Rapid early warning method for faults of ring main unit
By building a ring network cabinet fault warning system, using technical means such as real-time data dynamic update and similarity judgment, the problems of low accuracy of fault warning, high false alarm rate and warning delay in the existing technology have been solved, achieving more efficient and accurate fault warning.
Patent Information
- Application Number
- CN202411888348.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-20
- Publication Date
- 2025-02-18
AI Technical Summary
The existing ring network cabinet fault warning method has low accuracy when dealing with new fault types and abnormal situations, has a high false alarm rate, and has a large amount of data, resulting in warning delays.
Build a fault warning system, including data acquisition, preprocessing, preliminary warning, similarity judgment, accurate warning and model update modules. Through the real-time data dynamic update mechanism, sensor failures and environmental factors are eliminated, and preliminary warnings and high-precision and accurate warnings are adopted to reduce the amount of data and improve the warning efficiency.
It improves the accuracy and reliability of fault warning, reduces the false alarm rate, reduces the warning delay, and improves the efficiency and accuracy of fault warning of ring network cabinets.
Smart Images
Figure CN119469276A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of fault early warning, and in particular to a method for rapid early warning of a ring network cabinet fault. Background Art
[0002] As an important electrical equipment in the power supply and distribution system, the operating status of the ring main unit directly affects the stability and safety of the power system. Therefore, it is necessary to warn of its faults in time. The existing fault warning method mainly collects a large amount of historical data from various sensors, monitoring equipment and related power system records of the ring main unit, uses machine learning for data training after preprocessing, and then applies the trained model to the real-time monitoring data of the ring main unit to predict whether the ring main unit may fail and give corresponding warning information. This method has the following defects: First, historical data can only reflect the past operating conditions and failure modes, and cannot accurately identify new fault types or abnormal conditions, which leads to low reliability of the early warning system; second, the early warning results are highly dependent on the quality and accuracy of the data. Abnormal data may come from equipment failure, sensor failure or environmental factors. The existing method does not exclude sensor failure and environmental factors, resulting in a high false alarm rate; third, in order to improve accuracy and precision, the existing early warning method generally combines multimodal data fusion technology, but this also leads to excessive data processing, resulting in early warning delays. Summary of the invention
[0003] The object of the present invention is to provide a ring main unit fault rapid warning method to solve the problems raised in the above background technology.
[0004] To achieve the above object, the present invention provides the following technical solutions: a ring main unit fault rapid warning method, comprising the following steps: step one, building a fault warning system; step two, data collection; step three, data preprocessing; step four, preliminary warning; step five, similarity judgment; step six, accurate warning; step seven, model update; In the above step 1, the fault warning system constructed includes a data acquisition module, a data preprocessing module, a fault warning module, a similarity judgment module, a model updating module and a data storage module; In the above step 2, the data acquisition module is used to collect device data and send it to the data preprocessing module; In the above step 3, the data preprocessing module preprocesses the data collected in step 2, and then transmits it to the fault warning module and stores it in the equipment history database; In the above step 4, the fault warning module receives the pre-processed data and inputs it into the preliminary warning unit. The preliminary warning unit outputs the fault probability through the preliminary warning model. If the fault probability exceeds the threshold, step 5 is executed. If the fault probability is within the threshold, no action is taken. In the above step 5, the similarity judgment module performs similarity judgment on the preprocessed data and the data in the warning history database. If the similarity exceeds the threshold, the accurate warning result corresponding to the group of warning history data is directly output. If the similarity is within the threshold range, step 6 is executed; In the above step six, the precision warning unit uses a multimodal data fusion algorithm to extract feature data from the preprocessed data and fuses it, and inputs it into the precision warning model as an input feature vector. The precision warning model outputs the fault probability, fault type and severity, and updates the warning history database. If the fault probability exceeds the threshold, a fault alarm reminder is issued. If the fault probability is within the threshold range, no action is taken; In the above step seven, the model update module obtains the device data within the time window range stored in the device history database according to the set time window and the linearly decreasing weight distribution rule, and fuses the data using the weighted average method according to the assigned weights to obtain a fused data set. The fused data set is input into the preliminary warning model and the precision warning model, and the preliminary warning model and the precision warning model are updated using online learning.
[0005] Preferably, in step one, the data preprocessing module establishes data connections with the data acquisition module, the fault warning module and the data storage module respectively, the fault warning module establishes data connections with the similarity judgment module, and the model updating module establishes data connections with the fault warning module and the data storage module respectively.
[0006] Preferably, the fault warning module includes a preliminary warning unit and a precise warning unit.
[0007] Preferably, the data storage module stores an early warning history database and an equipment history database.
[0008] Preferably, in step 2, the device data includes electrical parameters, temperature data, humidity data, vibration data and sound data.
[0009] Preferably, in step three, the preprocessing specifically includes interference data elimination, outlier processing, data smoothing processing and normalization processing.
[0010] Preferably, the interference data exclusion is specifically as follows: monitoring the fluctuation of the data, determining whether the fluctuation amplitude of the relevant data exceeds the threshold, if it does not exceed the threshold, it means that there is no false alarm data, if it exceeds the threshold, it means that the data of this segment whose fluctuation exceeds the threshold is false alarm data, extracting the data and recording it, and at the same time, this segment of data does not participate in subsequent feature extraction, when the number of occurrences of a certain type of data in the recorded false alarm data reaches a preset value, it means that this type of data is an interference factor and needs to be manually excluded, and after the exclusion is completed, the number of occurrences of this type of data is cleared.
[0011] Preferably, in step 4, the preliminary warning model adopts one of logistic regression, decision tree and random forest algorithms.
[0012] Preferably, in step six, the accuracy warning model adopts a deep neural network.
[0013] Preferably, in step seven, the preliminary warning model and the precision warning model are updated by online learning as follows: when the fused data set is input into the model, the model first makes a prediction based on the current parameters, then compares the prediction result with the actual value, calculates the loss function, and finally adjusts the parameters of the model based on the loss function using an optimization algorithm to minimize the loss function.
[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention introduces a dynamic update mechanism for real-time data, so that the early warning model can timely adapt to changes in the operating status of the ring network cabinet, thereby improving the accuracy and reliability of the fault early warning results; the present invention eliminates false alarm data caused by sensor failure and environmental factors by analyzing the fluctuation of relevant data, thereby reducing the false alarm rate of fault early warning; the present invention reduces the amount of data that needs to be processed for early warning by combining efficient preliminary early warning with high-precision precision early warning, thereby improving the efficiency of ring network cabinet fault early warning and ensuring the accuracy of the early warning results. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 is a flow chart of the method of the present invention; Figure 2 This is a structural block diagram of the fault warning system of the present invention. DETAILED DESCRIPTION
[0016] 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.
[0017] Please see attached Figure 1 -Attached Figure 2 , an embodiment provided by the present invention: a ring network cabinet fault rapid warning method, comprising the following steps: step one, building a fault warning system; step two, data collection; step three, data preprocessing; step four, preliminary warning; step five, similarity judgment; step six, accurate warning; step seven, model update; In the above step 1, the fault warning system constructed includes a data acquisition module, a data preprocessing module, a fault warning module, a similarity judgment module, a model updating module and a data storage module; wherein the data preprocessing module establishes data connections with the data acquisition module, the fault warning module and the data storage module respectively, the fault warning module establishes data connections with the similarity judgment module, the model updating module establishes data connections with the fault warning module and the data storage module respectively, the fault warning module includes a preliminary warning unit and a precise warning unit, and the data storage module stores a warning history database and an equipment history database; In the above step 2, the data acquisition module is used to collect device data and send it to the data preprocessing module; wherein the device data includes electrical parameters, temperature data, humidity data, vibration data and sound data; In the above step three, the data preprocessing module preprocesses the data collected in step two, and then transmits it to the fault warning module and stores it in the equipment history database; wherein the preprocessing specifically includes interference data exclusion, outlier processing, data smoothing processing and normalization processing, and interference data exclusion specifically includes: monitoring the fluctuation of the data, judging whether the fluctuation amplitude of the data with correlation exceeds the threshold, if it does not exceed the threshold, it means that there is no false alarm data, if it exceeds the threshold, it means that the data of this section with fluctuation exceeding the threshold is false alarm data, extracting the data and recording it, and at the same time, the data does not participate in the subsequent feature extraction, when the number of occurrences of a certain type of data in the recorded false alarm data reaches a preset value, it means that this type of data is an interference factor and needs to be manually excluded, and after the exclusion is completed, the number of occurrences of this type of data is cleared; In the above step 4, the fault warning module receives the preprocessed data and inputs it into the preliminary warning unit. The preliminary warning unit outputs the fault probability through the preliminary warning model. If the fault probability exceeds the threshold, step 5 is executed. If the fault probability is within the threshold, no action is taken. The preliminary warning model adopts one of the logistic regression, decision tree and random forest algorithms. In the above step 5, the similarity judgment module performs similarity judgment on the preprocessed data and the data in the warning history database. If the similarity exceeds the threshold, the accurate warning result corresponding to the group of warning history data is directly output. If the similarity is within the threshold range, step 6 is executed; In the above step six, the precision warning unit uses a multimodal data fusion algorithm to extract feature data from the preprocessed data and fuses it, and inputs it into the precision warning model as an input feature vector. The precision warning model outputs the fault probability, fault type and severity, and updates the warning history database. If the fault probability exceeds the threshold, a fault alarm reminder is issued. If the fault probability is within the threshold range, no action is taken. The precision warning model adopts a deep neural network. In the above step seven, the model update module obtains the device data within the time window range stored in the device history database according to the set time window and the linearly decreasing weight distribution rule, and fuses the data using the weighted average method according to the assigned weights to obtain a fused data set, and inputs the fused data set into the preliminary warning model and the precision warning model, and updates the preliminary warning model and the precision warning model using online learning. Specifically, when the fused data set is input into the model, the model first makes a prediction based on the current parameters, then compares the prediction result with the actual value, calculates the loss function, and finally uses the optimization algorithm to adjust the parameters of the model based on the loss function to minimize the loss function.
[0018] Based on the above, the advantage of the present invention is that when the invention is used, a dynamic update mechanism for real-time data is introduced. By setting a time window, the latest real-time data is continuously added to the update of the model, and the weight of the old data is gradually reduced, so that the model can adapt to changes in the operating status of the ring network cabinet in time, thereby improving the adaptability of the model to new situations and timely discovering potential faults; the present invention analyzes the fluctuation of relevant data and eliminates false alarm data caused by sensor failure and environmental factors, thereby reducing the false alarm rate of early warning; the present invention adopts a combination of efficient preliminary warning and high-precision precision warning, so that accurate warning can be given when the preliminary warning result exceeds the threshold, and the result of the accurate warning can be directly output when similar data is encountered next time, thereby improving the efficiency and accuracy of ring network cabinet fault warning.
[0019] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.
Claims
1. A ring main unit fault rapid warning method, comprising the following steps: Step 1: Build a fault warning system; Step 2: data collection; Step 3: data preprocessing; Step 4: preliminary warning; Step 5: similarity judgment; Step 6: accurate warning; Step 7: model update; It is characterized by: In the above step 1, the fault warning system constructed includes a data acquisition module, a data preprocessing module, a fault warning module, a similarity judgment module, a model updating module and a data storage module; In the above step 2, the data acquisition module is used to collect device data and send it to the data preprocessing module; In the above step 3, the data preprocessing module preprocesses the data collected in step 2, and then transmits it to the fault warning module and stores it in the equipment history database; In the above step 4, the fault warning module receives the pre-processed data and inputs it into the preliminary warning unit. The preliminary warning unit outputs the fault probability through the preliminary warning model. If the fault probability exceeds the threshold, step 5 is executed. If the fault probability is within the threshold, no action is taken. In the above step 5, the similarity judgment module performs similarity judgment on the preprocessed data and the data in the warning history database. If the similarity exceeds the threshold, the accurate warning result corresponding to the group of warning history data is directly output. If the similarity is within the threshold range, step 6 is executed; In the above step six, the precision warning unit uses a multimodal data fusion algorithm to extract feature data from the preprocessed data and fuses it, and inputs it into the precision warning model as an input feature vector. The precision warning model outputs the fault probability, fault type and severity, and updates the warning history database. If the fault probability exceeds the threshold, a fault alarm reminder is issued. If the fault probability is within the threshold range, no action is taken; In the above step seven, the model update module obtains the device data within the time window range stored in the device history database according to the set time window and the linearly decreasing weight distribution rule, and fuses the data using the weighted average method according to the assigned weights to obtain a fused data set. The fused data set is input into the preliminary warning model and the precision warning model, and the preliminary warning model and the precision warning model are updated using online learning.
2. A ring main unit fault rapid warning method according to claim 1, characterized in that: In step one, the data preprocessing module establishes data connections with the data acquisition module, the fault warning module and the data storage module respectively, the fault warning module establishes data connections with the similarity judgment module, and the model updating module establishes data connections with the fault warning module and the data storage module respectively.
3. A ring main unit fault rapid warning method according to claim 2, characterized in that: The fault warning module includes a preliminary warning unit and a precise warning unit.
4. A ring main unit fault rapid warning method according to claim 2, characterized in that: The data storage module stores an early warning history database and an equipment history database.
5. A ring main unit fault rapid warning method according to claim 1, characterized in that: In the step 2, the device data includes electrical parameters, temperature data, humidity data, vibration data and sound data.
6. A ring main unit fault rapid warning method according to claim 1, characterized in that: In the step three, the preprocessing specifically includes interference data elimination, outlier processing, data smoothing processing and normalization processing.
7. A ring main unit fault rapid warning method according to claim 6, characterized in that: The interference data elimination is specifically as follows: monitoring the fluctuation of the data, determining whether the fluctuation amplitude of the relevant data exceeds the threshold, if it does not exceed the threshold, it means that there is no false alarm data, if it exceeds the threshold, it means that the data of this segment whose fluctuation exceeds the threshold is false alarm data, extracting the data and recording it, and at the same time, this segment of data does not participate in the subsequent feature extraction, when the number of occurrences of a certain type of data in the recorded false alarm data reaches a preset value, it means that this type of data is an interference factor and needs to be manually eliminated, and after the elimination is completed, the number of occurrences of this type of data is cleared.
8. A ring main unit fault rapid warning method according to claim 1, characterized in that: In the step 4, the preliminary warning model adopts one of the algorithms of logistic regression, decision tree and random forest.
9. A ring main unit fault rapid warning method according to claim 1, characterized in that: In step six, the accuracy warning model adopts a deep neural network.
10. A ring main unit fault rapid warning method according to claim 1, characterized in that: In the step seven, the preliminary warning model and the precision warning model are updated by online learning. Specifically, when the fused data set is input into the model, the model first makes a prediction based on the current parameters, then compares the prediction result with the actual value, calculates the loss function, and finally adjusts the parameters of the model based on the loss function using an optimization algorithm to minimize the loss function.