CN111931936A - Equipment fault diagnosis method based on collaborative case reasoning and semantic model reasoning - Google Patents

Abstract

The invention discloses an equipment fault diagnosis method for collaborative case reasoning and semantic model reasoning. Fault diagnosis ontology model; S3 generates the corresponding SWRL rules based on the knowledge obtained in the ontology model, combined with expert experience, and forms a fault diagnosis rule base after conflict detection of the generated SWRL rules. S4 performs fault detection according to the fault diagnosis ontology model, rule base and case base constructed above. Based on the combination of CBR and RBR, the present invention integrates the knowledge extracted by fuzzy logic and FMEA analysis into the construction of the ontology model, thereby improving the integrity of the ontology model, and the definition of uncertain knowledge is more reasonable; Based on shallow knowledge and deep knowledge, diagnostic rules are constructed, which improves the integrity and accuracy of the rule base, thereby improving the reasoning of the diagnostic framework.

Description

A Device Fault Diagnosis Method Based on Collaborative Case Reasoning and Semantic Model Reasoning

technical field

The invention relates to an equipment fault diagnosis method for collaborative case reasoning and semantic model reasoning, and belongs to the field of fault diagnosis and information technology in the industrial Internet of Things.

Background technique

Semantic model-based fault diagnosis methods have a wide range of applications in the context of the Industrial Internet of Things. Among them, the ontology model can integrate, share and reuse knowledge, overcome the problem of data heterogeneity, generate diagnostic rules according to the model, and can be combined with case-based reasoning (CBR) to improve the diagnostic efficiency. According to the working mode, equipment can be divided into different levels (such as equipment, systems, components, parts), and the failure of one level may affect the working mode of other levels. For such information, Failure Modes and Effects Analysis (FMEA) is an effective extraction method. Since FMEA has the problem of difficulty in acquiring knowledge and inconsistency in expression, integrating the extracted knowledge into the ontology model is an effective solution, but there are still some problems:

1) The accuracy of fault diagnosis based on the ontology model depends on the completeness of the model and the rationality of the generated rules, which requires the support of expert experience analysis (shallow knowledge) and the structure and data of the diagnostic object (deep knowledge);

2) In the inference process, the rigid rules generated by the ontology model may cause inaccurate inference results.

SUMMARY OF THE INVENTION

In order to improve the integrity of the fault diagnosis ontology model, improve the diagnosis rules, improve the problems caused by rigid rules, and improve the accuracy of the inference results and diagnosis, the present invention provides an equipment fault diagnosis method for coordinating case reasoning and semantic model reasoning. Based on the knowledge extracted from fuzzy logic and FMEA analysis, a fault diagnosis semantic model is constructed in the form of ontology, and CBR is integrated into the model to improve the diagnosis efficiency.

The technical scheme mainly adopted in the present invention is:

A device fault diagnosis method for collaborative case reasoning and semantic model reasoning, comprising the following steps:

S1 collects cases and builds a case library;

S2 combines the knowledge extracted by fuzzy logic and FMEA analysis method to construct a fault diagnosis ontology model with the methodological process of fuzzy ontology development;

Based on the knowledge obtained in the ontology model, S3 generates the corresponding SWRL rules in combination with expert experience, and performs conflict detection on the generated SWRL rules to form a fault diagnosis rule base;

S4 performs fault detection according to the above-built fault diagnosis ontology model, rule base and case base. The detection process is divided into a CBR module and an RBR module, wherein the CBR module includes an ontology model and a case base, and is carried out by reusing the experience of historical cases. For fault diagnosis, the RBR module includes an ontology model and a rule base, and is used for fault diagnosis through SWRL rule reasoning when the CBR module fails, that is, when the case matching similarity is lower than a threshold.

Preferably, in the construction of the fault diagnosis ontology model in the step S2, the fuzzy logic and the FMEA analysis method are combined, and the fuzzy ontology development methodology process is used to develop the ontology, and the specific steps are as follows:

S2-1: Use the FMEA analysis method to obtain the relationship between the equipment and components of the same level or different levels in the diagnostic object;

S2-2: For the part with uncertainty, define the degree of fuzzy requirements in the methodological process of fuzzy ontology development, and integrate the above knowledge into the fault diagnosis ontology model to improve the completeness of the relationship between its classes and attributes.

Preferably, in the step S4, the specific process of fault detection is as follows:

S4-1: Select the parameter features in the new case;

S4-2: First execute the CBR module, based on the fault diagnosis ontology model, and according to the parameter features selected in step S4-1, perform case retrieval in the case database. If the parameter features and the cases in the case database are successfully matched, the case is used. , the selected parameter features and the fault repair method in the matching case are regarded as a new case, and the similarity between it and the matching case is evaluated. If the similarity is lower than the threshold, it will be stored in the case database for case update. The threshold value is not updated. After the evaluation is completed, the CBR module is terminated to complete the fault diagnosis; if the parameter feature fails to match the case in the case database, the CBR module is terminated, and the process goes to step S4-3;

S4-3: Execute the RBR module, based on the fault diagnosis ontology model, according to the SWRL rules in the rule base to determine the fault cause, fault location and maintenance measures, complete the fault diagnosis, and add the case to the case database for updating; If the fault type is the fault type, firstly diagnose it through the SWRL rule in the rule base, if the diagnosis result is valid, then add the case to the case base for update.

Preferably, rules newly generated from the rule base according to expert experience or actual case analysis are added to the rule base for updating after rule conflict detection is performed.

Beneficial effects: The present invention provides a device fault diagnosis method for collaborative case reasoning and semantic model reasoning. On the basis of the combination of CBR and RBR, the knowledge extracted by fuzzy logic and FMEA analysis method is integrated into the construction of the ontology model, improving the performance of the ontology model. The integrity of the ontology model is improved, and the definition of uncertain knowledge is more reasonable; at the same time, the shallow knowledge and deep knowledge are used to construct diagnostic rules, which improves the integrity and accuracy of the rule base, thereby improving the reasoning of the diagnostic framework. .

Description of drawings

Figure 1 is a flowchart of fault diagnosis;

Figure 2 shows the construction process of the ontology model and rule base used by the RBR module;

Figure 3 shows the workflow of the CBR module.

Detailed ways

In order to make those skilled in the art better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present application, and Not all examples. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the scope of protection of the present application.

As shown in Figure 1, a device fault diagnosis method for collaborative case reasoning and semantic model reasoning includes the following steps:

S1 collects cases and builds a case library;

S2 combines the knowledge extracted by fuzzy logic and FMEA analysis method to construct a fault diagnosis ontology model with the methodological process of fuzzy ontology development;

Based on the knowledge obtained in the ontology model, S3 generates the corresponding SWRL rules combined with expert experience, and forms a fault diagnosis rule base after the conflict detection of the generated SWRL rules. (The construction of the rule base belongs to the existing technology. After the model is constructed, the rules are manually edited according to its structure and relationship, so it is not described in detail).

S4 performs fault detection according to the above-built fault diagnosis ontology model, rule base and case base. The detection process is divided into a CBR module and an RBR module, wherein the CBR module includes an ontology model and a case base, and is carried out by reusing the experience of historical cases. For fault diagnosis, the RBR module includes an ontology model and a rule base, and is used for fault diagnosis through SWRL rule reasoning when the CBR module fails, that is, when the case matching similarity is lower than a threshold.

Preferably, as shown in FIG. 2 , in the construction of the fault diagnosis ontology model in the step S2, fuzzy logic and FMEA analysis method are combined, and the fuzzy ontology development methodology process is used to develop the ontology. The specific steps are as follows:

S2-1: Use the FMEA analysis method to obtain the relationship between the equipment and components of the same level or different levels in the diagnostic object;

S2-2: For the part with uncertainty, define the degree of fuzzy requirements in the methodological process of fuzzy ontology development, and integrate the above knowledge into the fault diagnosis ontology model to improve the completeness of the relationship between its classes and attributes. (The uncertain part in the present invention needs to be determined according to the actual field, generally it is not clearly described in the case, or there is no fault case in the equipment but the relevant parameter part that may fail. The knowledge is the fuzzy ontology development methodology ( FODM) the definition of the uncertain part in the process, integration refers to the definition of vague related classes and parameters in the fault diagnosis ontology model).

Preferably, in the step S4, as shown in FIG. 3, the specific process of fault detection is as follows:

S4-1: Select the parameter features in the new case;

S4-2: First execute the CBR module, based on the fault diagnosis ontology model, and according to the parameter features selected in step S4-1, perform case retrieval in the case database. If the parameter features and the cases in the case database are successfully matched, the case is used. , the selected parameter features and the fault repair method in the matching case are regarded as a new case, and the similarity between it and the matching case is evaluated. If the similarity is lower than the threshold, it will be stored in the case database for case update. The threshold value is not updated. After the evaluation is completed, the CBR module is terminated to complete the fault diagnosis; if the parameter feature fails to match the case in the case database, the CBR module is terminated, and the process goes to step S4-3;

S4-3: Execute the RBR module, based on the fault diagnosis ontology model, according to the SWRL rules in the rule base to determine the fault cause, fault location and maintenance measures, complete the fault diagnosis, and add the case to the case database for updating; If the fault type is the fault type, firstly diagnose it through the SWRL rule in the rule base, if the diagnosis result is valid, then add the case to the case base for update.

Preferably, rules newly generated from the rule base according to expert experience or actual case analysis are added to the rule base for updating after rule conflict detection is performed.

In the present invention, the CBR and RBR modules are called for fault diagnosis, and the relevant information of the fault is determined by inputting parameters such as fault mode and related data values, and obtaining feedback.

1. Construction of ontology model

For the construction of ontology model, the process of Fuzzy Ontology Development Methodology (FODM) is adopted. Firstly, according to the FMEA analysis method, the relationship between equipment, systems and components and related parameter values are extracted as the basis for the construction of classes, object attributes and data attributes in the ontology model, mainly including failure mode number, failure mode, equipment components, failure effects, failure causes and maintenance measures. The information obtained above is defined as related classes++ and attributes in the ontology, and after extracting the FMEA knowledge, the building steps of the ontology model considering the ambiguity are as follows:

1) Determine the purpose and scope of the ontology, ask basic questions and answer them clearly.

2) Determine the need for ambiguity: further identify the domain or scope of the ontology, and determine whether to introduce ambiguity into the ontology design according to the clarity of the knowledge extracted from the FMEA, and identify different types of ambiguity according to the definition of the fuzzy ontology.

3) Determining the vaguely relevant information: Conduct further investigation on the part that needs vague definition, identify the information that really needs vague definition (if there is precise information, no need for ambiguity), put it in the expected field (that is, the equipment that is actually going to be faulted). The knowledge base is divided into precise information and fuzzy information.

4) Consider reusing existing ontologies: Examine existing ontologies related to the domain or scope and determine their reusability.

5) Reuse fuzzy ontology elements: Check the reusable ontology to determine whether the selected ontology elements are fuzzy, if only clear ontology elements can be reused, go to step 7, if only fuzzy ontology elements can be reused, go to step 6; If both exist, activate steps 6 and 7 at the same time.

6) Correct fuzzy ontology elements: The fuzzy specification and modeling provided by the existing fuzzy ontology elements may not match. It is necessary to improve the fuzzification of the ontology elements to meet the requirements of the target ontology, and to ensure that there is information with ambiguous meaning in the expected field or application. make a correct approximation.

7) Define Fuzzy Ontology Elements: Define different fuzzy ontology elements based on expertise or historical statistics to provide a correct approximation to the nature of vague and imprecise information in the domain.

8) Clearly define ontology elements: use a combination of top-down and bottom-up approaches to develop class hierarchies, define relationships (object properties) to link different concepts (classes), and develop data attributes, axioms, instance clarity Ontology elements.

9) Regularization: Formalize the ontology elements designed in step 8 into a machine-readable format.

10) Verification: Verification of the designed ontology in terms of correctness, consistency, completeness, rationality, understandability, and simplicity.

In addition, the case library is constructed in the form of ontology, and related classes and attributes are created and integrated into the ontology model of fault diagnosis to support the operation of the CBR module.

2. CBR module

In the event of a failure, the parameters obtained from the data acquisition device will change. Case retrieval refers to searching the case library based on parameter characteristics such as equipment type, fault phenomenon and operating parameters. The case base is a collection of fault diagnosis cases constructed by filing corresponding data for historical cases according to the categories and attributes of fault diagnosis ontology. Ontology models can be retrieved using semantically constructed cases, and case matching is achieved by matching the classes defined by the above features, with actual data values of parameters corresponding to the feature index between the ontology model and the case. The similarity value is used to judge the similarity between the fault to be diagnosed and the cases in the case database. If the similarity is greater than the set threshold, the result can be considered. After the case matching is successfully completed, the actual fault diagnosis result can be saved to in the case library.

3. RBR module

When the CBR module fails to match the case, the RBR module is triggered. According to the knowledge acquired in the ontology model and the SWRL rules generated by the expert experience, the rule base of the fault diagnosis model is formed. According to the failure parameters such as failure mode, detected related data attribute values and other elements, the relevant rules in the rule base are triggered to perform inference to achieve the purpose of fault diagnosis. The inference results show the possible failure causes and provide corresponding maintenance measures. For unknown types of faults, preliminary inferences are made according to the class and attribute relationships in the ontology and SWRL rules. If the inferred result conforms to the facts, the result can also be saved in the case library. In addition, the rule base can also be continuously improved, and newly generated rules based on expert experience or actual case analysis can be added to the rule base after rule conflict detection is performed.

The expert experience mentioned in the present invention is the knowledge about the structure of the equipment obtained by analyzing the equipment, or the judgment of the existing fault, which is a professional term.

The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, several improvements and modifications can be made. It should be regarded as the protection scope of the present invention.

Claims (4)

1. A device fault diagnosis method based on collaborative case reasoning and semantic model reasoning is characterized by comprising the following steps:

s1, collecting cases and constructing a case library;

s2, establishing a fault diagnosis ontology model by a fuzzy ontology development methodology process by combining knowledge extracted by fuzzy logic and FMEA analysis;

s3, generating a corresponding SWRL rule by combining with expert experience on the basis of the knowledge obtained in the ontology model, and performing conflict detection on the generated SWRL rule to form a fault diagnosis rule base;

s4 carries out fault detection according to the fault diagnosis body model, rule base and case base, the detection process is divided into a CBR module and an RBR module, wherein the CBR module comprises the body model and the case base, fault diagnosis is carried out by reusing experience of historical cases, and the RBR module comprises the body model and the rule base and is used for carrying out fault diagnosis through SWRL rule reasoning when the CBR module fails, namely when case matching similarity is lower than a threshold value.

2. The method for diagnosing equipment faults by coordinating case reasoning and semantic model reasoning with claim 1, wherein: in the fault diagnosis ontology model construction of the step S2, a fuzzy logic and FMEA analysis method are combined, and an ontology is developed by adopting a fuzzy ontology development methodology flow, which includes the following specific steps:

s2-1: obtaining the relation between the equipment and the components at the same level or different levels in the diagnostic object by using an FMEA (failure mode and effects analysis) method;

s2-2: and for the part with uncertainty, defining the fuzzy requirement degree of the fuzzy ontology in the development methodology process of the fuzzy ontology, and integrating the knowledge into the fault diagnosis ontology model to improve the relationship integrity degree of the classes and the attributes of the fault diagnosis ontology model.

3. The method for diagnosing equipment faults by coordinating case reasoning and semantic model reasoning with claim 1, wherein: in step S4, the specific process of fault detection is as follows:

s4-1: selecting parameter characteristics in the new case;

s4-2: firstly, executing a CBR module, based on a fault diagnosis body model, carrying out case retrieval in a case base according to the parameter characteristics selected in the step S4-1, if the parameter characteristics are successfully matched with the cases in the case base, adopting the cases, using the selected parameter characteristics and a fault maintenance method in the matching cases as a new case, evaluating the similarity between the new case and the matching cases, if the similarity is lower than a threshold value, storing the new case in the case base for case updating, if the similarity is higher than the threshold value, not updating, finishing the CBR module after the evaluation is finished, and finishing fault diagnosis; if the matching of the parameter characteristics and the cases in the case base fails, ending the CBR module and entering the step S4-3;

s4-3: executing an RBR module, judging the fault reason, fault position and maintenance measure according to the SWRL rule in the rule base based on the fault diagnosis body model, completing fault diagnosis, and adding the case into the case base for updating; if a new fault type occurs, the fault type is primarily diagnosed through the SWRL rule in the rule base, and if the diagnosis result is valid, the case is added into the case base for updating.

4. The method for diagnosing equipment faults by coordinating case reasoning and semantic model reasoning with claim 1, wherein: and the rule base is added into the rule base for updating after rule conflict detection is carried out according to the newly generated rule based on expert experience or actual case analysis.

Images