshold, setting the data which are greater than or equal to the threshold to be 1, and setting the data which are smaller than the threshold to be 0.
The step 2 of fault diagnosis based on case search comprises the following steps:
step 21: and (5) reasoning steps. Inputting basic information of current equipment, searching similar cases in a case library, giving fault diagnosis conclusions and solving measures if the similar cases exist, and transferring the similar cases into a time sequence-based deep neural network model if the similar cases do not exist. The case is searched by adopting a KNN (K-nearest neighbor) matching search mode. The KNN method is introduced as follows:
each case containing m kinds of features, fault case C i = (i =1,2.., n) can be represented by an m-dimensional vector: a. The i ＝(a i1 ，a i2 ，...，a im )，a ij (j =1,2,.. M) is fault case C i The value of the jth feature of (1).
The similarity between cases is defined as:
wherein 0 is less than or equal to Sim (C) i ，C j )≤1；ω k Represents the weight of the kth feature in the case feature vector, an
The time characteristic-based fault determination step 3 includes:
step 31: for the characteristic of time-series of the acquired device state information sequence, a generalized autoregressive conditional heteroscedasticity model (abbreviated as GARCH model) is implemented. In the invention, GARCH modeling is carried out on the data characteristic vector fed back by the equipment according to the time sequence, and whether the fault occurs is judged according to the obtained calculation result.
The generalized autoregressive conditional variance model (GARCH) is briefly described below.
Time series X t ：
X t ＝E{X t |Ψ t-1 }+ε t (3)
Wherein psi t-1 Represents all time series X obtained at time t-1 1 ,…,X t-1 ，ε t Representing the residual. And establishing a description equation for the residual error:
in the formula: z t Is a random variable with a mean value of zero and a variance of 1; p and q are the orders of the model respectively; alpha is alpha i And beta j Is the parameter to be estimated of the model. To make the conditional variance h t >, 0 requires alpha i And beta j Are both greater than 0. While in order to make the model broad and smooth, α i And beta j The conditions also need to be satisfied:
∑ i α i +∑ j β j <1 (6)
in the GARCH model of the invention, the parameter alpha of the conditional variance is determined by adopting the maximum likelihood principle 0 ，α 1 ，β 1 Make an estimate if { X } 1 ,X 2 ,…X T Is a signal produced by the GARCH model, then the likelihood function is defined by the formula:
where h is t The logarithm of the above formula is taken to obtain a log-likelihood function which is obtained by a recursion method as follows:
wherein: x = (X) 1 ,…,X j ) T ，h＝(h 1 ,…,h j ) T The limiting condition is equation (6), parameter α of the model 0 ，α 1 ，β 1 The maximum is obtained by the maximum formula (9).
And judging whether the equipment fails or not according to the result obtained by the model. If no fault is diagnosed, the information of no fault is directly returned, and if the fault is diagnosed, the step 4 is carried out.
The step 4 of predicting the fault of the deep neural network based on the hierarchy comprises the following steps:
step 41: the hierarchical neural network predicts a specific type of failure. The fault of the equipment establishes a tree type hierarchical model according to the master-slave hierarchy, and the schematic diagram of the hierarchical model is shown in figure 2. Implementing a neural network may construct a classification model corresponding to a hierarchy of fault classifications. Each artificial neural network adopts a three-layer BP network. The results calculated by the hierarchical neural network are matched with the specific type of the fault.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (5)
1. A two-stage equipment fault diagnosis method is characterized by comprising the following steps:
step 1, data processing: according to the data collected by the sensor, cleaning and unifying the data;
step 2, fault diagnosis based on case search: searching the acquired data in a case base established in advance through a K neighbor method, returning corresponding fault type information to the searched case based on a preset threshold value, ending, and continuing to execute the step 3 for the case which is not searched;
step 3, fault judgment based on time characteristics: calculating the obtained equipment state characteristic data according to a generalized autoregressive conditional variance model, and judging whether the equipment fails or not according to the value; if judging that the fault exists, continuing to execute the step 4, otherwise, returning the information that the fault does not exist, and ending;
step 4, deep neural network fault prediction based on hierarchy: according to the characteristics of the diagnosis object, decomposing the system, establishing a proper hierarchical classification model, and calculating by using the hierarchical classification model to obtain the corresponding fault type.
2. The method according to claim 1, wherein step 1 comprises in particular:
step 11: and (3) processing missing data by adopting a mean value-based data completion method: storing 5 pieces of data before and after the current time, and taking the average value as a filling basis of the data;
step 12: for redundant data, a sequence similarity method is adopted to eliminate the redundant data: the sequence similarity adopts the sum of Euclidean distances between the characteristic vectors as a measurement standard, a threshold value is set, data larger than the threshold value is judged to be redundant data to be discarded, and a sequence similarity calculation formula is as follows:
wherein, a f Representing the f-th received feature vector, a e Denotes a f The e-th, sim (a) of the previously received 10 feature vectors f ,a e ) Denotes a f And a e The Euclidean distance between;
step 13: discretizing the data by a 0-1 value method: and calculating an average value for each dimension of all the collected data to serve as a threshold, setting the data which are greater than or equal to the threshold to be 1, and setting the data which are smaller than the threshold to be 0.
3. The method according to claim 2, wherein step 2 specifically comprises:
step 21: the inference step comprises: inputting basic information of current equipment, searching similar cases in a case library, giving fault diagnosis results and solving measures if the similar cases exist, switching to a deep neural network model based on a time sequence if the similar cases do not exist, searching the cases by adopting a searching mode based on K-neighbor matching, and searching the cases by adopting a searching mode based on the K-neighbor matching specifically comprises the following steps:
each case containing m kinds of features, fault case C i (i =1,2.., n) may be represented by an m-dimensional vector: a. The i ＝(a i1 ，a i2 ，...，a im )，a ij (j =1,2,.., m) is fault case C i The value of the jth feature of (a);
the similarity between cases is defined as:
wherein 0 is less than or equal to Sim (C) i ，C j )≤1；ω k Represents the weight of the kth feature in the case feature vector, an
4. The method according to claim 3, wherein step 3 specifically comprises:
step 31: aiming at the characteristic of time serialization of the acquired equipment state information sequence, a generalized autoregressive conditional heteroscedasticity model GARCH is adopted to model a data feature vector fed back by equipment according to the time sequence, and whether a fault occurs is judged according to an obtained calculation result;
the modeling of the data characteristic vector fed back by the equipment according to the time sequence by adopting the generalized autoregressive conditional variance model GARCH specifically comprises the following steps:
time series X t ：
X t ＝E{X t |Ψ t-1 }+ε t (3)
Wherein psi t-1 Representing the acquisition at time t-1All time sequences X obtained 1 ,…,X t-1 ，ε t Representing residual errors, establishing a description equation for the residual errors:
in the formula: z t Is a random variable with a mean value of zero and a variance of 1; p and q are the orders of the model respectively; alpha is alpha i And beta j As a parameter to be estimated of the model, for making the conditional variance h t >, 0 requires alpha i And beta j Are all greater than 0, and alpha is used to make the model wide and smooth i And beta j The conditions also need to be satisfied:
∑ i α i +∑ j β j <1 (6)
in the GARCH model, the parameter alpha of the conditional variance is determined by adopting the maximum likelihood principle 0 ，α 1 ，β 1 Make an estimate if { X } 1 ,X 2 ,…X T Is a signal produced by the GARCH model, then the likelihood function is defined by the formula:
here h is t The logarithm of the above formula is taken to obtain a log-likelihood function which is obtained by a recursion method as follows:
wherein: x = (X) 1 ,…,X j ) T ，h＝(h 1 ,…,h j ) T The limiting condition is formula (6), and the parameter alpha of the model 0 ，α 1 ，β 1 By maximumSolving the formula (9);
judging whether the equipment fails or not according to the result obtained by the model; if no fault is diagnosed, the information of no fault is directly returned, and if the fault is diagnosed, the step 4 is carried out.
5. The method of claim 4, wherein in step 4, a neural network is used to construct a classification model corresponding to the classification level of the fault, each artificial neural network is a three-layer BP network, and the result calculated by the hierarchical neural network matches the specific type of the fault.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711033505.8A CN107862375A (en) | 2017-10-30 | 2017-10-30 | A kind of two stage equipment fault diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711033505.8A CN107862375A (en) | 2017-10-30 | 2017-10-30 | A kind of two stage equipment fault diagnosis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107862375A true CN107862375A (en) | 2018-03-30 |
Family
ID=61697478
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711033505.8A Pending CN107862375A (en) | 2017-10-30 | 2017-10-30 | A kind of two stage equipment fault diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107862375A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830306A (en) * | 2018-05-30 | 2018-11-16 | 东软集团股份有限公司 | The workflow method for diagnosing faults and device, medium and electronic equipment of business datum |
| CN109656818A (en) * | 2018-12-05 | 2019-04-19 | 北京计算机技术及应用研究所 | A kind of denseness system failure prediction method |
| CN109871975A (en) * | 2018-11-28 | 2019-06-11 | 国网浙江省电力有限公司台州供电公司 | Breakdown repair handling duration prediction technique based on data mining |
| CN114496201A (en) * | 2022-01-25 | 2022-05-13 | 山东浪潮工业互联网产业股份有限公司 | Medical equipment maintenance method, equipment and medium based on industrial internet identification |
| WO2023098372A1 (en) * | 2021-11-30 | 2023-06-08 | 无锡汇田水务科技有限公司 | Self-diagnosis method and non-negative pressure additive pressure water supply device |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040010733A1 (en) * | 2002-07-10 | 2004-01-15 | Veena S. | System and method for fault identification in an electronic system based on context-based alarm analysis |
| EP1703449A1 (en) * | 2005-03-18 | 2006-09-20 | BRITISH TELECOMMUNICATIONS public limited company | Fault diagnostics |
| CN101634605A (en) * | 2009-04-10 | 2010-01-27 | 北京工业大学 | Intelligent gearbox fault diagnosis method based on mixed inference and neural network |
| CN102043900A (en) * | 2010-11-24 | 2011-05-04 | 河海大学 | Failure prediction method of rod pumping system based on indicator diagram |
| CN102208028A (en) * | 2011-05-31 | 2011-10-05 | 北京航空航天大学 | Fault predicting and diagnosing method suitable for dynamic complex system |
| CN102340811A (en) * | 2011-11-02 | 2012-02-01 | 中国农业大学 | Method for carrying out fault diagnosis on wireless sensor networks |
| WO2013059039A1 (en) * | 2011-10-20 | 2013-04-25 | Nalco Company | Method for early warning chatter detection and asset protection management |
| US20140277910A1 (en) * | 2013-03-14 | 2014-09-18 | The Goodyear Tire & Rubber Company | Predictive peer-based tire health monitoring |
| CN104281130A (en) * | 2014-09-22 | 2015-01-14 | 国家电网公司 | Hydroelectric equipment monitoring and fault diagnosis system based on big data technology |
| CN105629962A (en) * | 2016-03-03 | 2016-06-01 | 中国铁路总公司 | Failure diagnosis method for high-speed railway train control equipment radio block center (RBC) system |
-
2017
- 2017-10-30 CN CN201711033505.8A patent/CN107862375A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040010733A1 (en) * | 2002-07-10 | 2004-01-15 | Veena S. | System and method for fault identification in an electronic system based on context-based alarm analysis |
| EP1703449A1 (en) * | 2005-03-18 | 2006-09-20 | BRITISH TELECOMMUNICATIONS public limited company | Fault diagnostics |
| CN101634605A (en) * | 2009-04-10 | 2010-01-27 | 北京工业大学 | Intelligent gearbox fault diagnosis method based on mixed inference and neural network |
| CN102043900A (en) * | 2010-11-24 | 2011-05-04 | 河海大学 | Failure prediction method of rod pumping system based on indicator diagram |
| CN102208028A (en) * | 2011-05-31 | 2011-10-05 | 北京航空航天大学 | Fault predicting and diagnosing method suitable for dynamic complex system |
| WO2013059039A1 (en) * | 2011-10-20 | 2013-04-25 | Nalco Company | Method for early warning chatter detection and asset protection management |
| CN102340811A (en) * | 2011-11-02 | 2012-02-01 | 中国农业大学 | Method for carrying out fault diagnosis on wireless sensor networks |
| US20140277910A1 (en) * | 2013-03-14 | 2014-09-18 | The Goodyear Tire & Rubber Company | Predictive peer-based tire health monitoring |
| CN104281130A (en) * | 2014-09-22 | 2015-01-14 | 国家电网公司 | Hydroelectric equipment monitoring and fault diagnosis system based on big data technology |
| CN105629962A (en) * | 2016-03-03 | 2016-06-01 | 中国铁路总公司 | Failure diagnosis method for high-speed railway train control equipment radio block center (RBC) system |
Non-Patent Citations (10)
| Title |
|---|
| SAROJ KR. BISWAS ET AL.: "Hybrid expert system using case based reasoning and neural network for classification", 《BIOLOGICALLY INSPIRED COGNITIVE ARCHITECTURE》 * |
| 侯玉梅等: "基于案例推理法研究综述", 《燕山大学学报（哲学社会科学版）》 * |
| 张晓莉等: "诊断推理中人工神经网络与基于案例推理的结合", 《上海铁道大学学报》 * |
| 方展画: "《中国都市教育竞争力研究》", 31 August 2011, 教育科学出版社 * |
| 李锋刚: "《基于案例推理的智能决策技术》", 31 January 2011, 安徽大学出版社 * |
| 梁华: "有杆抽油系统故障递阶诊断的故障识别研究", 《西南石油大学学报（自然科学版）》 * |
| 汤天浩等: "层次分类诊断模型的多重结构神经网络实现与应用", 《上海海运学院学报》 * |
| 潘红宇: "《金融时间序列模型》", 30 November 2008, 对外经济贸易大学出版社 * |
| 董磊等: "基于模型和案例推理的混合故障诊断方法", 《系统工程与电子技术》 * |
| 陆彦婷等: "层次分类方法综述", 《模式识别与人工智能》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108830306A (en) * | 2018-05-30 | 2018-11-16 | 东软集团股份有限公司 | The workflow method for diagnosing faults and device, medium and electronic equipment of business datum |
| CN109871975A (en) * | 2018-11-28 | 2019-06-11 | 国网浙江省电力有限公司台州供电公司 | Breakdown repair handling duration prediction technique based on data mining |
| CN109871975B (en) * | 2018-11-28 | 2021-04-09 | 国网浙江省电力有限公司台州供电公司 | Data mining-based fault first-aid repair processing duration prediction method |
| CN109656818A (en) * | 2018-12-05 | 2019-04-19 | 北京计算机技术及应用研究所 | A kind of denseness system failure prediction method |
| CN109656818B (en) * | 2018-12-05 | 2022-02-15 | 北京计算机技术及应用研究所 | Fault prediction method for software intensive system |
| WO2023098372A1 (en) * | 2021-11-30 | 2023-06-08 | 无锡汇田水务科技有限公司 | Self-diagnosis method and non-negative pressure additive pressure water supply device |
| CN114496201A (en) * | 2022-01-25 | 2022-05-13 | 山东浪潮工业互联网产业股份有限公司 | Medical equipment maintenance method, equipment and medium based on industrial internet identification |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110221225B (en) | Spacecraft lithium ion battery cycle life prediction method | |
| CN111914873B (en) | Two-stage cloud server unsupervised anomaly prediction method | |
| CN107862375A (en) | A kind of two stage equipment fault diagnosis method | |
| Haq et al. | Heart disease prediction system using model of machine learning and sequential backward selection algorithm for features selection | |
| CN111694879B (en) | Multielement time sequence abnormal mode prediction method and data acquisition monitoring device | |
| Nnamoko et al. | Predicting diabetes onset: an ensemble supervised learning approach | |
| CN111275288A (en) | XGboost-based multi-dimensional data anomaly detection method and device | |
| WO2019080367A1 (en) | Method for evaluating health status of mechanical device | |
| US20130204810A1 (en) | Discriminant model learning device, method and program | |
| CN108846512A (en) | Based on the water quality prediction method preferentially classified | |
| CN102521534B (en) | Intrusion detection method based on crude entropy property reduction | |
| CN111832825A (en) | Wind power prediction method and system integrating long-term and short-term memory network and extreme learning machine | |
| CN108803555B (en) | Sub-health online identification and diagnosis method based on performance monitoring data | |
| US20220245405A1 (en) | Deterioration suppression program, deterioration suppression method, and non-transitory computer-readable storage medium | |
| Fu et al. | MCA-DTCN: A novel dual-task temporal convolutional network with multi-channel attention for first prediction time detection and remaining useful life prediction | |
| WO2023231374A1 (en) | Semi-supervised fault detection and analysis method and apparatus for mechanical device, terminal, and medium | |
| Sadr et al. | An anomaly detection method for satellites using Monte Carlo dropout | |
| Gond et al. | A survey of machine learning-based approaches for missing value imputation | |
| CN115983087A (en) | Method for detecting time sequence data abnormity by combining attention mechanism and LSTM and terminal | |
| CN108282360B (en) | Fault detection method for long-term and short-term prediction fusion | |
| Tembhekar et al. | Cross-Domain Applications of MLOps: From Healthcare to Finance | |
| CN113674862A (en) | Acute renal function injury onset prediction method based on machine learning | |
| CN113780432B (en) | Intelligent detection method for operation and maintenance abnormity of network information system based on reinforcement learning | |
| CN114254738A (en) | Double-layer evolvable dynamic graph convolution neural network model construction method and application | |
| CN113884807A (en) | Power distribution network fault prediction method based on random forest and multi-layer architecture clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180330 |