CN116258206A - Distributed expert system based on Chinese natural language processing - Google Patents
Distributed expert system based on Chinese natural language processing Download PDFInfo
- Publication number
- CN116258206A CN116258206A CN202211600223.2A CN202211600223A CN116258206A CN 116258206 A CN116258206 A CN 116258206A CN 202211600223 A CN202211600223 A CN 202211600223A CN 116258206 A CN116258206 A CN 116258206A
- Authority
- CN
- China
- Prior art keywords
- expert system
- fault
- data
- local
- natural language
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/043—Distributed expert systems; Blackboards
-
- 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
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention relates to a distributed expert system based on Chinese natural language processing, which takes widely-required industrial equipment fault diagnosis as an object, and aims at an application scene of a distributed industrial park, and the fault diagnosis is carried out cooperatively by means of a Chinese natural language processing tool by depending on a central expert system and a local expert system; the central expert system is responsible for integrating, analyzing and processing fault data of equipment in each park, and the local expert system is responsible for carrying out field-based fault diagnosis based on a locally stored fault data basis; the local expert system data is updated by the central expert system. The invention solves the problems of low efficiency, insufficient robustness and insufficient accuracy existing in a 'engineer with bare' mode and a single independent fault diagnosis system in the past industrial fault diagnosis process by establishing a digital management platform of equipment fault data.
Description
Technical Field
The invention belongs to the technical field of information, and relates to a distributed expert system based on Chinese natural language processing.
Background
Industrial equipment is a fundamental sign of enterprise size and level of modernization, an important factor in social productivity. In recent years, with rapid improvement of social productivity and rapid rise of industries such as new energy, equipment manufacturing, and new electric power equipment, the installed amount of industrial equipment is rapidly increasing. Meanwhile, the existing equipment fault diagnosis mode gradually exposes the problems of low flexibility, low efficiency, unsatisfied accuracy and the like. The rapid diagnosis of the faults of the industrial equipment can reduce the equipment downtime, improve the production efficiency and improve the comprehensive benefit.
At present, the conventional industrial equipment fault diagnosis method mainly comprises modes of expert field equipment diagnosis, a single independent diagnosis system and the like. The expert field diagnosis mode is affected by factors such as time and space constraint, difficulty in long-term objective storage of expert knowledge, and dependence on a 'teacher with bare' mode in transmission of expert fault diagnosis knowledge, and the like, so that the efficiency is low. And a single independent expert diagnosis system lacks comprehensive utilization of equipment fault information of different industrial parks, is difficult to meet a modern distributed production scene, and the diagnosis process needs more strict text matching, so that the robustness and accuracy of the system are limited.
According to the analysis, the defects of time and space constraint, low comprehensive utilization degree of fault data, lack of consideration on modern distributed industrial scenes, poor system robustness, poor accuracy and the like exist in the conventional industrial equipment fault diagnosis process. To overcome these problems, the present invention proposes a distributed expert system based on chinese natural language processing.
Disclosure of Invention
Aiming at the technical problems, the invention provides a distributed expert system based on Chinese natural language processing. Aiming at industrial fault diagnosis application scenes, the system establishes two expert systems, namely a center and a local expert system, by means of methods and tools such as digital communication, a digital diagnosis platform, artificial intelligence Chinese natural language processing and the like; the fault diagnosis of the field industrial park field device is guided through summarizing and analyzing the fault types, fault analysis, fault solutions and the like of the industrial park device.
The technical scheme of the invention is as follows:
a distributed expert system based on Chinese natural language processing, which establishes a plurality of local expert systems and a central expert system aiming at specific scenes of fault diagnosis of industrial equipment;
the local expert system is responsible for guiding field fault diagnosis, new fault input and the like, is oriented to field operation and maintenance personnel and is deployed on a fault diagnosis field;
the central expert system is responsible for comprehensive management and coordination of fault data, standardization of fault data, statistics and analysis of fault data, updating of Chinese natural language models and the like, is oriented to a fault diagnosis system manager and is deployed in a fault data processing center.
In the system building process, the building flow of the central expert system is as follows:
11 Collecting historical fault base information of the device;
the historical fault base information includes, but is not limited to: fault equipment, equipment model, equipment number, fault components, fault type, fault occurrence time, fault cause analysis, fault solution, fault influence and the like;
12 Preprocessing and sorting the collected historical fault basic information to obtain normalized fault data, and establishing a normalized data set;
because the contents of fault information, personnel description modes and the like of different sites are often different, the fault data is normalized before a unified fault database is established;
13 Building a Chinese natural language processing tool and a Chinese natural language semantic matching pre-training model;
the Chinese natural language processing tool comprises, but is not limited to, a development environment, a Chinese corpus word segmentation tool, and a dictionary (hereinafter referred to as corpus dictionary) with Chinese corpus feature texts and computer representation symbols in one-to-one correspondence;
the Chinese natural language semantic matching pre-training model refers to a model which can be used for Chinese semantic recognition after extensive Chinese corpus unsupervised training, and can finish the 'complete shape filling' work;
14 Developing model migration training for the pre-training model obtained in the step 13) by utilizing the standardized data set established in the step 12);
the model migration training is to retrain the pre-training model in the step 13) to obtain a professional Chinese natural language semantic matching model aiming at a specific application scene;
15 Carrying out data statistical analysis based on the fault data set and constructing a knowledge graph;
aiming at a specific application scene, based on the normalized data set in the step 12), constructing a knowledge graph by using methods such as random forests, openUE, deepKE and the like or open source tools to obtain the overall distribution condition of fault data; the data statistics analysis assists in establishing a knowledge graph, and obtains the contents of a wider fault occurrence range, a larger occurrence probability, a fault type with a worse influence, fault cause analysis and the like;
16 Based on the normalized data set, chinese natural language semantic matching model and knowledge graph, building a central database and a central expert system.
In step 13), the word segmentation tool, the corpus dictionary and the pre-training model can be built for training by themselves, and also can be widely accepted by means of effects. For example, the word segmentation tools include jieba, hanLP, foolNLTK; the corpus dictionary comprises a jieba word stock, a HanLP word stock, a hundred-degree Chinese word segmentation word stock and the like; the Chinese natural language semantic matching model comprises a hundred-degree centroid model, a generating Pre-Training 2.0 Chinese model, a Bidirectional Encoder Representation from Transformers Chinese model and the like. Because the contents of word segmentation tools, corpus dictionary and the like in different languages are greatly different from each other, the distributed expert system in the invention takes Chinese as a language basis and a service basis.
In order to ensure that the Chinese natural language semantic matching model of the final specific scene has a good application effect, the pre-training model obtained in the step 13) should have a good effect of recognizing Chinese semantics (such as "complete filling" on a wide corpus); the normalized data set should be accurate and sufficient (the accurate means that the description language is concise and clear; the sufficient means that the normalized data set should contain more types of fault types, fault reasons, fault schemes and other fault basic information), and at least can characterize all fault basic information which can be collected when the distributed expert system is built.
In the system establishment process, the design of the local expert system is as follows:
21 A fault type input module is arranged for receiving a fault type to be diagnosed;
22 A fault diagnosis module is arranged for diagnosing the fault type to be diagnosed and giving a diagnosis result;
23 A fault input module is arranged for inputting the solved fault basic information, temporarily storing and centrally transmitting the fault basic information to a central expert system;
24 A local expert database is arranged, and the database is responsible for initialization and periodical update by the database of the central expert system;
25 The system is provided with a Chinese natural language semantic matching module which is used for assisting the local expert system to match the received fault type, and the Chinese natural language semantic matching module is responsible for initialization and periodical updating by the central expert system.
In the running process of the system, the data transmission and processing flow are as follows:
the local expert system is used as a fault data portal, and searches and matches the fault type to be diagnosed, which is input by equipment operation staff or other staff, in a local database based on a Chinese natural language semantic matching model; if the matching is successful, the fault has a data basis, and the local expert system feeds back information such as fault reasons, fault solutions, probability corresponding to the fault solutions and the like related to the type of the fault to be diagnosed to equipment operation and maintenance personnel to form a diagnosis result; otherwise, if the matching fails, indicating that the local database has no relevant fault data information, feeding back relevant data of the fault type to be diagnosed to operation and maintenance personnel to form a diagnosis result which needs to be determined by the personnel (such as equipment operation and maintenance personnel) and solves the problem; whether the equipment fault solving process is based on a solution provided by a local expert system or not, after the fault is actually solved on site, fault data information is input again through the local expert system, and the local expert system is responsible for recording and storing so that the follow-up central expert system synthesizes data and guides the follow-up diagnosis process; after the newly recorded fault diagnosis data in the local expert system are accumulated to a certain quantity, uniformly transmitting the fault diagnosis data to the central expert system in a message queue mode;
the central expert system regularly extracts and gathers fault diagnosis data accumulated by each local expert system from the message queue, synthesizes and processes data based on the normalized Chinese natural language semantic matching model, and stores the data in a central database. The central expert system periodically updates (iterates) the knowledge graph and the Chinese natural language semantic matching model according to the supplemented data condition, and periodically updates the fault data base and the Chinese natural language semantic matching model in the local database.
As described above, during system operation, the data interaction between the local expert system and the central expert system includes: firstly, the local expert system is responsible for accumulating new fault diagnosis data and sending the new fault diagnosis data to the central expert system; secondly, the local expert system receives fault data processed by the central expert system and updates a local database; thirdly, the Chinese natural language semantic matching model applied in the local expert system is derived from the central expert system, and the local expert system updates the corresponding model of the local expert system after receiving the model of the central expert system.
For data communication frequency requirements between the central expert system and the local expert system: (1) Because the frequency of the occurrence of the faults of the industrial equipment is not very frequent, the local expert system and the central expert system do not need to conduct data interaction in real time, and the data communication frequency between the local expert system and the central expert system can be determined according to the frequency of the occurrence of the faults according to different application scenes. Such as: for emerging park types (such as wind farms) with more equipment and higher failure occurrence frequency, more frequent data exchange frequency (such as once in two days) can be adopted; whereas for more traditional plants (e.g., petrochemical industry) lower data interaction frequencies (e.g., once a/two weeks) can be employed; (2) Or setting up data communication threshold values corresponding to the application scenes one by one according to parks of different application scenes, and performing data interaction when the number of fault data newly recorded by the local expert system reaches the threshold value. Similar to the foregoing, the emerging, less mature, may employ a smaller threshold.
For data communication tools between a central expert system and a local expert system: the data interaction between the central expert system and the local expert system is triggered, and a message queue processing method (such as RabbitMQ) can be specifically adopted as a data communication tool.
Basic tools and strategies involved in the development between the central expert system and the local expert system include, but are not limited to: platform development language determination, database type and database architecture design, database processing development tools, internet communication development tools and the like.
For the whole distributed expert system, each local expert system is an original fault data inlet, and the central expert system is responsible for comprehensively, standardizing and statistically analyzing each local fault data and recording. The knowledge graph and the Chinese natural language semantic matching model can be updated and iterated on line. The fault data and Chinese natural language semantic matching model of each local expert system are updated by the central expert system. And a closed loop iteration space is formed between the central expert system and the local expert system, and the updating is continued.
Compared with a fault diagnosis mode of common equipment, the invention has the following beneficial effects:
1. the distributed architecture design of the whole expert system synthesizes fault data of different parks and expands and analyzes, and guides the fault diagnosis process of the field device based on the whole fault distribution of the device and the actual fault condition of each park. The fault data information of the equipment is more reasonable in utilization mode and higher in utilization rate. Meanwhile, the distributed architecture design is more suitable for a modern production distributed park mode.
2. The robustness and accuracy of the system can be obviously improved by introducing the Chinese natural language identification model. By means of the Chinese natural language semantic matching model and artificial intelligence and computer computing power, automatic identification and matching of fault information are realized, and accuracy and efficiency can be greatly improved.
3. The introduction of the knowledge graph function can intuitively reflect fault data and equipment operation conditions, is beneficial to a manager to think and examine production operation conditions, guides the manager to optimize the production process, finally reduces production operation and maintenance cost and improves comprehensive benefits.
Drawings
FIG. 1 is a schematic diagram of a distributed expert system architecture according to the present invention.
FIG. 2 is a flow chart of distributed expert system data transfer in accordance with the present invention.
Fig. 3 is a schematic diagram of a knowledge graph of the present invention.
Detailed Description
As shown in FIG. 1, the invention designs a distributed expert system based on Chinese natural language processing, which comprises a plurality of local expert systems and a central expert system.
The central expert system is responsible for data comprehensive management and coordination, fault data standardization, statistics, analysis and the like, is oriented to a fault diagnosis system manager, is deployed in a fault data processing center, and has only one fault data processing center;
the local expert system is responsible for guiding field fault diagnosis, new fault recording and the like, is oriented to field operation and maintenance personnel and is deployed on a fault diagnosis field.
1. Functional design of central expert system
The method is used for data synthesis and management: the fault data of the equipment in each industrial park are reported to a central expert system through a local expert system, and the central expert system is responsible for integrating and managing all the fault data;
for normalizing fault data: for the newly generated fault information such as fault type, fault reason, fault solution and the like provided by the park, on the basis of Chinese natural language semantic matching model identification, if a central expert system does not have a relevant data basis, a system management expert is manually responsible for standardization and normalization of relevant data;
the method is used for analyzing fault data and constructing a knowledge graph: counting the occurrence frequency of faults, acquiring fault types and reasons which have wider occurrence range, larger occurrence probability and severe influence, guiding reasonable operation of the production operation and maintenance process, and reducing the occurrence times of the faults; establishing a knowledge graph of the architecture shown in fig. 3, and intuitively displaying the knowledge graph;
for updating local fault data: the method comprises the following steps: after the local new fault data is processed by the central expert system, the local data base is synchronously updated; secondly, the fault type which is obtained based on fault data analysis and has wider occurrence range, larger occurrence probability and more severe influence is required to synchronize related fault data information to a local database no matter whether each industrial park occurs or not.
2. Functional design of local expert system
For fault diagnosis: the user inputs fault data to be diagnosed, the local expert system performs fault category matching based on a Chinese natural language semantic matching model, and the fault data of the same type in the local database is extracted; if the same type of fault data exists, the fault reasons, the probabilities and the corresponding fault solutions are counted, and fault diagnosis results (including reasons, probabilities, processing schemes and the like) are output; if no fault data of the same type exist, outputting a fault diagnosis result without a relevant data base;
for fault entry: and (3) according to the actual condition of the site, after the site operation and maintenance personnel solve the fault based on the fault diagnosis result, inputting fault basic information such as fault equipment, equipment model and the like.
3. Online iterative update
Iterative updating of knowledge maps: with the migration of the device characteristics and the processing of the existing problems, the knowledge graph needs to be changed continuously to meet the changing device characteristics, that is, the central expert system updates the knowledge graph of the architecture shown in fig. 3 based on the latest fault database.
Iterative upgrade of Chinese natural language semantic matching model: along with the accumulation of equipment fault data, the model is updated on line to ensure that the model keeps higher robustness and accuracy. To avoid over-fitting the model on the equipment failure data, one possible updating strategy is to migrate-train all failure data based on the pre-trained model of step 13) of the technical scheme, instead of training for the latest failure data based on the model in the application. Iterative upgrade of models
Taking wind field fan fault diagnosis as an example, the method comprises the following practical processes:
1) Fan fault data preparation and pretreatment:
1.1 Collecting wind farm fan historical fault basic information, including but not limited to wind farm number, fan number, fault equipment, equipment model, fault type, fault occurrence time, fault cause analysis, fault solution, fault influence and other basic information;
1.2 Preprocessing normalized finishing of the collected information: the text having the same meaning as described is unified as a fixed expression text.
2) Acquiring a Chinese natural language processing tool and a Chinese natural language semantic matching pre-training model; :
2.1 Acquiring a Chinese natural language processing tool and a pre-training model: based on Python language, using the hundred-degree text model which is approved by the effect as a Chinese text matching pre-training model;
2.2 Based on the pre-trained model in the step 2.1), performing model migration training on the model aiming at a specific application scene by utilizing the collected equipment fault data. Namely: and (3) retraining the pre-trained model in the step (2.1) by utilizing the wind field fault data subjected to the normalization processing in the step (1) to obtain a model aiming at the fan fault diagnosis application scene.
3) Establishing and validating center, local and inter-two communication facilities, tools, policies:
3.1 Normal and unobstructed internet communication infrastructure between the validation center and the local area.
3.2 A central, local two-system data communication frequency is determined. Wind power belongs to an emerging fast development industry, has a wide wind field range, relates to more fans, and adopts more frequent data exchange frequency, namely once in two days.
3.3 Data interaction between the center and the local is triggered, and a message queue processing method RabbitMQ which is open in source and subjected to a large amount of industrial verification is selected as a data communication tool.
4) The basic tools and strategies involved in the development of the central and local expert systems are selected and established. Platform development language: because the system is a distributed platform, the distributed development language Golang is adopted as a development tool; the database type adopts a relational database MySQL; the database architecture design is shown in figure 1, and the center comprises an original fault data set, a normalized data set and a comprehensive analysis data set; the database processing development tool selects a widely applied third party open source package GORM; the Internet communication development tool selects a widely applied third party open source package Echo.
5) And a development center expert system application platform.
6) And developing a local expert system application platform.
6.1 Developing fault diagnosis, fault input and Chinese natural language semantic matching modules;
6.2 Data interaction with a central expert system. Setting up a timing task trigger, namely, periodically sending new fault data collected by the local expert system to the central expert system (every three days); secondly, the receiving center expert system processes the fault data and updates a local database (2 hours later than the data transmission task every three days); thirdly, receiving the Chinese natural language semantic matching model updated by the center and updating the local model (every two weeks).
7) Online iterative update
Iterative updating of knowledge maps: and periodically updating the knowledge graph of the framework shown in fig. 3 based on the latest central expert system fault database to form an image report.
Iterative upgrade of Chinese natural language semantic matching model: since the model is trained on a large number of texts, a longer period of time update, here two weeks, can be used. Retraining all fault data of the central expert system based on the pre-trained model in step 2.1).
In the fault diagnosis process of industrial equipment, due to factors such as a plurality of fault information and differences of different fault data entry personnel expression modes, the efficiency is often lower by simply recording and analyzing the fault data manually. The single independent fault diagnosis system lacks comprehensive overall arrangement among fault data of different parks, and the fault diagnosis process needs more strict text matching, so that the data utilization rate is lower. The distributed expert system based on Chinese natural language processing improves the robustness of the system and reduces the use threshold by means of a Chinese natural language processing tool; meanwhile, fault data of different parks are integrated in series, the comprehensive utilization rate of the data is higher, and the method is more suitable for the distributed parks in modern production. In addition, the design of the knowledge graph function is helpful for guiding and optimizing the production process of each park, and the production efficiency is improved.
Claims (9)
1. A distributed expert system based on Chinese natural language processing is characterized in that: based on the Chinese natural language processing tool, establishing a plurality of local expert systems and a central expert system; the local expert system is responsible for guiding field fault diagnosis and new fault recording and is deployed on a fault diagnosis field; the central expert system is responsible for data comprehensive management and coordination, fault data standardization, data statistics, knowledge graph construction, chinese natural language semantic matching model updating and local expert system data updating and is deployed in a fault data processing center;
the construction flow of the center expert system is as follows:
11 Collecting historical fault base information of the device;
12 Preprocessing and sorting the collected historical fault basic information to obtain normalized fault data, and establishing a normalized data set;
13 Building a Chinese natural language processing tool and a Chinese natural language semantic matching model pre-training model;
14 Utilizing the fault standardization data set obtained in the step 12) to perform model migration training on the pre-training model obtained in the step 13) aiming at the application scene to obtain a Chinese natural language semantic matching model suitable for the application scene;
15 Carrying out data statistical analysis based on the normalized data set and constructing a knowledge graph;
16 Based on the normalized data set, the knowledge graph and the Chinese natural language semantic matching model obtained in the steps, a central database is established, and a central expert system is established;
the local expert system is functionally designed to:
21 A fault type input module is arranged for receiving a fault type to be diagnosed;
22 A fault diagnosis module is arranged for diagnosing the fault type to be diagnosed and giving a diagnosis result;
23 A fault input module is arranged for inputting and temporarily storing the solved fault basic information and then intensively transmitting the fault basic information to a central expert system;
24 A local expert database is arranged, and the database is responsible for initialization and periodical update by a central expert system;
25 The system is provided with a Chinese natural language semantic matching module for assisting the local expert system in matching the received fault types, wherein the Chinese natural language semantic matching module is responsible for initialization and periodical updating by the central expert system.
2. The distributed expert system of claim 1, wherein: in step 11), the historical fault basic information at least comprises fault equipment, equipment model, equipment number, fault components, fault type, fault occurrence time, fault reason analysis, fault solution and fault influence.
3. The distributed expert system of claim 1, wherein: in step 13), the Chinese natural language processing tool and the Chinese natural language semantic matching model pre-training model are built by self or other tools and models which are allowed to be used are applied.
4. A distributed expert system as claimed in claim 1 or 3, wherein: the Chinese natural language semantic matching model pre-training model obtained in the step 13) has the effect of recognizing Chinese semantics; the normalized data set is capable of characterizing at least all equipment failure data that can be collected when the distributed expert system is built.
5. The distributed expert system of claim 1, wherein during system operation, the data transfer flow of the distributed expert system is as follows:
the local expert system is used as a fault data portal, and searches and matches the input fault type to be diagnosed in a local database based on a Chinese natural language semantic matching model; if the matching is successful, the fault has a data basis, and the local expert system feeds back the fault reason, the fault solution and the probability corresponding to the fault solution which are related to the type of the fault to be diagnosed, so as to form a diagnosis result; otherwise, if the matching fails, indicating that the local database has no relevant fault data information, feeding back relevant data of the fault type to be diagnosed, and forming a diagnosis result which needs to determine the reason and solve the problem by the human body; whether the equipment fault solving process is based on a solution provided by a local expert system or not, after the fault is actually solved on site, fault data information is input again through the local expert system, and the local expert system is responsible for recording and storing so that the follow-up central expert system synthesizes data and guides the follow-up diagnosis process; after the newly recorded fault diagnosis data in the local expert system are accumulated to a certain quantity, uniformly transmitting the fault diagnosis data to the central expert system in a message queue mode;
the central expert system regularly extracts and gathers fault diagnosis data accumulated by each local expert system from the message queue, synthesizes and processes data based on the normalized Chinese natural language semantic matching model, and stores the data in a central database; and the central expert system periodically updates the iterative knowledge graph and the Chinese natural language semantic matching model according to the supplemented data condition, and periodically updates the fault data base and the Chinese natural language semantic matching model in the local database.
6. The distributed expert system of claim 5, wherein the data interaction of the local expert system with the central expert system during system operation comprises: firstly, the local expert system is responsible for accumulating new fault diagnosis data and sending the new fault diagnosis data to the central expert system; secondly, the local expert system receives fault data processed by the central expert system and updates a local database; thirdly, the Chinese natural language semantic matching model applied in the local expert system is derived from the central expert system, and the local expert system updates the corresponding model of the local expert system after receiving the model of the central expert system.
7. The distributed expert system of claim 5, wherein: and determining the data communication frequency between the local expert system and the central expert system according to the occurrence frequency of the faults according to different application scenes.
8. The distributed expert system of claim 5, wherein: the data interaction between the central expert system and the local expert system is triggered, and a message queue processing method is adopted as a data communication method.
9. The distributed expert system of claim 1, wherein for an overall distributed expert system: each local expert system is an original fault data inlet, and the central expert system is responsible for comprehensively, standardizing and statistically analyzing each local fault data and recording; the knowledge graph and the Chinese natural language semantic matching model can be updated and iterated on line; fault data of each local expert system and a Chinese natural language semantic matching model are updated by the central expert system; and a closed loop iteration space is formed between the central expert system and the local expert system, and the updating is continued.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211600223.2A CN116258206A (en) | 2022-12-14 | 2022-12-14 | Distributed expert system based on Chinese natural language processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211600223.2A CN116258206A (en) | 2022-12-14 | 2022-12-14 | Distributed expert system based on Chinese natural language processing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116258206A true CN116258206A (en) | 2023-06-13 |
Family
ID=86685297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211600223.2A Pending CN116258206A (en) | 2022-12-14 | 2022-12-14 | Distributed expert system based on Chinese natural language processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116258206A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116932148A (en) * | 2023-09-19 | 2023-10-24 | 山东浪潮数据库技术有限公司 | Problem diagnosis system and method based on AI |
-
2022
- 2022-12-14 CN CN202211600223.2A patent/CN116258206A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116932148A (en) * | 2023-09-19 | 2023-10-24 | 山东浪潮数据库技术有限公司 | Problem diagnosis system and method based on AI |
| CN116932148B (en) * | 2023-09-19 | 2024-01-19 | 山东浪潮数据库技术有限公司 | Problem diagnosis system and method based on AI |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110046562B (en) | Health monitoring method and device for wind power system | |
| US11840998B2 (en) | Hydraulic turbine cavitation acoustic signal identification method based on big data machine learning | |
| CN108376151A (en) | Question classification method, device, computer equipment and storage medium | |
| CN110929918A (en) | 10kV feeder line fault prediction method based on CNN and LightGBM | |
| US20240094693A1 (en) | Prediction method of wind power output, electronic device, storage medium, and system | |
| CN112419096B (en) | Automatic circulation method of user electricity appeal work order based on NLP information extraction and few-sample self-learning | |
| CN110766212A (en) | Ultra-short-term photovoltaic power prediction method for historical data missing electric field | |
| CN111178699B (en) | Method for constructing intelligent check system for dispatching operation ticket | |
| CN113435759B (en) | Primary equipment risk intelligent assessment method based on deep learning | |
| CN111611218A (en) | Distributed abnormal log automatic identification method based on deep learning | |
| CN110110095A (en) | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network | |
| CN112818008A (en) | Intelligent diagnosis method, system, medium and electronic equipment for nuclear power debugging faults | |
| CN116258206A (en) | Distributed expert system based on Chinese natural language processing | |
| CN116910633A (en) | Power grid fault prediction method based on multi-modal knowledge mixed reasoning | |
| CN110457450A (en) | Answer generation method and relevant device based on neural network model | |
| CN115563968A (en) | Water and electricity transportation and inspection knowledge natural language artificial intelligence system and method | |
| CN116862477A (en) | Intelligent maintenance auxiliary method, system, equipment and storage medium | |
| CN114625837A (en) | Intelligent operation and maintenance method and system for railway station interlocking system | |
| US20220404778A1 (en) | Intellectual quality management method, electronic device and computer readable storage medium | |
| CN116804691B (en) | Fault monitoring method for dispatching automation equipment of power system | |
| CN109033378A (en) | A kind of application method of Zero-shot Learning in intelligent customer service system | |
| CN117409890A (en) | Transformer fault self-identification method and system based on two-way long and short-time memory | |
| CN117198446A (en) | Ward-round report automatic generation method based on medical care intercom equipment | |
| CN117235638A (en) | Police condition content multilayer classification method based on pre-training model | |
| CN116304112A (en) | Intelligent monitoring method based on big data technology |
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 |