Electrical circuit fire risk early warning method and system based on multi-source data fusion

Abstract

The invention discloses an electrical line fire risk early warning method and system based on multi-source data fusion, and relates to the technical field of electrical line fire risk early warning. Electrical line parameter data are collected in real time, and line parameter risk abnormal grade judgment is carried out; then carrying out pretreatment; key features are extracted from the preprocessed electrical line parameter data; constructing a fire risk prediction model, carrying out training prediction, and carrying out analysis and calculation according to a prediction result to obtain an electrical line fire risk probability value and a predicted occurrence area; according to the electrical circuit fire risk probability value, whether the electrical circuit has a fire risk is judged in combination with a set threshold value, and if it is determined that the fire risk occurs, an alarm signal is sent out and emergency measures are taken. According to the method, the parameter index change of the electrical circuit is monitored in real time, the fire risk induced by the electrical fault is predicted through intelligent analysis of the deep machine learning algorithm, intervention measures are automatically triggered, and the safety and reliability of an electrical system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric line fire risk early warning, and more particularly to an electric line fire risk early warning method and system based on multi-source data fusion. Background Art

[0002] At present, with the rapid development of modern society, the power system has become an indispensable infrastructure to support the operation of all walks of life. In the power system, the low-voltage distribution cabinet is a key equipment for power distribution and conversion. Its operation stability and safety are directly related to the stable operation of the entire power system and the safety of users' electricity use. However, with the continuous increase in the load of electrical equipment and the extension of operation time, the fire risk caused by aging, overload, short circuit, poor contact and other problems of the internal electrical circuit of the low-voltage distribution cabinet has become increasingly prominent. Traditional low-voltage distribution cabinets have many shortcomings in fire warning, such as reliance on manual inspections, slow response speed, low warning accuracy, etc., which makes it difficult to timely and effectively discover and deal with fire hazards, resulting in frequent fire accidents, causing huge casualties and property losses to the society.

[0003] In the complex operating environment of low-voltage distribution cabinets, the occurrence and development of electrical line faults often become the main cause of electrical fire accidents. In the process of electrical fault evolution, the changes in electrical, thermal and gas parameters are used as critical criteria for sensing and predicting line faults or fire risks. Sensor technology is used to perceive the multi-index parameters in the process of electrical line faults in real time and combined with deep machine learning algorithms for intelligent analysis and prediction, which will provide a method for predicting electrical line faults and fire risks in low-voltage distribution cabinets.

[0004] Therefore, for the low-voltage distribution cabinet, the core equipment of the power system, it is urgent to develop an electrical line fire risk warning method that can monitor the changes of multiple indicators of electrical lines, integrate real-time data analysis and intelligent warning, and promptly detect and deal with potential abnormal problems, so as to achieve early identification of electrical line faults and accurate prediction of fire risks, and take early prevention measures. Summary of the invention

[0005] In view of this, the present invention provides an electrical line fire risk warning method and system based on multi-source data fusion to solve the problems in the background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] An electrical line fire risk early warning method based on multi-source data fusion, comprising:

[0008] Real-time collection of electrical circuit parameter data;

[0009] Perform preprocessing operations on the acquired electrical circuit parameter data;

[0010] Extract key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit;

[0011] Use machine learning algorithms to build a fire risk prediction model, use preprocessed electrical line parameter data for training and prediction, analyze and calculate based on the prediction results, and obtain the electrical line fire risk probability value and predicted occurrence area;

[0012] Based on the fire risk probability value of the electrical line and the set threshold, it is determined whether there is a fire risk in the electrical line. If a fire risk is determined, an alarm signal is issued and emergency measures are taken.

[0013] Optionally, the real-time collection of electrical circuit parameter data specifically includes:

[0014] Sensors are placed in low-voltage distribution cabinets to monitor and collect temperature, smoke, residual current and pyrolysis particle data of electrical lines in real time, including change values ​​and transient values; data under normal conditions within a fixed time is collected, as well as historical line data and disaster data; a standard data set is constructed, and the evaluation benchmark is set through mean values ​​and specification standards.

[0015] Optionally, it also includes analyzing the abnormal level of line parameter risk and issuing corresponding sound and light alarms, and taking corresponding specific operation measures according to the corresponding alarm signals; wherein the abnormal level determination includes:

[0016] The collected line temperature data is compared with the standard data set, and the risk level R1 is determined based on the difference, temperature rise law and standard specifications;

[0017] The collected residual current data is compared with the standard data set, and the risk level R2 is determined based on the difference and standard specifications;

[0018] The collected line smoke data is compared with the standard data set, and the risk level R3 is determined based on the difference and smoke concentration standards;

[0019] The collected line pyrolysis particle data are compared with the standard data set, and the risk level R4 is determined based on the difference and the corresponding risk standards of the pyrolysis particle concentration.

[0020] Optionally, the preprocessing specifically includes:

[0021] The electrical line parameter data is cleaned to remove abnormal values ​​and missing values, and the data obtained by the mean method are used to replace the abnormal values ​​and missing values. The mean method is expressed as:

[0022]

[0023] In the formula, a i Indicates whether the data is missing. i =1 means that the data is not missing, a i =n means data is missing, y i is the corresponding value of the data set, n i is the number of non-missing items in the data set;

[0024] Normalize the cleaned and eliminated data and compress the values ​​into the range of [0,1];

[0025] Specifically, the expression of the standardized formula is:

[0026]

[0027] In the formula, x i-min is the i-th parameter data array x i The minimum value in (k), x i-max is the i-th parameter data array x i (k) is the maximum value, i represents the parameter number, k is the kth data, k = 1, 2, 3, …, n.

[0028] Optionally, the method also includes extracting the preprocessed electrical circuit parameter data to obtain the temperature, flue gas concentration, residual current, and pyrolysis particle change trends; and using 70% of the preprocessed electrical circuit parameter data set as a training set and 30% of the data set as a test set.

[0029] Optionally, the building of the fire risk prediction model specifically includes:

[0030] The LSTM-SVM model is selected to predict the risk of electrical fires, and the training set data is used for training; the test set data is tested, and the test set verifies that the model prediction accuracy meets the standard, so as to obtain a trained fire risk prediction model and determine the fire judgment threshold based on the training results; the processed data values ​​are used as input values ​​and sent to the trained fire risk prediction model to predict the intact, damaged, and fire conditions of the electrical lines, and then the fuzzy reasoning is used to calculate the probability of electrical line fires, and the prediction fitting accuracy of the fire risk prediction model is obtained; finally, based on the predicted characteristics and probability values, the fire occurrence area of ​​the line to be tested is obtained, and the fire risk status of the line to be tested is judged.

[0031] Optionally, judging whether a fire risk occurs in the electrical circuit based on the fire risk probability value of the electrical circuit in combination with a set threshold value specifically includes:

[0032] According to the fire risk probability value of the electrical line obtained by the fire risk prediction model, it is judged whether there is a risk in the current line; when the predicted probability value exceeds the preset threshold range, it is judged that there is a fire hazard in the electrical line and a fire alarm signal is issued. When the alarm is issued, the light flashes and the buzzer sounds;

[0033] Priority processing is performed based on the probability value, and intervention measures are automatically taken as needed. The specific area where the fire occurs is obtained through the fire risk prediction model, that is, the point where the fire risk prediction model measures abnormal temperature, smoke, residual current and pyrolysis particles, and human intervention is promptly notified to take corresponding measures.

[0034] An electrical line fire risk early warning system based on multi-source data fusion, including:

[0035] Data acquisition module, real-time acquisition of electrical circuit parameter data;

[0036] A data processing module performs preprocessing operations on the acquired electrical circuit parameter data;

[0037] The feature processing module extracts key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit;

[0038] The training prediction and analysis module uses machine learning algorithms to build a fire risk prediction model, uses preprocessed electrical line parameter data for training and prediction, and analyzes and calculates based on the prediction results to obtain the electrical line fire risk probability value and predicted occurrence area;

[0039] The early warning module determines whether there is a fire risk in the electrical circuit based on the fire risk probability value of the electrical circuit and the set threshold. If a fire risk is determined, an alarm signal is issued and emergency measures are taken.

[0040] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method and system for early warning of electrical line fire risk based on multi-source data fusion, which has the following beneficial effects:

[0041] The present invention uses sensors to monitor the temperature, flue gas, residual current, pyrolysis particle multi-index parameter data of electrical lines in real time, ensures the real-time and accuracy of the data, predicts possible accidents in the future and takes timely measures, and effectively ensures the normal operation of the line.

[0042] The present invention first performs abnormality judgment through the data collected by the sensor. When the line parameter risk abnormality level exceeds 2, it will be further verified through the prediction model. If the predicted probability value exceeds the threshold range, it is judged that there is a fire hazard and a fire confirmation signal is issued, further reducing errors and reducing the false alarm rate.

[0043] The present invention can predict the probability of fire in electrical lines and the specific location of fire through machine learning models, determine whether danger will occur in combination with set thresholds, and issue an alarm signal to confirm the occurrence of fire. Human intervention measures can be taken in a timely manner to achieve accurate prediction of fire risks and effectively prevent the occurrence of fire accidents in electrical lines. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0045] Figure 1 A schematic diagram of the method provided by the present invention;

[0046] Figure 2 This is a schematic diagram of the system structure provided by the present invention. DETAILED DESCRIPTION

[0047] 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.

[0048] The embodiment of the present invention discloses an electrical line fire risk early warning method based on multi-source data fusion, such as Figure 1 As shown, including:

[0049] S1. Real-time collection of electrical circuit parameter data;

[0050] Place multiple types of sensors in the low-voltage distribution cabinet to monitor and collect temperature, flue gas, residual current and pyrolysis particle data of electrical lines in real time, including change values ​​and transient values; collect relevant data under normal conditions within a certain period of time, as well as collect historical data and disaster data of the lines; establish a normal state judgment data set and set its data average value as the standard judgment value. The data includes wire specification standards, relevant parameter data collected under normal conditions, and historical line normal data collected; and judge the abnormal level of line parameter risk, issue corresponding sound and light alarms, and perform corresponding specific operation measures according to the corresponding alarm signals.

[0051] Abnormal level determination includes:

[0052] The collected line temperature data is compared with the normal state judgment data set, and the risk level R1 is determined based on the difference, combined with the temperature rise law and relevant standards and specifications;

[0053] The collected line smoke data is compared with the normal state judgment data set, and the risk level R2 is determined based on the difference and smoke concentration standards;

[0054] The collected line residual current data is compared with the normal state judgment data set, and the risk level R3 is determined based on the difference and relevant standards and specifications;

[0055] The collected line pyrolysis particle data is compared with the normal state judgment data set, and the risk level R4 is determined based on the difference and relevant standards and specifications;

[0056] The risk abnormality level R=R1+R2+R3+R4, its data is assigned a value of 0 within the normal state range, and is assigned values ​​of 1, 2, and 3 according to the difference size combined with the corresponding range standard; when it is higher than the safety threshold, that is, the highest value of the relevant standard is the lowest threshold, and the lowest value of the line data value under abnormal conditions is the highest threshold, it is assigned a value of 3; the maximum risk does not exceed 3, and when it exceeds 3, it is recorded as 3, where risk=0 is safe, risk=1 is a slight abnormality, risk=2 is a moderate risk, and risk=3 is a severe risk; when risk=0, it shows that the current state is safe; when R=1, after receiving the alarm signal, manual confirmation of the abnormal state is performed; when R≥2, there may be extremely large temperature abnormalities or smoke abnormalities or abnormal changes in residual current and pyrolysis particles. After receiving the alarm signal, an abnormal data set is automatically generated, and then the abnormal data set is processed and input into the electrical fire risk prediction model for prediction verification to determine whether there is a fire hazard in the electrical line, and take timely intervention measures after confirming the hazard.

[0057] S2. Preprocessing the acquired electrical circuit parameter data;

[0058] The specific operations of data processing on the obtained electrical circuit parameter data set are as follows: data cleaning is performed on the obtained electrical circuit parameter data set to obtain the cleaned electrical circuit parameter data set, and any missing or abnormal data in the real-time temperature, flue gas, residual current and pyrolysis particle data are searched, and the missing or abnormal data are eliminated, and the data obtained by the mean method are used to replace the abnormal values ​​and missing values.

[0059] The mean value method is expressed as:

[0060]

[0061] In the formula, a iIndicates whether the data is missing. i =1 means that the data is not missing, a i =n means data is missing, y i is the corresponding value of the data set, n i is the number of non-missing items in the data set;

[0062] Normalize the cleaned and eliminated data. Since the attributes, units and sizes of various parameters are different, in order to make the values ​​of various parameters comparable and facilitate unified analysis and comparison, it is necessary to eliminate the differences in the collected original parameter data, that is, to normalize the original parameter data and compress the values ​​into the interval of [0,1].

[0063] Specifically, the expression of the standardized formula is:

[0064]

[0065] In the formula, x i-min is the i-th parameter data array x i The minimum value in (k), x i-max is the i-th parameter data array x i (k) is the maximum value, i represents the parameter number, k is the kth data, k = 1, 2, 3, …, n.

[0066] S3. Extract key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit; also include extracting the preprocessed electrical circuit parameter data to obtain the temperature, flue gas concentration, residual current, and pyrolysis particle change trends; and use 70% of the preprocessed electrical circuit parameter data set as a training set and 30% of the data set as a test set.

[0067] S4. Use machine learning algorithms to build a fire risk prediction model, use preprocessed electrical line parameter data for training and prediction, analyze and calculate based on the prediction results, and obtain the electrical line fire risk probability value and predicted occurrence area;

[0068] The LSTM-SVM model is selected to predict the risk of electrical fires, and the training set data is used for training; the test set data is tested, and the test set verifies that the model prediction accuracy reaches the standard, so as to obtain a trained fire risk prediction model and determine the fire judgment threshold based on the training results; the processed data value is used as the input value, and sent to the trained fire risk prediction model to predict the intact, damaged, and ignited conditions of the electrical line, and then the probability of electrical line fire is calculated using fuzzy reasoning, and the prediction fitting accuracy of the fire risk prediction model is obtained; finally, based on the predicted characteristics and probability values, the fire occurrence area of ​​the line to be tested is obtained, and the fire risk status of the line to be tested is judged. That is, the point where the parameter index measured by the prediction model is abnormal is the point where the fire occurs, and the fire risk status of the line to be tested is judged in combination with the probability value.

[0069] The process of prediction by the trained LSTM-SVM prediction model is as follows:

[0070] The LSTM model is used to predict the temporal changes of the temperature, smoke, residual current and pyrolytic particles of the electrical line based on the collected data set, and the points where the temperature, smoke, residual current and pyrolytic particle concentration on the line are abnormal are obtained;

[0071] The SVM model is used to predict the probability of the electrical line being intact, damaged, or on fire. Finally, based on the prediction features, fuzzy reasoning is used to calculate the probability of a fire failure in the electrical line. Based on the obtained probability value, it is determined whether there is a fire risk in the electrical line.

[0072] The trained LSTM-SVM prediction model is used to predict the fire area, predict the time-varying patterns of the temperature, smoke, residual current and pyrolytic particles of the electrical lines, and predict the points where the temperature, smoke, residual current and pyrolytic particles data are abnormal, so as to determine the location of the fire in the electrical lines.

[0073] The data set is input into the trained LSTM-SVM prediction model to obtain the probability prediction values ​​of intact, damaged, and on-fire conditions of the electrical line and the prediction fitting accuracy. Finally, based on the prediction characteristics, the probability value of the occurrence of electrical line fire failure is calculated using fuzzy reasoning, and the obtained probability value is used to determine whether the electrical line has risks;

[0074] The fuzzy reasoning used is specifically operated as follows:

[0075] Use fuzzy rules for logical editing, and set the fuzzy rule membership function as the normal distribution function. The input fuzzy level is divided into three levels: large (L), medium (M), and small (S). The output fuzzy level is two levels: large (L) and small (S). The input and output are the trained LSTM-SVM network model output probability values ​​p1, p2, p3 and the output electrical line fire probability value p. Finally, the defuzzification centroid method is used to calculate the final electrical line fire probability:

[0076]

[0077] Where p is the probability of fire in electrical circuits, i = 1, 2, 3. i They are respectively the probability of intact, damaged and fire of electrical circuits.

[0078] S5. Based on the fire risk probability value of the electrical circuit and the set threshold, determine whether there is a fire risk in the electrical circuit. If a fire risk is determined, issue an alarm signal and take emergency measures.

[0079] Specifically, the probability value of the line fire risk obtained by the fire risk prediction model is used to determine whether the current line has a risk. When the predicted probability value exceeds the preset threshold range, a fire alarm signal is issued, and the light flashes and the buzzer sounds during the alarm. Combined with the predicted fire area and probability value, after receiving the early warning signal, timely measures are taken, such as cutting off the power supply, tripping, starting the fire extinguishing equipment, etc., and priority processing is performed according to the probability value to prevent the occurrence of fire accidents.

[0080] An electrical line fire risk warning system based on multi-source data fusion, such as Figure 2 As shown, including:

[0081] Data acquisition module, real-time acquisition of electrical circuit parameter data;

[0082] A data processing module performs preprocessing operations on the acquired electrical circuit parameter data;

[0083] The feature processing module extracts key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit;

[0084] The training prediction and analysis module uses machine learning algorithms to build a fire risk prediction model, uses preprocessed electrical line parameter data for training and prediction, and analyzes and calculates based on the prediction results to obtain the electrical line fire risk probability value and predicted occurrence area;

[0085] The early warning module determines whether there is a fire risk in the electrical circuit based on the fire risk probability value of the electrical circuit and the set threshold. If a fire risk is determined, an alarm signal is issued and emergency measures are taken.

[0086] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0087] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An electrical line fire risk early warning method based on multi-source data fusion, characterized in that: include: Real-time collection of electrical circuit parameter data; Perform preprocessing operations on the acquired electrical circuit parameter data; Extract key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit; Use machine learning algorithms to build a fire risk prediction model, use preprocessed electrical line parameter data for training and prediction, analyze and calculate based on the prediction results, and obtain the electrical line fire risk probability value and predicted occurrence area; Based on the fire risk probability value of the electrical line and the set threshold, it is determined whether there is a fire risk in the electrical line. If a fire risk is determined, an alarm signal is issued and emergency measures are taken.

2. The electrical line fire risk early warning method based on multi-source data fusion according to claim 1 is characterized in that: The real-time collection of electrical line parameter data specifically includes: Sensors are placed in low-voltage distribution cabinets to monitor and collect temperature, smoke, residual current and pyrolysis particle data of electrical lines in real time, including change values ​​and transient values; data under normal conditions within a fixed time is collected, as well as historical line data and disaster data; a standard data set is constructed, and the evaluation benchmark is set through mean values ​​and specification standards.

3. The electrical line fire risk early warning method based on multi-source data fusion according to claim 2 is characterized in that: It also includes analyzing the abnormal risk level of line parameters and issuing corresponding sound and light alarms, and taking corresponding specific operational measures according to the corresponding alarm signals; The abnormal level determination includes: The collected line temperature data is compared with the standard data set, and the risk level R1 is determined based on the difference, temperature rise law and standard specifications; The collected residual current data is compared with the standard data set, and the risk level R2 is determined based on the difference and standard specifications; The collected line smoke data is compared with the standard data set, and the risk level R3 is determined based on the difference and smoke concentration standards; The collected line pyrolysis particle data are compared with the standard data set, and the risk level R4 is determined based on the difference and the corresponding risk standards of the pyrolysis particle concentration.

4. The electrical line fire risk early warning method based on multi-source data fusion according to claim 1 is characterized in that: The pre-processing specifically includes: Clean the electrical line parameter data, remove outliers and missing values, and use the data obtained by the mean method to replace outliers and missing values; The cleaned and eliminated data are normalized and the values ​​are compressed to the range of [0,1].

5. The electrical line fire risk early warning method based on multi-source data fusion according to claim 1 is characterized in that: It also includes extracting the pre-processed electrical circuit parameter data to obtain the temperature, smoke concentration, residual current, and pyrolysis particle change trends; And 70% of the preprocessed electrical line parameter data set is used as a training set, and 30% of the data set is used as a test set.

6. The electrical line fire risk early warning method based on multi-source data fusion according to claim 5 is characterized in that: The construction of the fire risk prediction model specifically includes: The LSTM-SVM model is selected to predict the risk of electrical fires, and the training set data is used for training; the test set data is tested, and the test set verifies that the model prediction accuracy meets the standard, so as to obtain a trained fire risk prediction model and determine the fire judgment threshold based on the training results; the processed data values ​​are used as input values ​​and sent to the trained fire risk prediction model to predict the intact, damaged, and fire conditions of the electrical lines, and then the fuzzy reasoning is used to calculate the probability of electrical line fires, and the prediction fitting accuracy of the fire risk prediction model is obtained; finally, based on the predicted characteristics and probability values, the fire occurrence area of ​​the line to be tested is obtained, and the fire risk status of the line to be tested is judged.

7. The electrical line fire risk early warning method based on multi-source data fusion according to claim 1 is characterized in that: The method of judging whether a fire risk occurs in an electrical circuit based on the fire risk probability value of the electrical circuit and the set threshold value specifically includes: According to the fire risk probability value of the electrical line obtained by the fire risk prediction model, it is judged whether there is a risk in the current line; when the predicted probability value exceeds the preset threshold range, it is judged that there is a fire hazard in the electrical line and a fire alarm signal is issued. When the alarm is issued, the light flashes and the buzzer sounds; Priority processing is performed based on the probability value, and intervention measures are automatically taken as needed. The specific area where the fire occurs is obtained through the fire risk prediction model, that is, the point where the fire risk prediction model measures abnormal temperature, smoke, residual current and pyrolysis particles, and human intervention is promptly notified to take corresponding measures.

8. An electrical line fire risk warning system based on multi-source data fusion, characterized in that: An electrical line fire risk early warning method based on multi-source data fusion as described in any one of claims 1 to 7 is applied, comprising: Data acquisition module, real-time acquisition of electrical circuit parameter data; A data processing module performs preprocessing operations on the acquired electrical circuit parameter data; The feature processing module extracts key features from the preprocessed electrical circuit parameter data to characterize the working status of the electrical circuit; The training prediction and analysis module uses machine learning algorithms to build a fire risk prediction model, uses preprocessed electrical line parameter data for training and prediction, and analyzes and calculates based on the prediction results to obtain the electrical line fire risk probability value and predicted occurrence area; The early warning module determines whether there is a fire risk in the electrical circuit based on the fire risk probability value of the electrical circuit and the set threshold. If a fire risk is determined, an alarm signal is issued and emergency measures are taken.

Images