Converter station important event diagnosis method and system based on correlation causality

Abstract

The invention discloses a converter station important event diagnosis method and system based on correlation causality. According to the method, historical messages and equipment information of the converter station are collected and preprocessed, and SER event modeling is carried out; recording sequence events in the database, namely SER events occurring according to a time sequence, and mining association rules according to event levels by utilizing an Apriori method; constructing an event task atlas based on an SER model in combination with an association rule event set and a converter station sequential control operation task process; converting the event atlas into a causal model by utilizing a causal framework based on a graph model; and carrying out real-time analysis on data in the converter station message system by utilizing a causal atlas model, and judging the occurrence probability of an emergency event after a non-emergency event occurs. According to the method, the important events of the converter station are diagnosed based on the correlation causal method, and the method is small in data demand and high in diagnosis accuracy.

Description

Technical Field

[0001] The present invention relates to the field of abnormal event diagnosis of converter stations, and in particular to a method and system for important events of converter stations based on correlation causality. Background Art

[0002] As an important part of the power system, the converter station has been increasing in complexity and equipment intelligence, which has put forward higher requirements for the intelligent operation and maintenance of the converter station. Existing converter station event diagnosis methods mostly rely on experience and traditional monitoring methods, which are difficult to effectively identify complex cause-and-effect relationships. When a DC system fails, the equipment in the converter station will generate a large number of SER events. After the event occurs, if the operation and maintenance personnel cannot promptly analyze the correlation SER event missing in the SER event in a short time, it may lead to the occurrence of power safety events.

[0003] In the event diagnosis of power system converter stations, the current mainstream technology mainly relies on correlation analysis. However, this method can only reveal the statistical relationship between events, and cannot deeply explore the causal laws behind them, resulting in inaccurate or misleading diagnostic results in complex fault situations. In addition, when it comes to important or emergency fault events, sample data is often scarce, which makes conventional solutions incapable of identifying and analyzing faults and unable to effectively cope with the complexity and diversity of emergencies. Therefore, there is an urgent need for a new method that can simultaneously consider correlation and causality to improve the accuracy and reliability of fault diagnosis.

[0004] This project proposes a "correlation-causality" analysis technology for SER event analysis, diagnoses important operation and maintenance events of converter stations, prompts converter station abnormal events, reveals the implicit logic of converter station operation, and improves the operation and maintenance efficiency of converter stations. The algorithm is used to analyze the correlation of converter station events and abnormal feedback, and the occurrence of important events in converter stations can be inferred by mining the correlation and abnormal feedback of converter station events. Summary of the invention

[0005] In the past, the analysis and prediction of abnormal events in converter stations only focused on the correlation between events, but did not pay attention to the causality between events and between events and equipment. By adopting the abnormal event analysis method described in the present invention, the relationship between events can be discovered through correlation analysis, and then causal analysis can be used as an auxiliary to reveal the implicit logic of converter station operation, so as to improve the accuracy and efficiency of diagnosis.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is a method and system for diagnosing important events in converter stations based on correlation causality.

[0007] The technical solution of the method of the present invention is a method for diagnosing important events of converter stations based on correlation causality, and the specific steps are as follows: Step 1: Collect historical messages and equipment information of the converter station and perform preprocessing to perform SER event modeling; Step 2: Record the sequential events in the database, i.e., the SER events that occur in time sequence, and use the Apriori method to mine association rules according to the event level; Step 3: Construct an event task map based on the SER model, combining the association rule event set and the converter station sequential control operation task process; Step 4: Use the causal framework based on the graph model to transform the event graph into a causal model; Step 5: Use the cause-effect graph model to analyze the data in the converter station message system in real time to determine the probability of an emergency event after a non-emergency event occurs; Preferably, the pretreatment described in step 1 is as follows: Eliminate empty sets in the converter station's historical message data, classify the collected equipment information to form clear equipment categories, and record and collect the converter station's sequential control operation task process; The principal component analysis method is used to extract key features from the historical message data of the converter station; The key features of the converter station historical message data include: Time information, event level information, specific event information, and device information; Continue to use PCA to build the SER event model; The SER event model is understood as collecting redundant data in the original message after simple preprocessing into a data set after data screening and cleaning; The principal component analysis method is used to reduce the dimension of the converter station historical message data to form a processed data set, as follows:

[0008] in, Indicates the message data after dimensionality reduction; It means that the original historical message data is standardized to obtain a standardized data set after removing the empty set; : is the matrix weight. During the analysis, we assign higher weights to features such as event information, time information, and event level in the message data. These features play a key role in the dimensionality reduction process; Preferably, the content of mining association rules in step 2 is: The association relationship between events with an event level of emergency and non-emergency events, and the mined antecedent non-emergency events and subsequent emergency events are constructed into an association rule event set; Preferably, the step 3 is as follows: For the association rule event set, the previous event is connected with the subsequent event in a directionally manner to form a directed acyclic graph. By matching the event with the device corresponding to the event in the SER model, the event is connected with the device to construct an event map. For the sequential control operation task flow of the converter station, each operation step in the task is used as a node, and the nodes are connected in the order of the task flow to form a task directed acyclic graph. According to the equipment corresponding to each operation step in the task, the operation steps are connected with the equipment nodes to construct a task map. Preferably, the step 4 is as follows: The event map is used as the experimental group, the task map is used as the control group, and the causality between the event and the operation task is determined by comparing the two results to obtain the causal map inference result. The causal relationship between events, equipment, and tasks is clarified through the causal graph, and the association between events and events, events and equipment, and tasks is obtained - the causal model; This model not only helps to understand the dynamic interactions within the system, but also provides effective decision support for the operation and maintenance of converter stations; Preferably, the step 5 is specifically as follows: collecting relevant message data from the converter station message system in real time, matching the message data with the event occurrence law obtained from the cause-effect graph model, and comparing the timing of event nodes and task operation nodes; If the time sequence of the two events is exactly the same, the preceding event in the association rule event set has a high probability of causing the subsequent event, i.e., the emergency event, to occur, and an early warning is issued.

[0009] The present invention also provides a converter station important event diagnosis system based on correlation causality, comprising: SER event modeling module, used to collect historical messages and equipment information of converter stations and perform preprocessing to perform SER event modeling; The association rule mining module is used to record the sequential events in the database, i.e., the SER events that occur in time sequence, and mine association rules according to the event level using the Apriori method; An event task graph construction module is used to construct an event task graph based on the SER model combined with the association rule event set and the converter station sequential control operation task process; A causal model building module, which is used to transform an event graph into a causal model using a causal framework based on a graph model; The emergency event determination module uses the cause-effect graph model to perform real-time analysis on the data in the converter station message system to determine the probability of an emergency event after a non-emergency event occurs; The present invention has the advantage that it proposes a method based on correlation causality to diagnose important events in a converter station, and this method has small data requirements and high diagnostic accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 : Flow chart of the method of an embodiment of the present invention.

[0011] Figure 2 : Comparison chart of prediction accuracy of embodiments of the present invention. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] In specific implementation, the method proposed in the technical solution of the present invention can be implemented by technical personnel in this field using computer software technology to realize automatic operation process. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of the present invention and computer equipment running the corresponding computer program, should also be within the scope of protection of the present invention.

[0014] The converter station is a key facility in the high-voltage direct current (HVDC) transmission system. If the DC system fails or complex operations occur, a large number of events will occur in a short period of time. If the operation and maintenance personnel cannot promptly discover and analyze key events and information in a short period of time, it may cause the accident to expand and cause safety accidents. It is necessary to mine the key information in the SER event. Prior to this, it has been proposed to use association rules to mine SER events and predict abnormal events. However, since this method only focuses on the correlation between events, the analysis and diagnosis of complex events in the converter station is not accurate enough. Based on this, the present invention proposes to use the association-causal method to diagnose important events in the converter station.

[0015] Combine the following Figure 1-2 The specific implementation of the present invention is a method for important events of converter stations based on correlation causality, which is as follows: like Figure 1 FIG. 1 is a flow chart of a method according to an embodiment of the present invention.

[0016] Step 1: Collect historical messages and equipment information of the converter station and perform preprocessing to perform SER event modeling; The pretreatment described in step 1 is as follows: Eliminate empty sets in the converter station's historical message data, classify the collected equipment information to form clear equipment categories, and record and collect the converter station's sequential control operation task process; The principal component analysis method is used to extract key features from the historical message data of the converter station; The key features of the converter station historical message data include: Time information, event level information, specific event information, and device information; Continue to use PCA to build the SER event model; The SER event model is understood as collecting redundant data in the original message after simple preprocessing into a data set after data screening and cleaning; The principal component analysis method is used to reduce the dimension of the converter station historical message data to form a processed data set, as follows:

[0017] in, Indicates the message data after dimensionality reduction; It means that the original historical message data is standardized to obtain a standardized data set after removing the empty set; : is the matrix weight. During the analysis, we assign higher weights to features such as event information, time information, and event level in the message data. These features play a key role in the dimensionality reduction process; Step 2: Record the sequential events in the database, i.e., the SER events that occur in time sequence, and use methods such as Apriori to mine association rules according to event levels; The main content of the mining is the association between the events with the event level of emergency and non-emergency, and the mined former non-emergency events and latter emergency events are constructed into an association rule event set; Step 3: Construct an event-task graph based on the SER model, combining the association rule event set and the converter station sequential control operation task process; The step 3 is specifically as follows: For the association rule event set, the previous event is connected with the subsequent event in a directionally manner to form a directed acyclic graph. By matching the event with the device corresponding to the event in the SER model, the event is connected with the device to construct an event map. For the sequential control operation task flow of the converter station, each operation step in the task is used as a node, and the nodes are connected in the order of the task flow to form a task directed acyclic graph. According to the equipment corresponding to each operation step in the task, the operation steps are connected with the equipment nodes to construct a task map. Step 4: Use the causal framework based on the graph model to transform the event graph into a causal model; The step 4 is specifically as follows: The event map is used as the experimental group, the task map is used as the control group, and the causality between the event and the operation task is determined by comparing the two results to obtain the causal map inference result. The causal relationship between events, equipment, and tasks is clarified through the causal graph, and the association between events and events, events and equipment, and tasks is obtained - the causal model; This model not only helps to understand the dynamic interactions within the system, but also provides effective decision support for the operation and maintenance of converter stations; Step 5: Use the cause-effect graph model to analyze the data in the converter station message system in real time to determine the probability of an emergency event after a non-emergency event occurs; The step 5 is specifically as follows: collecting relevant message data from the converter station message system in real time, matching the message data with the event occurrence rules obtained from the causal graph model, and comparing the timing of event nodes and task operation nodes; If the time sequence of the two events is exactly the same, the preceding event in the association rule event set has a high probability of causing the subsequent event, i.e., the emergency event, to occur, and an early warning is issued.

[0018] The specific implementation of the system of the present invention is a converter station important event diagnosis system based on correlation causality, including: SER event modeling module, used to collect historical messages and equipment information of converter stations and perform preprocessing to perform SER event modeling; The association rule mining module is used to record the sequential events in the database, i.e., the SER events that occur in time sequence, and mine association rules according to the event level using the Apriori method; An event task graph construction module is used to construct an event task graph based on the SER model combined with the association rule event set and the converter station sequential control operation task process; A causal model building module, which is used to transform an event graph into a causal model using a causal framework based on a graph model; The emergency event determination module uses the cause-effect graph model to perform real-time analysis on the data in the converter station message system to determine the probability of an emergency event after a non-emergency event occurs; The present invention proposes a method based on correlation causality to diagnose important events in converter stations. This method has small data requirements and high diagnostic accuracy. Figure 2 shown.

[0019] This is a comparison of the prediction accuracy of the emergency event prediction method in this paper and the Apriori algorithm. By continuously adding non-emergency events to the SER data set to complicate the SER data set, it is determined whether the two methods are accurate in predicting emergency events. As can be seen from the figure, compared with the two methods, the method in this paper can still maintain a high prediction accuracy for emergency events when the data set is highly redundant, and the event prediction accuracy remains above 80%, while the simple Apriori algorithm has a downward trend in prediction accuracy when the redundancy of the data set continues to increase. Therefore, it is shown that the proposed converter station operation and maintenance event detection method based on correlation-causality has a strong prediction performance and can improve the reliability of converter station operation.

[0020] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0021] It should be understood that the above description of the preferred embodiment is relatively detailed and cannot be regarded as limiting the scope of patent protection of the present invention. Under the enlightenment of the present invention, ordinary technicians in this field can also make substitutions or modifications without departing from the scope of protection of the claims of the present invention, which all fall within the scope of protection of the present invention. The scope of protection requested for the present invention shall be based on the attached claims.

Claims

1. A method for diagnosing important events in converter stations based on correlation causality, characterized in that: The following steps are involved: Step 1: Collect historical messages and equipment information of the converter station and perform preprocessing to perform SER event modeling; Step 2: Record the sequential events in the database, i.e., the SER events that occur in time sequence, and use the Apriori method to mine association rules according to the event level; Step 3: Construct an event task map based on the SER model, combining the association rule event set and the converter station sequential control operation task process; Step 4: Use the causal framework based on the graph model to transform the event graph into a causal model; Step 5: Use the causal graph model to perform real-time analysis on the data in the converter station message system to determine the probability of an emergency event after a non-emergency event occurs.

2. The converter station important event diagnosis method based on correlation causality according to claim 1 is characterized in that: The pretreatment described in step 1 is as follows: Eliminate empty sets in the converter station's historical message data, classify the collected equipment information to form clear equipment categories, and record and collect the converter station's sequential control operation task process; The principal component analysis method is used to extract the key features of the historical message data of the converter station.

3. The converter station important event diagnosis method based on correlation causality according to claim 2 is characterized in that: The key features of the converter station historical message data include: Time information, event level information, specific event information, and device information; Continue to use PCA to build the SER event model; The SER event model is understood as collecting redundant data in the original message after simple preprocessing into a data set after data screening and cleaning; The principal component analysis method is used to reduce the dimension of the converter station historical message data to form a processed data set, as follows: in, Indicates the message data after dimensionality reduction; It means that the original historical message data is standardized to obtain a standardized data set after removing the empty set; : is the matrix weight. During the analysis, we assign higher weights to features such as event information, time information, and event level in the message data. These features play a key role in the dimensionality reduction process.

4. The converter station important event diagnosis method based on correlation causality according to claim 3 is characterized in that: The content of mining association rules described in step 2 is: The association relationship between events with an event level of emergency and non-emergency events is constructed into an association rule event set by mining the former non-emergency events and the latter emergency events.

5. The converter station important event diagnosis method based on correlation causality according to claim 4 is characterized in that: The step 3 is specifically as follows: For the association rule event set, the previous event is connected with the subsequent event in a directionally manner to form a directed acyclic graph. By matching the event with the device corresponding to the event in the SER model, the event is connected with the device to construct an event map. For the sequential control operation task flow of the converter station, each operation step in the task is taken as a node, and the nodes are connected in chronological order based on the task flow sequence to form a task directed acyclic graph. According to the equipment corresponding to each operation step in the task, the operation steps are connected to the equipment nodes to construct a task map.

6. The converter station important event diagnosis method based on correlation causality according to claim 5 is characterized in that: The step 4 is specifically as follows: The event map is used as the experimental group, the task map is used as the control group, and the causality between the event and the operation task is determined by comparing the two results to obtain the causal map inference result. The causal relationship between events, equipment, and tasks is clarified through the causal graph, and the association between events and events, events and equipment, and tasks is obtained - the causal model; This model not only helps to understand the dynamic interactions within the system, but also provides effective decision support for the operation and maintenance of converter stations.

7. The converter station important event diagnosis method based on correlation causality according to claim 6 is characterized in that: The step 5 is specifically as follows: collecting relevant message data from the converter station message system in real time, matching the message data with the event occurrence rules obtained from the causal graph model, and comparing the timing of event nodes and task operation nodes; If the time sequence of the two events is exactly the same, the preceding event in the association rule event set has a high probability of causing the subsequent event, i.e., the emergency event, to occur, and an early warning is issued.

8. A converter station important event diagnosis system based on correlation causality, characterized in that: include: SER event modeling module, used to collect historical messages and equipment information of converter stations and perform preprocessing to perform SER event modeling; The association rule mining module is used to record the sequential events in the database, i.e., the SER events that occur in time sequence, and mine association rules according to the event level using the Apriori method; An event task graph construction module is used to construct an event task graph based on the SER model combined with the association rule event set and the converter station sequential control operation task process; A causal model building module, which is used to transform an event graph into a causal model using a causal framework based on a graph model; The emergency event determination module uses a cause-effect graph model to perform real-time analysis on the data in the converter station message system to determine the probability of an emergency event occurring after a non-emergency event occurs.

Images