CN112905799A - Intelligent method and system for deducing system or product quality abnormity - Google Patents
Intelligent method and system for deducing system or product quality abnormity Download PDFInfo
- Publication number
- CN112905799A CN112905799A CN201911229817.5A CN201911229817A CN112905799A CN 112905799 A CN112905799 A CN 112905799A CN 201911229817 A CN201911229817 A CN 201911229817A CN 112905799 A CN112905799 A CN 112905799A
- Authority
- CN
- China
- Prior art keywords
- data
- abnormal
- machine learning
- inference
- product quality
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000010801 machine learning Methods 0.000 claims abstract description 50
- 238000004458 analytical method Methods 0.000 claims abstract description 22
- 238000007405 data analysis Methods 0.000 claims abstract description 4
- 230000002159 abnormal effect Effects 0.000 claims description 70
- 230000005856 abnormality Effects 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 17
- 238000011058 failure modes and effects analysis Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 7
- 238000005457 optimization Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 230000019771 cognition Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000008676 import Effects 0.000 description 2
- 230000001939 inductive effect Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 230000000246 remedial effect Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
Abstract
The specification discloses an intelligent method and system for deducing system or product quality abnormity, the method runs in a client system executing failure mode and influence analysis, a machine learning algorithm is run, data is collected and imported into a knowledge base, wherein various information influencing system abnormity is recorded in a text report, an expert article and the like, then character exploration is carried out on the collected data, data which are not beneficial to establishing a model are deleted, unstructured data are processed into structured data, big data analysis is carried out aiming at the data, the relevance of the data is learned, an abnormity deduction model is established, then abnormity deduction is carried out by the abnormity deduction model, and after feedback information is received, the abnormity deduction model is verified, and the abnormity deduction model is optimized.
Description
Technical Field
The invention relates to a technology for deducing system or product quality abnormity, in particular to a method and a system for deducing the phenomenon of system or product quality abnormity by intelligent means such as data acquisition, character exploration and algorithm and the like by using a machine learning method.
Background
As technology advances, the manufacturing process of an industrial product becomes more complex with higher product complexity, and more functions and a larger number of components make a manufacturing system more complicated, so that when a problem occurs in a product and a systematic problem needs to be found, the problem of execution difficulty is caused by considering too many details.
When a system is faced with Failure, the prior art proposes a concept of Failure Mode and Effects Analysis (FMEA), which is a method for gradually identifying possible errors in the system, and can be applied to the product manufacturing process or service flow for checking the problem that may cause the system Failure. One of the existing methods is to debug various links of the system in a tabular and documentary manner, and when a problem occurs, the table lookup manner can be used to determine which link may have an error.
However, the traditional FMEA still faces a lot of pain points, when new errors or factors are generated, such error elimination mode in a form or file mode needs to be revised at any time, and the interpretation and cognition of the word semantics depend on human interpretation and cognition, so that everyone has different interpretation and cognition of the word semantics, and the interpretation and cognition are difficult to be summarized by a definite standard on a rating number, so that no standard exists and the efficiency is not high; in addition, with the higher system complexity, the complex products of multiple functions and a large number of components have too many details after the system is decomposed, and the system is complex and difficult to execute; if more complex factors cause system failure, the simultaneous action or mutual influence of multiple failure modes is difficult to analyze by using a table or a file.
Moreover, when the FMEA is applied to enterprises, the problems of insufficient analysis, incomplete measures, inaccurate risk assessment, difficult team cooperation and the like may exist in the implementation of the enterprises; the implementer may also implement as a task, which cannot generate practical benefit; the management of knowledge and risk cannot be unified along with the fluctuation of factors such as personnel and ability due to different ability levels and knowledge habits of the enterprises participating in the FMEA, so that standardized data cannot be formed, and the conventional FMEA cannot meet the standard requirements.
Disclosure of Invention
According to the embodiment disclosed in the specification, an intelligent method for deducing system or product quality abnormity implemented by a computer system is provided, and one purpose of the method is to provide a method for automatically inducing, analyzing and eliminating failure influence by using an intelligent method of machine learning in view of the operation failure of a traditional failure and influence judgment method (such as FMEA). Then, when the system receives new data, the system inputs the abnormal inference model, executes the abnormal inference, outputs an inference result, and can verify and optimize the abnormal inference model after receiving feedback information.
Preferably, the knowledge base includes text reports (which may include quality inference reports, quality management analysis, problem solution and countermeasure files, 8D files (8D problem solution), CAR files (corrective action request), expert articles, various information that the existing FMEA file affects the system and the product quality abnormality, and information of system operation and environmental information.
Further, the collected data includes failure data from a generalized analysis of fault pattern effects and criticality.
Furthermore, the step of performing character exploration on the acquired data or file description comprises the steps of screening and eliminating data which are unfavorable for establishing an abnormal inference model, processing unstructured data into structured data and the like, and can comprise the steps of performing vocabulary unification on the received data to establish a vocabulary library, selecting the data serving as a training sample, and establishing the abnormal inference model through the training sample.
Preferably, when a machine learning algorithm for running the method is selected from a plurality of machine learning algorithms, the method for selecting the machine learning algorithm includes: and respectively establishing individual abnormal inference models by a plurality of machine learning algorithms, and grading each abnormal inference model to select one of the machine learning algorithms.
Further, to verify the anomaly inference model based on the feedback information and the actual failure condition, and to modify the parameters as needed, a lookup table for inferring system anomalies can be generated.
The specification also discloses an intelligent system for deducing the quality abnormality of the system or the product, and an intelligent method for executing the system or the product, wherein the intelligent method comprises an algorithm module realized by software or matched hardware, a plurality of machine learning algorithms and a machine learning module are provided, one of the machine learning algorithms is used for analyzing data provided by a client system and learning and training the data through a neural network, a model establishing module is used for establishing an abnormal inference model according to the training result of the machine learning module, verifying each abnormal inference model established by each machine learning algorithm, selecting a better machine learning algorithm from the abnormal inference model and executing evaluation and optimization, and a database module is used for searching data for the client system and establishing a knowledge base.
Drawings
FIG. 1 illustrates a schematic diagram of an embodiment of an intelligent system architecture for inferring anomalies in system or product quality;
FIG. 2 illustrates a schematic diagram of an embodiment of an intelligent system for inferring anomalies in system or product quality;
FIG. 3 illustrates a flow chart of a text exploration applied in an intelligent method of inferring system or product quality anomalies;
FIG. 4 is a diagram illustrating an embodiment of an algorithm flow that operates for the purpose of inferring system or product quality anomalies;
FIG. 5 illustrates a flow diagram of an embodiment of an algorithm for selecting for use in an intelligent method of inferring system or product quality anomalies;
FIG. 6 illustrates a flow diagram of an embodiment of establishing an anomaly inference model in an intelligent method of inferring system or product quality anomalies; and
FIG. 7 illustrates a flow diagram for an embodiment of performing exception inference after importing data into artificial intelligence.
Detailed Description
The specification provides an intelligent method and system for deducing system or product quality abnormity, in particular to a method realized by a computer system, which is a system suitable for establishing a knowledge base and needing to predict abnormal events, and is a method for automatically inducing, analyzing and eliminating failure influence by using an intelligent method of machine learning. Such systems may have the capability to cope with common abnormal events, for example, by looking up a table to compare specific situations to obtain a solution for the abnormal event, or relying on experts in the related field to perform abnormal judgment, but often suffer from the problems of difficult judgment due to complicated systems, insufficient analysis, incomplete measures, inaccurate evaluation, and increased difficulty in handover due to too much reliance on experts. Therefore, the proposed intelligent method and system for deducing system or product quality abnormity introduces machine learning technology of Artificial Intelligence (AI) into the knowledge base, constructs the knowledge base and model capable of deducing system or product quality abnormity in advance through the capability of processing a large amount of data by the computer system, and effectively optimizes the knowledge base and revises the model according to feedback information at any time in the operation process, thereby achieving the purposes of reducing errors of human factors, executing automatic accurate deduction system or product quality abnormity and strengthening the prevention of system abnormity in advance.
In order to import a client (such as each enterprise and factory) to perform failure mode judgment, impact evaluation after failure, and failure reason searching in an intelligent manner, so as to establish a model of inference system or product quality abnormality, and further perform subsequent processing and establish a knowledge base, an artificial intelligence engine is imported into the inference system or product quality abnormality method, and a customized abnormality inference model is established by a machine learning method, which can refer to an inference system or an intelligent system architecture implementation illustration of product quality abnormality shown in fig. 1.
An inference system or an intelligent system 10 with abnormal product quality is shown, in which a plurality of artificial intelligence related function modules implemented by software programs in cooperation with hardware data processing capabilities, such as an algorithm module 101, are provided, in which a plurality of machine learning algorithms are provided, and in cooperation with a machine learning (learning) technique implemented by the machine learning module 103, one of the machine learning algorithms analyzes data provided by client systems (client one 111, client two 112, and client three 113), and performs learning and training through a neural network, and a model building module 105 builds an abnormal inference model according to training results of the machine learning module 103, and also verifies various abnormal inference models built by the machine learning algorithms. And then, different algorithms can be used for verifying the inference results, a better algorithm is selected, and the evaluation and optimization of the abnormal inference model are executed.
The intelligent system 10 for deducing system or product quality abnormity links the knowledge bases of the clients through a network or a specific mode, such as schematically shown client one 111 (knowledge base 115), client two 112 (knowledge base 116) and client three 113 (knowledge base 117), the intelligent system 10 for deducing system or product quality abnormity gathers data to the client system through the database module 107, establishes the knowledge base of the system end, and can perform the purpose of abnormity deduction such as failure mode and influence analysis and the like aiming at different clients.
FIG. 2 illustrates a schematic diagram of an embodiment of an intelligent system applying inference systems or product quality anomalies. The intelligent system for deducing system or product quality abnormity is provided with a knowledge base 20, wherein the recorded matters at least comprise expert knowledge 201, historical records 202, abnormity factors 203 and a comparison table 204, and in the intelligent method for deducing system or product quality abnormity executed by the system, the main steps comprise graphic data acquisition 21, character exploration 22, pointer scoring 23, abnormity diagnosis 24, optimization tracking 25 and the like.
The knowledge base 20 records information for system anomaly determination, such as expert knowledge 201, which refers to expert knowledge about the relevant fields of the applied system, especially knowledge about how to eliminate system anomalies, and the sources of the knowledge, such as experts inside and outside an enterprise, can be various unstructured text reports, articles, etc. imported into the system. The history record 202 records past abnormal events of the system and how to solve the abnormal events, and becomes a main data source for constructing an abnormal inference model by an intelligent system for inferring system or product quality abnormality. The exception factor 203 records the factors associated with the occurrence of a particular exception event for use in resolving an exception. The knowledge base 20 derives a system anomaly to solution look-up table 204 in which various anomaly information is recorded, including severity, anomaly frequency, and proportion of anomalies detected.
In the intelligent method for deducing system or product quality abnormality by using the knowledge base 20, the data collection 21 aims to reduce the difficulty of data collection and enhance the capability of collecting and clearing system data (such as process data), and the data sources include machine networking, MES, ERP, customer complaint and the like besides the knowledge base 20.
The step of text exploration 22 is to solve the problem of text recognition, and the objective is to reduce the burden of text work analysis on the personnel and to reduce the problem of inconsistency due to the problem of text recognition, i.e. to standardize some phrases for operation. In one embodiment, unstructured text can be structured into keywords through a Support-Vector Machine (SVM) text analysis, which facilitates the establishment of an abnormal inference model.
The step of index scoring 23 is to solve the problem that no objective scoring standard exists due to too large scoring difference caused by subjective judgment of different people, and according to an embodiment, a Fuzzy Decision System (Fuzzy Decision System), Decision Tree Analysis, bayesian network, and an FTA-SVM (Fault Decision Tree Analysis-Support Vector Machine) algorithm can be adopted to successfully score SOD (severity-occurrence rate-detection measure). The Bayesian network and the FTA-SVM algorithm are used for establishing an intelligent failure diagnosis model for training a prediction model with high accuracy and strong self-adjusting capability.
In the step of anomaly diagnosis 24, the customer complaint cause is analyzed on-line for a fast speed, and a step of improving a strategy is proposed to reinforce the capability of solving the failure.
The optimization tracking 25 is one of the technical purposes of an inference system or an intelligent system with abnormal product quality, and is used for optimizing the knowledge base 20, and the optimization target may be to automatically optimize a traditional failure mode and impact analysis (FMEA) index, optimize a Risk Priority Number (RPN) or an Action Priority (AP), and dynamically analyze Action priorities of various failure modes to continuously improve product quality.
It should be noted that the above-mentioned character exploration 22 is a method for processing text data, and the related process can refer to the character exploration process shown in fig. 3 applied to an intelligent method for deducing system or product quality abnormality.
Initially, in step S301, failure data obtained from an analysis of failure modes and criticality analysis (FMECA) is collected, wherein the FMECA is augmented with criticality analysis in addition to failure modes and impact analysis (FMEA) to highlight failure modes with higher probability and higher severity, thereby maximizing the effectiveness of their remedial actions.
Then, in step S303, data unfavorable for model building in the collected data is screened out and a data set for model training is eliminated, and in step S305, the sorted unstructured data is processed into structured data that can be used for SVM text analysis. In step S307, a part of the preprocessed data is selected as a training sample by the software program, and the remaining part is a test sample, and in step S309, a prediction model, that is, an abnormal inference model in an inference system or an intelligence of abnormal product quality is established by the training sample, and then the data is used to predict an output failure mode through the abnormal inference model.
Next, in step S311, the established abnormal inference model is verified, and one method is to verify the accuracy of the abnormal inference model by testing the data of the sample, and if necessary, the optimal model can be adjusted by modifying the parameters. Then, a relevant report related to system abnormal inference is generated.
In operating the intelligent method for inferring system or product quality anomalies, one feature is in validating various machine learning algorithms, wherein the operation of the algorithms has several major steps as shown in fig. 4 for the purpose of inferring system or product quality anomalies.
One of a plurality of algorithms provided by the system is selected (step S401), a machine learning algorithm is executed through a hardware processing circuit, meanwhile (step S403), a knowledge base of each client is imported, and through big data analysis, the association between input (such as the knowledge base, existing data and the like) and output data (such as inference results, client feedback, user feedback and the like) which are closely connected and run repeatedly is learned and obtained (step S405), and an abnormal inference model is established through learning the association between various data.
In step S407, the system receives the new data and also inputs the abnormal inference model, in step S409, performs abnormal inference, outputs an inference result (step S411), and receives feedback information (step S413), in step S415, the system verifies the abnormal inference model obtained by the machine learning algorithm selected this time according to the output result, and the verification method can verify the abnormal inference model by using historical data already recorded by the system, and can further continuously optimize the abnormal inference model by correcting. In the optimization process, various machine learning algorithms learn the abnormal conditions mentioned in the feedback information and establish the relevance with the data, thereby highlighting the known and tiny factors in the past and obtaining the unknown information related to the system abnormality in the past so as to optimize the abnormality inference model.
Wherein the machine learning algorithm process can refer to the flowchart of fig. 5 for an embodiment of selecting an algorithm to be applied in an intelligent method for deducing system or product quality abnormality.
Initially, in step S501, an inference system or an intelligent system with abnormal product quality collects client data, including a knowledge base of the client system, which may include texts, video and audio contents, and records, including various text reports (including quality inference reports, quality management analysis, Problem resolution and countermeasure files, 8D files (8D Problem solutions), CAR files (corrective action request), expert articles, various information that the existing FMEA file affects the system and the product quality, and information about system operation, environment information, and the like. Next, in step S503, a text exploration is performed on the collected content, which can refer to the above embodiments, and one of the main purposes is to perform a text exploration and vocabulary unification operation on the received data, and to establish a vocabulary library for each system.
Next, according to the results of the character exploration, a plurality of algorithms, such as algorithm one (step S505), algorithm two (step S506) and algorithm three (step S507) in the illustrated flow, are executed, and the inference model of the system anomaly, that is, the inference model one (step S511), the inference model two (step S512) and the inference model three (step S513) corresponding to each algorithm, can be established by using the machine learning technique, such as various machine learning methods corresponding to each algorithm (steps S508, S509 and S510), and by using the neural network to obtain the correlation between the system anomaly data and various input data.
Next, in step S515, the system scores the abnormal inference models, and finally selects an algorithm (step S517). According to an embodiment, the scoring criteria mainly refers to a machine learning algorithm for determining a system suitable for an inference result generated by an abnormal inference model, such as a Classification metric (Classification metrics) and a Regression metric (Regression metrics) adopted in the prior art.
When the selection of the algorithm is completed, a flowchart of an embodiment of establishing an abnormal inference model as shown in fig. 6 is started.
In step S601, data is initially received, a system-side knowledge base is built based on client data or an already existing knowledge base, and in step S603, historical data is sorted, including data that is not conducive to model building is deleted, and unstructured data is processed into structured data that can be easily analyzed. In step S605, learning relevance in data by a selected machine learning algorithm, analyzing data by the algorithm, building an abnormal inference model by mass data and algorithm learning data features, wherein a training custom machine learning model (AutoML) can be employed, which can be automatically constructed and deployed on structured data automatically for a given problem (labeled data).
Then, in step S607, the generated abnormal inference model is verified, whether the abnormal event of the system is accurately inferred is evaluated, and if the evaluation result is successful, the generated model is stored in step S609; otherwise, the model parameters are further adjusted according to the verification result, as in step S611, and the step of adjusting the model should be repeated if necessary.
After completing the creation of the abnormal inference model, as shown in fig. 7, the import of data (step S701) and the clearing of data (step S703) will be started, such as described in the above embodiments, including deleting data that is not conducive to the creation of the model, and processing unstructured data into structured data that can be easily analyzed. Then, a text preprocessing (step S705) and a data preprocessing (step S707) are performed, an application is performed, the samples are classified (step S709), an abnormality inference model is input (step S711), the steps S709 and S711 are repeated as necessary, the abnormality inference model is verified according to the feedback information and the actual failure condition, and a comparison table for finally estimating the system abnormality is generated (step S715) by modifying parameters (step S713).
According to the embodiment, the abnormal inference model established through the steps can be used for predicting the output failure mode, at the beginning of the operation of the client system, the abnormal inference model is understood for the client system, and the abnormal inference model comprises a failure item needing to define a system, a failure factor and related items of the system are confirmed, the potential failure mode can be analyzed based on each functional requirement in the system, and a database is established according to the reason and the influence result generated by the failure mode. Historical experience trainings are then gathered, including test results, analytical data, and after-market data for a particular product, to assist in potential failure mode analysis.
With respect to the evaluation of the inference model, this is possible in combination with the impact of the failure to evaluate the severity and severity level of the failure, with the frequency of occurrence being evaluated in combination with current preventive measures, and with current sounding measures to evaluate the sounding level.
And finally, optimizing the abnormal inference model, for example, taking a series of improvement measures, re-evaluating the occurrence frequency and the detection degree according to the implementation result condition of the measures, thereby reducing the overall risk, and continuously and circularly optimizing until the RPN (AP) reaches an acceptable range.
In summary, according to the embodiments of the intelligent method and system for inference system or product quality abnormality described in the specification, a method for automatically summarizing, analyzing and eliminating the influence of failure by using the intelligent method of machine learning is proposed in view of the operation failure of the conventional failure and influence judgment method, wherein the abnormal inference model is repeatedly verified and corrected by using the artificial intelligent machine learning technology in combination with the text exploration and data exploration capabilities and the knowledge established by the experience and the creative of people (such as experts), so as to achieve the goal of preventing the system abnormality in advance.
The disclosure is only a preferred embodiment of the invention and should not be taken as limiting the scope of the invention, so that the invention is not limited by the disclosure of the invention.
Claims (10)
1. An intelligent method implemented in a computer system for inferring anomalies in system or product quality, operating in a client system that performs failure mode and impact analysis, the method comprising: collecting data by a machine learning algorithm and importing the data into a knowledge base;
performing character exploration on the collected data and the knowledge base;
performing big data analysis on data subjected to character exploration, learning the relevance of the data, and establishing an abnormal inference model;
when the system receives new data, the system inputs the abnormal inference model, executes abnormal inference and outputs an inference result; and
and after receiving the feedback information, verifying the abnormal inference model obtained by the machine learning algorithm so as to optimize the abnormal inference model.
2. An intelligent method for deducing system or product quality abnormality according to claim 1 wherein said knowledge base comprises text reports, expert articles, FMEA files, various information affecting said system abnormality, information about the operation of said system and environmental information.
3. An intelligent method of inferring system or product quality anomalies as claimed in claim 2, wherein the data gathered further includes failure data derived from a generalized analysis of fault pattern effects and criticality analysis.
4. An intelligent method for inference system or product quality anomaly as claimed in claim 1, wherein in the step of performing text exploration on the acquired data, comprising:
screening and eliminating data which are unfavorable for establishing the abnormal inference model;
processing unstructured data into structured data;
unifying the vocabulary of the received data and establishing a vocabulary library; and
and selecting data serving as a training sample to establish the abnormal inference model through the training sample.
5. An intelligent method for inference system or product quality anomaly as claimed in claim 1, wherein selecting said machine learning algorithm to run said method from a plurality of machine learning algorithms is provided, said method for selecting said machine learning algorithm comprising: and respectively establishing individual abnormal inference models by the plurality of machine learning algorithms, and grading each abnormal inference model to select one of the machine learning algorithms.
6. The intelligent method as claimed in any one of claims 1 to 5, wherein a look-up table for inferring system anomalies is generated by validating the anomaly inference model based on feedback information and actual failure conditions and modifying parameters as necessary.
7. An intelligent system for inferring anomalies in system or product quality, said system comprising: an algorithm module, in which a plurality of machine learning algorithms are prepared;
the machine learning module analyzes data provided by a client system by using one of the machine learning algorithms and learns and trains through a neural network;
a model establishing module, establishing an abnormal inference model according to the training result of the machine learning module, verifying each abnormal inference model established by each machine learning algorithm, selecting a better machine learning algorithm from the abnormal inference models, and executing evaluation and optimizing the abnormal inference model; and
the system comprises a database module, a client system and a database module, wherein the database module is used for searching data from the client system by the inference system or the intelligent system with abnormal product quality and establishing a knowledge base;
wherein the inference system or the intelligent system of product quality anomaly performs an intelligent method of inference system or product quality anomaly, the method comprising:
collecting data by the selected machine learning algorithm and importing the data into the knowledge base;
performing character exploration on the collected data and the knowledge base;
performing big data analysis on data subjected to character exploration, learning the relevance of the data, and establishing an abnormal inference model;
when the system receives new data, the system inputs the abnormal inference model, executes abnormal inference and outputs an inference result; and
and after receiving the feedback information, verifying the abnormal inference model obtained by the machine learning algorithm so as to optimize the abnormal inference model.
8. An inference system or an intelligent system of product quality anomaly according to claim 7, wherein said knowledge base records items including at least expert knowledge, history, anomaly factors and a look-up table.
9. An inference system or an intelligent system of product quality anomaly according to claim 8, wherein the collected data further includes failure data from a generalized analysis of failure mode effects and criticality analysis.
10. The system for intelligently deducing a quality abnormality of a product according to any one of claims 7 to 9 wherein in the intelligent method for deducing a quality abnormality of a product, the step of performing a text exploration on the acquired data comprises:
screening and eliminating data which are unfavorable for establishing the abnormal inference model;
processing unstructured data into structured data;
unifying the vocabulary of the received data and establishing a vocabulary library; and
and selecting data serving as a training sample to establish the abnormal inference model through the training sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911229817.5A CN112905799A (en) | 2019-12-04 | 2019-12-04 | Intelligent method and system for deducing system or product quality abnormity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911229817.5A CN112905799A (en) | 2019-12-04 | 2019-12-04 | Intelligent method and system for deducing system or product quality abnormity |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112905799A true CN112905799A (en) | 2021-06-04 |
Family
ID=76110718
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911229817.5A Pending CN112905799A (en) | 2019-12-04 | 2019-12-04 | Intelligent method and system for deducing system or product quality abnormity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112905799A (en) |
-
2019
- 2019-12-04 CN CN201911229817.5A patent/CN112905799A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5734961B2 (en) | Apparatus and method for estimating model quality and adapting model in multivariable process control | |
| CN110335168B (en) | Method and system for optimizing power utilization information acquisition terminal fault prediction model based on GRU | |
| US7562054B2 (en) | Method and apparatus for automated feature selection | |
| CN112906764B (en) | Communication safety equipment intelligent diagnosis method and system based on improved BP neural network | |
| KR20190110084A (en) | Esg based enterprise assessment device and operating method thereof | |
| US11580425B2 (en) | Managing defects in a model training pipeline using synthetic data sets associated with defect types | |
| CN111340269B (en) | Real-time optimization method for process industrial process | |
| Vogel-Heuser et al. | Operator knowledge inclusion in data-mining approaches for product quality assurance using cause-effect graphs | |
| Lyu et al. | A data-driven approach for identifying possible manufacturing processes and production parameters that cause product defects: A thin-film filter company case study | |
| Nagahara et al. | Toward data-driven production simulation modeling: dispatching rule identification by machine learning techniques | |
| CN109523030B (en) | Telemetering parameter abnormity monitoring system based on machine learning | |
| CN114118982A (en) | Cloud computing and edge computing combined production control system | |
| TWM592123U (en) | Intelligent system for inferring system or product quality abnormality | |
| TWI761719B (en) | Intelligent method for inferring system or product quality abnormality and system thereof | |
| CN116705272A (en) | Comprehensive evaluation method for equipment health state based on multidimensional diagnosis | |
| CN116308295A (en) | Industrial production data management method and system | |
| CN116361191A (en) | Software compatibility processing method based on artificial intelligence | |
| CN108363738B (en) | Recommendation method for industrial equipment data analysis algorithm | |
| CN112905799A (en) | Intelligent method and system for deducing system or product quality abnormity | |
| JPH09233700A (en) | Method of evaluating reliability on estimation of day maximum demand power | |
| CN114970674A (en) | Time sequence data concept drift adaptation method based on relevance alignment | |
| Haug et al. | Standardized Evaluation of Machine Learning Methods for Evolving Data Streams | |
| CN114139589A (en) | Fault diagnosis method, device, equipment and computer readable storage medium | |
| Brasen et al. | Comparison between data maturity and maintenance strategy: A case sutdy | |
| Turkoglu et al. | Application of data mining in failure estimation of cold forging machines: An industrial research |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20221010 Address after: Chinese Taiwan New Taipei City Applicant after: DATA SYSTEMS CONSULTING CO.,LTD. Address before: TaiWan, China Applicant before: Zhilue Information Integration Co.,Ltd. |