Cascade fault diagnosis method based on root cause intensity power set belief rule base

Abstract

The invention discloses a cascade fault diagnosis method based on a root cause intensity power set belief rule base. According to the technical scheme, the method comprises the steps of 1, constructing a power set belief rule base; the conclusion hypothesis unit can clearly describe a single-point fault state, a composite fault state and a causal chain fault state. And 2, introducing a root cause strength parameter: introducing the root cause strength parameter into the conclusion hypothesis unit constructed in the step 1 for quantifying the causal influence capability of the fault state as the root cause. And step 3, reasoning fusion: calculating the matching degree and the activation degree of each rule, and applying an evidence theory-based reasoning fusion algorithm to an output result to obtain power set confidence distribution of each fault hypothesis. And step 4, fault diagnosis and fault root cause positioning are carried out, and a main fault root cause is output, and the main fault root cause is the root cause with the maximum root cause strength. The method is mainly used for realizing accurate diagnosis of cascade faults in a complex system by popularizing a traditional BRB model to power set reasoning and introducing a root cause strength concept.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to a cascaded fault diagnosis method based on a root cause strength power set confidence rule base. Background Technology

[0002] In modern complex systems, components are typically tightly coupled physically and logically, leading to a high degree of complexity in failure modes. Cascading failures are a particularly challenging and devastating type of failure. They occur when a failure of a local component triggers a series of subsequent component malfunctions or performance degradations due to the interconnectedness of the system, ultimately resulting in widespread system paralysis or even complete collapse. Unlike independent single-point failures, cascading failures are characterized by dynamic propagation and causal relationships. Therefore, quickly and accurately diagnosing the root cause of cascading failures is a core challenge in system operation and maintenance.

[0003] Current fault diagnosis methods can be broadly classified into three categories: analytical model-based methods, data-driven methods, and knowledge-based methods. However, these methods still have some problems. For example, mathematical model-based methods rely on precise physical or mathematical models of the system. They generate residual signals by constructing observers or filters to detect and isolate faults. However, for modern large and complex systems, it is usually very difficult to build accurate and complete mathematical models. The uncertainty, nonlinearity, and high modeling cost of the models greatly limit the practical application of this method, especially for cascading faults, whose dynamic propagation process involves multiple subsystems, making it difficult to describe their causal chain with a unified mathematical model. Data-driven methods, such as deep learning and support vector machines, learn the complex mapping relationship between fault modes and monitoring signals from massive historical data to achieve fault classification. However, although these methods have high classification accuracy in some cases, their decision-making process lacks transparency and interpretability. Their performance is heavily dependent on a large amount of complete and accurately labeled historical data. When faced with small samples, noisy data, or novel fault chains that have never appeared before, the generalization ability and robustness of the model will significantly decrease. Furthermore, it is impossible to directly embed the causal knowledge and experience of domain experts into the model, resulting in a waste of knowledge resources. As a typical knowledge-based approach, confidence rule base expert systems successfully combine IF-THEN rule structures with DS evidence theory and fuzzy set theory. They effectively handle uncertain information and possess good model transparency and interpretability, making them particularly suitable for complex system modeling scenarios involving small samples and multi-source information fusion. However, when applied to cascading fault diagnosis, traditional confidence rule bases reveal inherent structural flaws: their rule conclusions are confined to an identification framework consisting of single mutually exclusive fault types, failing to express the concept of fault propagation. In other words, they can only provide the confidence scores of two faults separately, but cannot express them as an intrinsically related whole. Essentially, a traditional confidence rule base is a fault classifier, not a causal analyzer. When a system exhibits multiple symptoms due to cascading faults, a traditional confidence rule base may provide high confidence scores for multiple faults, but cannot identify which fault is the root cause of the entire chain reaction.

[0004] Therefore, there is an urgent need to construct a new diagnostic method that, while inheriting the advantages of confidence rule bases in handling uncertainty, breaks through the limitations of their conclusion expression capabilities, so as to achieve accurate description of the fault propagation chain and root cause localization. Summary of the Invention

[0005] To address the above shortcomings, this invention provides a cascaded fault diagnosis method based on a root cause strength power set confidence rule base. This method can analyze fault propagation based on root cause strength and is applicable to fault diagnosis and root cause localization in complex systems such as power systems, communication networks, and industrial manufacturing.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a cascaded fault diagnosis method based on a root cause strength power set confidence rule base, comprising the following steps: Step 1: Construct a power set confidence rule base; the power set confidence rule base uses all non-empty subsets of the set of potential root causes of system failures as conclusion hypothesis units, and the conclusion hypothesis units can clearly describe single-point failure states, compound failure states, and causal chain failure states.

[0007] Step 2: Introduce root cause strength parameters: In the conclusion hypothesis unit constructed in Step 1, root cause strength parameters are introduced to quantify the causal influence of the fault state as a root cause, and the root cause strength of each fault is set to be equal for compound faults in order to realize compound fault modeling.

[0008] Step 3: Inference Fusion. Based on the power set confidence rule base described in Step 1, calculate the matching degree and activation degree of each rule to form the rule activation weight. Apply the evidence theory-based inference fusion algorithm to the output results of all extended rules in the power set confidence rule base to obtain the power set confidence degree distribution of each fault hypothesis. Extract the cascaded fault propagation path candidate set based on the inference results and calculate the root cause strength parameter for each fault.

[0009] Step 4: Fault diagnosis and root cause localization. Based on the reasoning results of Step 3, determine the most likely fault cascading propagation path; and output the main fault root cause, which is the root cause with the greatest root cause intensity.

[0010] Furthermore, the construction of the power set confidence rule base in step one is as follows: S101: Define the identification framework and attributes; assuming the identification framework is defined as follows: ; in, It is the first Types of fault states The power set P is composed of It consists of a subset.

[0011] The power set is defined as follows: ; Each non-empty subset serves as a conclusion in the power set confidence rule base model.

[0012] The power set features are: ; Among them, a single-element subset represents a single point of failure; a multi-element subset represents a compound failure or a causal chain.

[0013] Suppose we have the following set of attributes: Each attribute corresponding A set of reference levels .

[0014] S102: Construct a power-set confidence rule base; perform similarity matching between actual observations and reference levels to obtain the corresponding evidence distribution: in, This indicates the degree of matching between the observed value and the reference level.

[0015] Furthermore, the rule definitions of the power set confidence rule base in step S102 are as follows: No. Rules Represented as: Furthermore, in step two, a root cause strength parameter is introduced into the conclusion assumption unit to describe the root cause strength in the fault propagation chain, including the following steps: S201: For each composite conclusion Introducing the root cause strength vector And it satisfies the normalization constraint. The higher the root cause strength, the more likely the fault is the starting point of the cascading chain in that subset.

[0016] S201: Some compound faults do not have a clear causal directionality. In this case, the root cause strength of each fault should be considered equal, as shown below: Furthermore, the reasoning fusion process in step three is as follows: S301: Calculate the matching degree and activation degree of each rule.

[0017] S302: Evidence Fusion; The consequent evidence generated by each rule is fused using an Evidence Reasoning (ER) algorithm. For subset conclusions... The confidence level of the fused subset conclusions is: .

[0018] Furthermore, in step S301, the matching degree is calculated by: calculating the input attributes in the power set root cause confidence rule base. degree of matching with the reference value of this attribute in the k-th rule ,in, Let be the weight of the j-th attribute; Calculate the activation degree: Calculate the activation degree for the first... The activation degree of each rule, with a rule activation weight of 1. ;in It is the rule confidence score, with a value range of [value range missing]. .

[0019] Furthermore, in step four, the fault cascading propagation path is to output the final conclusion of the synthetic inference by selecting the subset of conclusions with the highest confidence level.

[0020] Furthermore, the primary root cause of the fault mentioned in step four is the root cause with the highest root cause intensity. Based on the root cause intensity vector in each conclusion, a comprehensive root cause score is calculated for each fault. The scoring formula is as follows: ; The main fault root is such that The largest .

[0021] On the other hand, a cascaded fault diagnosis system based on a confidence rule base includes: a confidence rule base module, a data processing module, an inference module, and a root cause strength module.

[0022] Compared with the prior art, the beneficial effects of the present invention are: 1. It realizes the explicit expression and diagnosis of cascading fault causal chains. By redefining the identification framework, the fault causal chain (such as F1→F2) is used as the basic assumption unit, so that the model conclusion can directly output the confidence level of "F2 was caused by root cause F1". This fundamentally breaks through the bottleneck of traditional BRB, which can only give the confidence level of isolated faults and cannot describe the causal relationship between faults.

[0023] 2. It provides structured and highly interpretable root cause localization capabilities. By introducing a root cause strength vector for each complex fault state, the model can not only identify "which faults occurred," but also accurately quantify "which fault is the starting point of the entire chain reaction." This achieves a leap from "fault classification" to "root cause tracing," providing a direct and clear basis for operational decisions.

[0024] 3. A transparent and reliable "knowledge neural network" was constructed. This invention combines the transparency of knowledge-based methods with the inferential capabilities of data-driven methods. The "IF-THEN" structure of the rules makes the decision-making process fully traceable, while the parallel reasoning mechanism based on evidence theory ensures efficient information fusion and computation, overcoming the shortcomings of "black box" models such as deep learning in terms of unclear decision-making logic in critical tasks.

[0025] 4. Improved robustness and adaptability of the model in complex scenarios. This method has low dependence on precise mathematical models and does not rely on massive amounts of labeled fault data. By embedding causal knowledge from domain experts, the model still exhibits good diagnostic performance and generalization ability when faced with small samples, noisy data, or novel fault propagation paths that have never appeared before. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0027] Figure 1 The flowchart below shows a cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to the present invention. Figure 2 This is a schematic diagram illustrating the principle of a cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to the present invention. Figure 3 This is a flowchart illustrating the diagnostic process of a cascaded fault diagnosis method based on a root cause strength power set confidence rule base, as described in this invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0029] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0030] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Furthermore, the technical features involved in the different embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0031] A cascaded fault diagnosis method based on a root cause strength power set confidence rule base, such as Figure 1 , 2 As shown, it includes the following steps: Step 1: Construct a power set confidence rule base; the power set confidence rule base uses all non-empty subsets of the set of potential root causes of system failures as conclusion hypothesis units, and the conclusion hypothesis units can clearly describe single-point failure states, compound failure states, and causal chain failure states.

[0032] The steps for constructing the power set confidence rule base in step one are as follows: S101: Define the identification framework and attributes; assuming the identification framework is defined as follows: ; in, It is the first Types of fault states The power set P is composed of It consists of a subset.

[0033] The power set is defined as follows: ; Each non-empty subset serves as a conclusion in the power set confidence rule base model.

[0034] The power set features are: ; Among them, a single-element subset represents a single point of failure; a multi-element subset represents a compound failure or a causal chain.

[0035] Suppose we have the following set of attributes: Each attribute corresponding A set of reference levels .

[0036] S102: Construct a power-set confidence rule base; perform similarity matching between actual observations and reference levels to obtain the corresponding evidence distribution: in, This indicates the degree of matching between the observed value and the reference level.

[0037] The rules of the power set confidence rule base are defined as follows: No. Rules Represented as: Step 2: Introduce root cause strength parameters: In the conclusion hypothesis unit constructed in Step 1, root cause strength parameters are introduced to quantify the causal influence of the fault state as a root cause, and the root cause strength of each fault is set to be equal for compound faults in order to realize compound fault modeling.

[0038] The root cause strength parameter is introduced into the conclusion assumption unit to describe the root cause strength in the fault propagation chain, including the following steps: S201: For each composite conclusion Introducing the root cause strength vector And it satisfies the normalization constraint. The higher the root cause strength, the more likely the fault is the starting point of the cascading chain in that subset.

[0039] S201: Some compound faults do not have a clear causal directionality. In this case, the root cause strength of each fault should be considered equal, as shown below: Step 3: Inference Fusion. Based on the power set confidence rule base described in Step 1, calculate the matching degree and activation degree of each rule to form the rule activation weight. Apply the evidence theory-based inference fusion algorithm to the output results of all extended rules in the power set confidence rule base to obtain the power set confidence degree distribution of each fault hypothesis. Extract the cascaded fault propagation path candidate set based on the inference results and calculate the root cause strength parameter for each fault.

[0040] The reasoning fusion process is as follows: S301: Calculate the matching degree and activation degree of each rule.

[0041] Calculate the matching degree: Calculate the input attributes in the power set root cause confidence rule base. degree of matching with the reference value of this attribute in the k-th rule ,in, Let be the weight of the j-th attribute; Calculate the activation degree: Calculate the activation degree for the first... The activation degree of each rule, with a rule activation weight of 1. ;in It is the rule confidence score, with a value range of [value range missing]. .

[0042] S302: Evidence Fusion; The consequent evidence generated by each rule is fused using an Evidence Reasoning (ER) algorithm. For subset conclusions... The confidence level of the fused subset conclusions is: .

[0043] Step 4: Fault diagnosis and root cause localization. Based on the reasoning results of Step 3, determine the most likely fault cascading propagation path; and output the main fault root cause, which is the root cause with the greatest root cause intensity.

[0044] The fault cascade propagation path is to output the final conclusion of the synthetic inference by selecting the subset of conclusions with the highest confidence level.

[0045] The primary root cause of the failure is the root cause with the highest root cause intensity. Based on the root cause intensity vectors in each conclusion, a comprehensive root cause score is calculated for each failure. The scoring formula is as follows: ; The main fault root is such that The largest .

[0046] Figure 3 This is a flowchart illustrating the diagnostic process of a cascaded fault diagnosis method based on a root cause strength power set confidence rule base, as described in this invention.

[0047] The diagnostic process includes the following steps in sequence: First, input the real-time observation data of the system and perform necessary data preprocessing to form a normalized input vector.

[0048] Next, the pre-built power set confidence rule library is invoked to match the preprocessed input vector with the premises of the rules in the library, and the corresponding rule set is activated based on the matching result.

[0049] Subsequently, the evidence reasoning algorithm is used to fuse the evidence of the activated rule conclusions to obtain the comprehensive confidence distribution on the power set identification framework, and this is used as the diagnostic conclusion output.

[0050] Finally, decision tree analysis is performed based on the output diagnostic conclusions: First, it is determined whether the conclusion with the highest confidence level is a single-element subset. If so, a single point of failure is determined to have occurred in the system, and this failure is the root cause; if not, the multi-element subset analysis path is entered to check the root cause strength of each failure within the subset: if all root cause strengths are equal, it is determined to be a concurrent failure with no root cause; otherwise, the failure with the highest root cause strength is identified as the root cause of the cascading failure.

[0051] On the other hand, a cascaded fault diagnosis system based on a confidence rule base includes: a confidence rule base module, a data processing module, an inference module, and a root cause strength module.

[0052] The confidence rule base module is used to store the power set confidence rule base constructed in step one of a cascaded fault diagnosis method based on the root cause strength power set confidence rule base; The data processing module is used to collect and preprocess system monitoring data and calculate rule matching degree; The reasoning module is used to perform evidence theory reasoning to calculate the confidence level of the failure hypothesis; The root cause strength module is used to calculate the strength score of each fault root cause based on the reasoning results and identify the main root cause.

[0053] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A cascaded fault diagnosis method based on a root cause strength power set confidence rule base, characterized in that, Includes the following steps: Step 1: Construct a power set confidence rule base; the power set confidence rule base uses all non-empty subsets of the set of potential root causes of system failures as conclusion hypothesis units, and the conclusion hypothesis units can clearly describe single-point failure states, compound failure states, and causal chain failure states; Step 2: Introduce root cause strength parameters: In the conclusion hypothesis unit constructed in Step 1, root cause strength parameters are introduced to quantify the causal influence of the fault state as a root cause, and the root cause strength of each fault is set to be equal for compound faults in order to achieve compound fault modeling. Step 3: Inference Fusion. Based on the power set confidence rule base described in Step 1, calculate the matching degree and activation degree of each rule to form the rule activation weight. Apply the inference fusion algorithm based on evidence theory to the output results of all extended rules in the power set confidence rule base to obtain the power set confidence degree distribution of each fault hypothesis. Extract the cascaded fault propagation path candidate set based on the inference results and calculate the root cause strength parameter for each fault. Step 4: Fault diagnosis and root cause localization. Based on the reasoning results of Step 3, determine the most likely fault cascading propagation path; and output the main fault root cause, which is the root cause with the greatest root cause intensity.

2. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 1, characterized in that, The steps for constructing the power set confidence rule base in step one are as follows: S101: Define the identification framework and attributes; assuming the identification framework is defined as follows: ； in, It is the first Types of fault states The power set P is composed of Composed of subsets; The power set is defined as follows: ； Each non-empty subset serves as a conclusion in the power set confidence rule base model; The power set features are: ； Among them, a single-element subset represents a single point of failure; a multi-element subset represents a compound failure or a causal chain. Suppose that the set of attributes is as follows: Each attribute corresponding A set of reference levels ; S102: Construct a power-set confidence rule base; perform similarity matching between actual observations and reference levels to obtain the corresponding evidence distribution: in, This indicates the degree of matching between the observed value and the reference level.

3. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 2, characterized in that, The rules defined in the power set confidence rule base in step S102 are as follows: No. Rules Represented as: 。 4. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 1, characterized in that, Step two introduces a root cause strength parameter into the conclusion assumption unit to describe the root cause strength in the fault propagation chain, including the following steps: S201: For each composite conclusion Introducing the root cause strength vector And it satisfies the normalization constraint. The higher the root cause strength, the more likely the fault is the starting point of the cascading chain in that subset; S201: Some compound faults do not have a clear causal directionality. In this case, the root cause strength of each fault should be considered equal, as shown below: 。 5. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 1, characterized in that, The reasoning fusion process described in step three is as follows: S301: Calculate the matching degree and activation degree of each rule; S302: Evidence Fusion; fusing the consequential evidence generated by each rule using an evidence reasoning algorithm; for subset conclusions... The confidence level of the fused subset conclusions is: .

6. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 5, characterized in that, In step S301, the matching degree is calculated by: calculating the input attributes in the power set root cause confidence rule base. degree of matching with the reference value of this attribute in the k-th rule ,in, Let be the weight of the j-th attribute; Calculate the activation degree: Calculate the activation degree for the first... The activation degree of each rule, with a rule activation weight of 1. ;in It is the rule confidence score, with a value range of [value range missing]. .

7. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 1, characterized in that, The fault cascading propagation path in step four is to output the final conclusion of the synthetic inference by selecting the subset of conclusions with the highest confidence level.

8. The cascaded fault diagnosis method based on a root cause strength power set confidence rule base according to claim 7, characterized in that, The primary root cause of the fault, as described in step four, is the root cause with the highest root cause intensity. Based on the root cause intensity vectors in each conclusion, a comprehensive root cause score is calculated for each fault. The scoring formula is as follows: ； The main fault root is such that The largest .

9. A cascaded fault diagnosis system based on a confidence rule base, comprising: The module consists of a confidence rule base, a data processing module, an inference module, and a root cause strength module.

Images