Substation fault diagnosis method and system for constructing decision tree based on sequential logic
By constructing a decision tree based on time-series logic, redundant alarm signals are eliminated and the time-series relationships of alarm signals are utilized. This solves the problem of insufficient utilization of redundancy attributes and time-series characteristics in substation fault diagnosis, and achieves efficient and interpretable fault diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-10
- Publication Date
- 2026-03-31
AI Technical Summary
Existing substation fault diagnosis methods suffer from redundancy sensitivity, inability to utilize the timing characteristics of alarm signals, and poor robustness, resulting in low diagnostic efficiency and limited accuracy.
A decision tree-based approach based on temporal logic is adopted. By eliminating redundant alarm signals, a decision tree is constructed using the temporal logic relationship of the alarm signals, generating concise decision rules to achieve rapid fault type localization.
It improves the accuracy of fault diagnosis and system robustness, reduces computational complexity, enhances interpretability, and is suitable for complex and ever-changing substation operating environments.
Smart Images
Figure CN121765242A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of fault diagnosis technology, specifically to a substation fault diagnosis method and system based on a decision tree constructed using time-series logic. Background Technology
[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.
[0003] In power systems, substations serve as core hubs, and their operational status directly impacts the safety and stability of the power grid. When a substation malfunctions, the automation system generates a large number of fault alarm signals in a very short time, creating an "alarm storm." These alarm signals contain rich fault information, but also suffer from redundancy, incompleteness, or disordered timing, placing a significant analytical burden on operators and severely affecting the efficiency of fault location and emergency response. Therefore, there is an urgent need for an intelligent and efficient fault diagnosis method to improve the processing capacity of alarm information and the accuracy of diagnosis.
[0004] Currently, substation fault diagnosis mainly relies on expert systems, neural networks, and decision trees. Expert systems depend on manually constructed knowledge bases, resulting in high maintenance costs and difficulty in handling novel faults. Neural networks possess nonlinear modeling capabilities, but their "black box" nature makes the diagnostic process difficult to interpret. Traditional decision tree algorithms (such as ID3 and C4.5), while logically clear, suffer from three shortcomings when processing substation alarm information: first, they are sensitive to redundancy attributes, easily generating lengthy and complex tree structures, leading to overfitting; second, they cannot effectively utilize the temporal characteristics of alarm signals, losing important fault propagation path information; and third, they exhibit poor robustness and limited diagnostic accuracy when faced with uncertain information such as missing or out-of-order alarms. Summary of the Invention
[0005] To address the aforementioned issues, this invention proposes a substation fault diagnosis method and system based on time-series logic to construct a decision tree. By integrating attribute simplification and alarm signal time-series logic, a highly efficient decision tree model with fault propagation interpretation capabilities is constructed, enabling more accurate, interpretable, and robust substation fault diagnosis.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: One or more embodiments provide a substation fault diagnosis method based on time-series logic to construct a decision tree, including the following steps: Obtain historical fault data of the power plant, use alarm signal sequence as conditional attribute, and fault type or component as decision attribute to construct a fault sample decision table; Based on the data in the fault sample decision table, redundant alarm signals are removed to obtain the core attribute set; The core alarm signal attributes in the core attribute set are used as candidate splitting attributes, and the temporal logical relationship of the alarm signals is used as the basis for node splitting to construct a decision tree; Traverse the path from the root node to each leaf node of the constructed decision tree, and combine the splitting attributes and branch conditions on the path to form decision rules; The alarm signal sequence received in real time from the substation is matched with the obtained decision rules to determine the fault type.
[0007] One or more embodiments provide a substation fault diagnosis system based on timing logic to construct a decision tree, including: The decision table construction module is configured to acquire historical fault data of the power plant, use alarm signal sequences as conditional attributes, and use fault type or component as decision attributes to construct a fault sample decision table. The reduction module is configured to remove redundant alarm signals from the data in the fault sample decision table to obtain the core attribute set. The decision tree construction module is configured to use the core alarm signal attributes in the core attribute set as candidate splitting attributes and the temporal logical relationship of the alarm signals as the basis for node splitting to construct a decision tree; The decision rule generation module is configured to traverse the path from the root node to each leaf node of the constructed decision tree, combine the splitting attributes and branch conditions on the path to form decision rules; The fault diagnosis module is configured to obtain the real-time received alarm signal sequence from the substation and match it with the obtained decision rules to determine the fault type.
[0008] An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the processor executes the computer instructions, it performs the steps in the substation fault diagnosis method based on timing logic to construct a decision tree, as described above.
[0009] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps in the above-described substation fault diagnosis method based on timing logic to construct a decision tree.
[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: Compared to traditional ID3 or C4.5 decision tree algorithms, the method of this invention has significant advantages in fault diagnosis accuracy and system robustness. Firstly, through attribute simplification, it effectively compresses alarm dimensions, avoiding overfitting caused by redundant attributes, while simultaneously improving the computational efficiency of the diagnostic system. Secondly, by integrating the temporal sequence information of alarm signals, the structure of the decision tree better reflects the actual fault propagation mechanism, enhancing interpretability and diagnostic accuracy. Furthermore, the decision rules are simpler, facilitating understanding and maintenance by engineers, and improving the system's fault tolerance and adaptability in abnormal situations such as missing or out-of-order alarms, making it suitable for complex and ever-changing substation operating environments.
[0011] The advantages of the present invention, as well as its additional advantages, will be described in detail in the following specific embodiments. Attached Figure Description
[0012] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute a limitation thereof.
[0013] Figure 1 This is a flowchart illustrating the substation fault diagnosis method based on time-series logic to construct a decision tree, according to Embodiment 1 of the present invention. Figure 2 The decision tree constructed by the substation fault diagnosis method based on time-series logic in Embodiment 1 of the present invention is a decision tree constructed by the method. Figure 3 This is a schematic diagram of the traditional decision tree construction method in Embodiment 1 of the present invention for constructing a decision tree. Detailed Implementation
[0014] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0015] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0016] It should be noted that the terminology used herein is for describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. It should be noted that, without conflict, the various embodiments and features within those embodiments can be combined with each other. The embodiments will now be described in detail with reference to the accompanying drawings.
[0017] Example 1 In one or more of the technical solutions disclosed in the embodiments, such as Figures 1 to 3 As shown, a substation fault diagnosis method based on time-series logic to construct a decision tree includes the following steps: Step 1: Obtain historical fault data of the power plant, use alarm signal sequence as conditional attribute, and fault type or component as decision attribute to construct a fault sample decision table. Step 2: For the data in the fault sample decision table, remove redundant alarm signals to obtain the core attribute set; Step 3: Select the core alarm signal attributes from the core attribute set as candidate splitting attributes, and use the temporal logical relationship of the alarm signals as the basis for node splitting to construct a decision tree; Step 4: Generate decisions: Traverse the path from the root node to each leaf node of the constructed decision tree, combine the splitting attributes and branch conditions on the path to form decision rules; Step 5: Obtain the real-time received alarm signal sequence from the substation and match it with the obtained decision rules to determine the fault type.
[0018] The method in this embodiment first uses historical fault alarm signals and corresponding fault types collected during actual operation of the substation as data foundation to construct a fault sample decision table containing time-series information. Based on this, through dependency analysis of alarm signal attributes, redundant or irrelevant signals are identified and eliminated, retaining the core alarm attribute set closely related to fault classification. Furthermore, considering the temporal order of alarm signals, a decision tree model with temporal logic judgment capabilities is constructed. Node partitioning is no longer based on static information gain, but comprehensively considers the order of alarms and their contribution to fault judgment, thereby reconstructing the fault development path. Finally, decision rules are generated by traversing the tree structure, and in real-time applications, the current alarm sequence of the substation is matched with the rule base to quickly locate the fault type or corresponding faulty component, achieving efficient and interpretable fault diagnosis.
[0019] Compared to traditional ID3 or C4.5 decision tree algorithms, this method offers significant advantages in fault diagnosis accuracy and system robustness. Firstly, by simplifying the attribute process, it effectively compresses the alarm dimensions, avoiding overfitting caused by redundant attributes and improving the computational efficiency of the diagnostic system. Secondly, by integrating the temporal sequence information of alarm signals, the structure of the decision tree better reflects the actual fault propagation mechanism, enhancing interpretability and diagnostic accuracy. Furthermore, the decision rules are simpler, facilitating understanding and maintenance by engineers, and improving the system's fault tolerance and adaptability in abnormal situations such as missing or out-of-order alarms, making it suitable for complex and ever-changing substation operating environments.
[0020] In step 1, historical fault cases or simulation data of substations are collected, and the sequence of accident alarm signals generated in each fault is used as a condition attribute, and the final fault type or faulty component is used as a decision attribute to construct a fault sample decision table.
[0021] Suppose a substation has two fault types D={d1, d2}, and its alarm signal set C={c1, c2, c3, c4, c5}. u1, u2, u3, u4, u5 are fault samples, and c1, c2, c3, c4, c5 are alarm signals. The decision table is shown in Table 1. Table 1. Examples of decision representations;
[0022] In Table 1, 1 indicates that the signal has appeared, and 0 indicates that it has not appeared; In step 2, the positive domain attribute simplification algorithm based on rough set theory is used to remove redundant alarm signals (while maintaining classification ability) to obtain the core attribute set. Specifically: the positive domain attribute simplification algorithm is used to calculate the positive domain POS_C(D) of the complete alarm signal attribute set C relative to the decision attribute D, which is used as the core attribute set; Step 21: Initialize the simplified attribute set R to be an empty set; Step 23: Iteratively select an attribute 'a' from the complete alarm signal attribute set (C). If adding 'a' to R results in the largest increase in the number of samples in the positive domain POS_(R∪{a})(D) of the new attribute set (R∪{a}) relative to the original positive domain POS_R(D), then add attribute 'a' to R. Repeat this process until the reduced attribute set R is equal to the positive domain of the complete set, completing the reduction process and obtaining the core attribute set. Select one attribute a from the complete set of alarm signal attributes (C), including alarm signals c1, c2, c3, c4, and c5; Wherein, the positive domain of decision attribute D is the set of samples that can be accurately classified into the decision category; Specifically, after selecting an attribute a and adding it to the selected attribute set R, the new attribute set (R∪{a}) has the largest increase in the number of samples or coverage of the positive domain of the decision attribute D compared to the positive domain of the original attribute set R. Thus, the corresponding attribute a is added to the core attribute set. A concrete example of reduction based on the attribute reduction algorithm of positive domains: Initial state: Complete attribute set C = {c1, c2, c3, c4, c5}; Decision set D = {d1, d2}; Calculate the positive domain: Suppose that through calculation, it is found that using only the attribute subset R={c1, c4} is sufficient to unambiguously partition all samples, that is, POS_{c1,c4}(D) = {u1, u2, u3, u4, u5}, and is equal to the positive domain POS_C(D) of the complete attribute set C; Simplified result: The simplified core alarm signal attribute set is obtained as R = {c1, c4}. This means that c2, c3, and c5 are redundant attributes.
[0023] In the above implementation of this embodiment, the conditional attributes that contribute the most to improving classification ability are selected first, the core attribute set is gradually constructed, and finally redundant attributes are eliminated while keeping the classification ability unchanged (the positive domain is equal to the complete attribute set). In step 3, the simplified core alarm signal attribute set R obtained in step 2 is used as the candidate splitting attribute. The temporal logical relationship of the alarm signals is used instead of the traditional information entropy as the basis for node splitting to construct a decision tree. The specific process includes: Step 31, Feature Statistics (Intra-Node Statistics): For the current node of the decision tree, for all samples contained in the current decision tree node, calculate the average occurrence time and frequency of each core alarm signal. Specifically, the average occurrence time of each core alarm signal, i.e., the time sequence relative to the occurrence of the fault; and the frequency of occurrence of each core alarm signal, i.e., the number of times or probability of occurrence.
[0024] Step 32: Perform feature selection based on time priority, and select the alarm signal with the highest time priority as the optimal splitting attribute of the current node; Specifically, the alarm signal with the highest timing priority is the alarm signal with the earliest average occurrence time or the highest frequency of the first alarm signal to appear.
[0025] Step 33: Node division based on alarm signal occurrence: Determine if the optimal splitting attribute appears in the fault sample, divide the current node's fault sample into two child nodes, and generate two branches: Specifically, if the alarm information corresponding to the optimal splitting attribute appears in the sample, it is treated as one branch; if the alarm information corresponding to the optimal splitting attribute does not appear in the sample, it is treated as another branch. Step 34, Recursive Construction: For each newly generated child node, repeat steps 31 to 33 until the recursive construction termination condition is met, and obtain the constructed decision tree; The recursive construction termination condition is satisfied if any of the following conditions are met: (1) The fault samples in the nodes belong to the same fault type; (2) No core alarm signal in the node fault sample can be used as a split attribute; for example, if no corresponding alarm signal is found in any sample in the current node, that is, all available "core alarm signals" in the samples in the current node have been used. (3) The number of node fault samples is less than the preset threshold to prevent overfitting.
[0026] The process of constructing the decision tree in this step is illustrated based on the example in step 2. The core attribute set R={c1, c4} is used to construct the decision tree, as follows: Figure 2 As shown; First, determine the root node: Assume that in all samples where fault d1 occurs, the average occurrence time of alarm signal c1 is 0.1 seconds; In all samples where fault d2 occurred, the average occurrence time of alarm signal c4 was 0.05 seconds; In summary, alarm signal c4 has a higher timing priority (occurs earlier) when distinguishing faults. Therefore, signal c4 is selected as the root node of the decision tree. First split: Alarm c4 has a branch (c4=1): This branch contains fault samples {u3,u4}, all of which belong to fault type d2. This branch becomes a leaf node and is marked as "fault d2".
[0027] The alarm signal c4 did not appear (c4=0) branch: This branch contains fault samples {u1,u2,u5}, all of which belong to fault type d1. This branch also becomes a leaf node, marked as "fault d1".
[0028] Since all branches become pure leaf nodes after the first split, the decision tree is complete. The final decision tree is very concise, with only one layer of judgment. It accurately distinguishes fault types d1 and d2 based on the presence or absence of alarm c4, without further splitting. This decision tree structure is simple and highly efficient, demonstrating the advantages of temporal priority in feature selection. In practical applications, only the triggering status of the c4 signal needs to be monitored to quickly locate the fault category, improving diagnostic efficiency. This method has good interpretability and deployment value in engineering practice, especially suitable for industrial systems with high real-time requirements. When a new fault sample is input, the c4 alarm status only needs to be judged layer by layer along the decision tree path, and fault classification can be completed within 0.1 seconds.
[0029] Figure 2 The decision tree structure based on temporal logic proposed in this embodiment is shown. Figure 3 The traditional decision tree structure relies on multiple redundant alarm signals (such as alarms C2, C3, and C5) during the splitting process, resulting in a lengthy tree structure, complex decision paths, and a tendency to overfit. Figure 2For the decision tree constructed by the method in this embodiment, by using attribute reduction and the temporal relationship of alarm signals, it can effectively distinguish between faults d1 and d2 based only on the key alarm C4. The structure is significantly more concise and the logic is clearer. This improvement not only reduces the model complexity but also enhances the efficiency and interpretability of fault diagnosis.
[0030] From the constructed decision tree, the following two rules can be extracted: IF c4 appears THEN the fault type is d2.
[0031] IF c4 does not appear THEN the fault type is d1.
[0032] In step 4, to generate decision rules, traverse each path from the root node to each leaf node of the constructed decision tree, and combine the splitting attributes and branch conditions on the path to form a series of decision rules in the form of "IF-THEN".
[0033] During actual diagnosis, match the sequence of alarm signals received in real time with these decision rules to quickly determine the fault type.
[0034] In step 5, the method for obtaining the matching of the sequence of alarm signals of the substation received in real time with the obtained decision rules includes the following steps: Step 51: Obtain the sequence of alarm signals received in real time; Step 52: Check whether the currently received alarm sequence contains the alarm signals specified in the rule; Step 53: Check whether the temporal order or time range of the alarm signals is the same as the order or time range of the decision rule to determine the fault type; For example, an IF-THEN decision rule generated based on the decision tree is as follows: "IF c1 appears, AND c2 appears, AND the occurrence time of c1 < the occurrence time of c2 THEN the fault type = d1;" The obtained alarm sequence is: [(c1, 12:01:10), (c2, 12:01:15), (c4, 12:01:18)]; The matching process is as follows: Check whether the alarm signals c1 and c2 are included, and the result is yes; Check whether the occurrence time of the alarm signal c1 is earlier than the alarm signal c2, and the result is yes; Then, the matching is successful and it is determined that the fault type is d1; Compared to the traditional ID3 algorithm, which requires calculating information gain for each attribute and constructing a complex decision tree with redundant judgments when using a complete attribute set (e.g., {c1, c2, c3, c4, c5}), this embodiment effectively optimizes the decision tree construction process by introducing two improvements: attribute simplification and temporal logical priority. On one hand, preprocessing filters out representative core alarm attributes, significantly reducing the input dimension and avoiding interference from redundant features in splitting decisions. On the other hand, using the order of alarm signal occurrence as the splitting basis makes the decision tree closer to the actual fault propagation process, thus providing greater interpretability and accuracy in the splitting path. The resulting decision tree has the advantages of fewer nodes, simpler rules, and higher diagnostic efficiency, significantly improving the response speed and reliability of fault identification, and is suitable for real-time diagnostic needs in complex alarm environments in substations.
[0035] Example 2 Based on Embodiment 1, this embodiment provides a substation fault diagnosis system based on a decision tree constructed using time-series logic, including: The decision table construction module is configured to acquire historical fault data of the power plant, use alarm signal sequences as conditional attributes, and use fault type or component as decision attributes to construct a fault sample decision table. The reduction module is configured to remove redundant alarm signals from the data in the fault sample decision table to obtain the core attribute set. The decision tree construction module is configured to use the core alarm signal attributes in the core attribute set as candidate splitting attributes and the temporal logical relationship of the alarm signals as the basis for node splitting to construct a decision tree; The decision rule generation module is configured to traverse the path from the root node to each leaf node of the constructed decision tree, combine the splitting attributes and branch conditions on the path to form decision rules; The fault diagnosis module is configured to obtain the real-time received alarm signal sequence from the substation and match it with the obtained decision rules to determine the fault type.
[0036] It should be noted that each module in this embodiment corresponds one-to-one with each step in embodiment 1, and their specific implementation process is the same, so it will not be repeated here.
[0037] Example 3 Based on Embodiment 1, this embodiment provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, they complete the steps in the substation fault diagnosis method based on timing logic to construct a decision tree as described in Embodiment 1.
[0038] Example 4 Based on Embodiment 1, this embodiment provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, they complete the steps in the substation fault diagnosis method based on timing logic to construct a decision tree as described in Embodiment 1.
[0039] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
[0040] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.
Claims
1. A substation fault diagnosis method based on decision tree construction using time-series logic, characterized in that, Includes the following steps: Obtain historical fault data of the power plant, use alarm signal sequence as conditional attribute, and fault type or component as decision attribute to construct a fault sample decision table; Based on the data in the fault sample decision table, redundant alarm signals are removed to obtain the core attribute set; The core alarm signal attributes in the core attribute set are used as candidate splitting attributes, and the temporal logical relationship of the alarm signals is used as the basis for node splitting to construct a decision tree; Traverse the path from the root node to each leaf node of the constructed decision tree, and combine the splitting attributes and branch conditions on the path to form decision rules; The alarm signal sequence received in real time from the substation is matched with the obtained decision rules to determine the fault type.
2. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 1, characterized in that: The positive domain attribute simplification algorithm based on rough set theory is used to remove redundant alarm signals and obtain the core attribute set. The process is as follows: The simplified attribute set R is an empty set after initialization; Iteratively select an attribute 'a' from the complete set of alarm signal attributes. If adding 'a' to the attribute set R results in the largest increase in the number of samples in the positive domain of the new attribute set relative to the original positive domain, then add attribute 'a' to R. Iteratively select the attribute that results in the largest increase in the positive domain until the reduced attribute set R is equal to the positive domain of the complete set. This completes the reduction process and yields the core attribute set.
3. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 1, characterized in that, The core alarm signal attributes in the core attribute set are selected as candidate splitting attributes, and the temporal logical relationship of the alarm signals is used as the basis for node splitting to construct a decision tree. The specific process includes: Step 31: For the current node of the decision tree, for all samples contained in the current decision tree node, calculate the average occurrence time and frequency of each core alarm signal; Step 32: Perform feature selection based on time priority, and select the alarm signal with the highest time priority as the optimal splitting attribute of the current node; Step 33: Determine whether the optimal splitting attribute appears in the faulty sample, divide the current node's faulty sample into two child nodes, and generate two branches: For each newly generated child node, repeat steps 31 to 33 until the recursive construction termination condition is met, and the constructed decision tree is obtained.
4. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 3, characterized in that: The recursive construction termination condition is satisfied if any of the following conditions are met: The fault samples in the node belong to the same fault type; The absence of core alarm signals in node failure samples can be used as a splitting attribute. The number of node failure samples is less than the preset threshold.
5. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 3, characterized in that: If the alarm information corresponding to the optimal splitting attribute appears in the sample, it is treated as a branch; If the alarm information corresponding to the optimal splitting attribute does not appear in the sample, it is treated as another branch.
6. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 3, characterized in that: The alarm signal with the highest timing priority is the alarm signal with the earliest average occurrence time or the alarm signal with the highest frequency of the first occurrence.
7. The substation fault diagnosis method based on time-series logic to construct a decision tree as described in claim 1, characterized in that: A method for matching real-time received substation alarm signal sequences with obtained decision rules includes the following steps: Obtain the real-time received alarm signal sequence; Check if the current alarm sequence contains the alarm signal specified in the rule; Check whether the timing sequence or time range of the alarm signals is the same as the timing or time range of the decision rules, and then determine the fault type.
8. A substation fault diagnosis system based on decision tree construction using time-series logic, characterized in that, include: The decision table construction module is configured to acquire historical fault data of the power plant, use alarm signal sequences as conditional attributes, and use fault type or component as decision attributes to construct a fault sample decision table. The reduction module is configured to remove redundant alarm signals from the data in the fault sample decision table to obtain the core attribute set. The decision tree construction module is configured to use the core alarm signal attributes in the core attribute set as candidate splitting attributes and the temporal logical relationship of the alarm signals as the basis for node splitting to construct a decision tree; The decision rule generation module is configured to traverse the path from the root node to each leaf node of the constructed decision tree, combine the splitting attributes and branch conditions on the path to form decision rules; The fault diagnosis module is configured to obtain the real-time received alarm signal sequence from the substation and match it with the obtained decision rules to determine the fault type.
9. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the processor executes the computer instructions, it completes the steps in the substation fault diagnosis method based on timing logic to construct a decision tree as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, complete the steps in the substation fault diagnosis method based on time-series logic to construct a decision tree as described in any one of claims 1-7.