Power distribution box fire early warning method and system based on multiple sensors

Abstract

The power distribution box fire early warning method and system based on multiple sensors, the method comprises the following steps: acquiring the real-time environmental characteristics, the ontology characteristics of the power distribution box and the corresponding segmented threshold; according to the cross-validation result, determining the segmented early warning level of the current pyrolysis particle concentration, the power distribution box environment temperature, the power distribution box current and the cable temperature, and recording the cumulative point number of the corresponding segmented early warning level; according to the preset upgrade rule and the prediction model, whether there is a segmented early warning level alarm level within a certain time is predicted, and if there is, an early warning is sent. The cross-validation between the environmental characteristics and the ontology characteristics is used to determine whether an abnormal point is a cumulative point of the corresponding segmented early warning, which reduces the possibility of false positives and effectively eliminates false alarms; at the same time, the early warning method of hierarchical accumulation is adopted, so that the staff can focus on solving important problems, and the safety of the fire early warning is ensured.

Description

Technical Field

[0001] This invention relates to the field of fire early warning, and in particular to a method and system for early warning of fires in distribution boxes based on multiple sensors. Background Technology

[0002] Existing early warning systems all require the collection and analysis of fire characteristics for prediction. The selection of these characteristic data often focuses on several specific scenarios, such as wood placed at 60–100°C (CN113538838A, Temperature monitoring method for identifying pyrolysis particles in cultural relics buildings), or experimental scenarios clearly in the process of large-scale smoldering (CN104766433A, Temperature alarm system based on data fusion). These technologies do not address the detection of early-stage hazards. In existing fire early warning systems, the most frequent false alarms occur in the early stages. Therefore, accurate identification of risks at this stage is a pain point of existing technologies that has not been resolved.

[0003] Existing fire alarm systems need to handle numerous abnormal scenarios and issue warnings for various situations. These warnings cover different scenarios, risk exposures, and expected occurrences, consuming significant time and effort for staff in daily alarm processing. In particular, a large number of false alarms and non-critical alarms can overwhelm important alarms, delaying problem resolution. In practice, identifying genuine alarms and classifying alarms according to their urgency is a major challenge that urgently needs to be addressed.

[0004] Therefore, those skilled in the art are dedicated to developing a multi-sensor-based early warning classification processing method to overcome the problems existing in the prior art. Summary of the Invention

[0005] To address the aforementioned problems, this invention proposes a method and system for early warning of fires in distribution boxes based on multiple sensors.

[0006] The main contents of this invention include:

[0007] A multi-sensor-based method for early warning of fires in distribution boxes includes:

[0008] The real-time environmental and physical characteristics of the distribution box are obtained. The environmental characteristics include the concentration of pyrolysis particles and the ambient temperature of the distribution box. The physical characteristics include the current of the distribution box and the cable temperature.

[0009] The segmented thresholds corresponding to the pyrolysis particle concentration, the ambient temperature of the distribution box, the current of the distribution box, and the cable temperature are obtained; the segmented thresholds include cumulative thresholds and alarm thresholds; the cumulative thresholds include a first threshold, a second threshold, and a third threshold with progressively increasing levels.

[0010] Based on the segmented thresholds corresponding to pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and based on the results of cross-validation, the segmented warning levels for the current pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature are determined, and the cumulative number of corresponding segmented warning levels is recorded. The segmented warning levels include normal level, first warning level, second warning level, third warning level, and alarm level.

[0011] Based on the preset upgrade rules and prediction model, it is predicted whether there will be a segmented warning level of alarm level within a certain period of time. If so, an alarm is issued. The preset upgrade rules refer to the rules for upgrading the current segmented warning level to the next level.

[0012] Preferably, the preset upgrade rules include:

[0013] When the current segmented warning level is non-alarm level, obtain the cumulative number of points of the current segmented warning level within a certain period of time.

[0014] Obtain the current cumulative threshold for upgrading the segmented warning level;

[0015] If the cumulative number of points for the current segmented warning level exceeds the threshold for upgrading the current segmented warning level within a certain period of time, the cumulative number of points for the next level of the current segmented warning level is incremented by 1.

[0016] Preferably, based on the segmented thresholds corresponding to pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and according to the results of cross-validation, the segmented early warning levels for the current pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature are determined, including:

[0017] When the current pyrolysis particle concentration exceeds the corresponding segment threshold, obtain the pyrolysis particle concentration, distribution box current, and cable temperature at the previous time point;

[0018] If the difference between the current pyrolysis particle concentration and the previous time point exceeds the set value, then calculate the difference between the current distribution box current and / or cable temperature and the previous time point.

[0019] If the difference between the current current and / or cable temperature of the current distribution box and the current and cable temperature of the previous time point exceeds the set value, the cross-validation passes and is recorded as an accumulation point of a corresponding segmented warning level; otherwise, the cross-validation fails and is recorded as a special event.

[0020] Preferably, when the current pyrolysis particle concentration exceeds the corresponding segmentation threshold, the method further includes obtaining current weather data from the cloud, wherein the weather data includes haze concentration;

[0021] Based on the smog concentration, adjust the preset segmented threshold of pyrolysis particle concentration or calculate the actual pyrolysis particle concentration;

[0022] Compare the current pyrolysis particle concentration with the adjusted pyrolysis particle concentration segment threshold, or compare the actual pyrolysis particle concentration with the preset pyrolysis particle concentration segment threshold. If the difference exceeds the preset value, then calculate the difference between the current distribution box current and / or cable temperature and the previous time point's distribution box current and cable temperature.

[0023] If the difference between the current current and / or cable temperature of the current distribution box and the current and cable temperature of the previous time point exceeds the set value, the cross-validation passes and is recorded as an accumulation point of a corresponding segmented warning level; otherwise, the cross-validation fails and is recorded as a special event.

[0024] Preferably, based on the segmented thresholds corresponding to the pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and using cross-validation, the segmented warning levels for the current pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature are determined, further including:

[0025] When the current current of the distribution box exceeds the corresponding segment threshold, obtain the current of the distribution box and the concentration of pyrolysis particles at the previous time point;

[0026] If the difference between the current current in the distribution box and the current at the previous time point exceeds a set value, then the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point is calculated.

[0027] If the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point exceeds the set value, the cross-validation passes and is recorded as an accumulation point of a corresponding segmented warning level; otherwise, the cross-validation fails and is recorded as a special event.

[0028] Preferably, the first threshold is the high point value of the corresponding normal value obtained by the environmental feature or the ontology feature through the normal model, and the second threshold is the 75th percentile value of the corresponding abnormal value obtained by the environmental feature or the ontology feature through the normal model. The normal model is an AI model that uses the normal environmental data of the distribution box in a safe operating environment as the training set.

[0029] The third threshold is the sampling mean of the environmental feature or the ontology feature obtained through an anomaly model, where the anomaly model is an AI model that uses the labeled data of the destructive experiment as the training set.

[0030] Preferably, the alarm segment threshold is the average of the third segment threshold and the set standard threshold of the alarm.

[0031] Preferably, when the current pyrolysis particle concentration, the ambient temperature of the distribution box, the current of the distribution box, and the cable temperature exceed the corresponding alarm threshold, an alarm is issued based on the cross-validation results.

[0032] This invention also proposes a multi-sensor-based fire early warning system for distribution boxes, comprising:

[0033] The data acquisition module includes a pyrolysis particle sensor for acquiring the concentration of pyrolysis particles in the distribution box, an electrical fire sensor for acquiring the characteristics of the distribution box itself, and a temperature sensor for acquiring the temperature of the distribution box and the cable temperature.

[0034] The processing module, connected to the acquisition module, is used to execute the fire early warning method described above;

[0035] A storage module, connected to the processing module, is used to store the accumulated points of the corresponding segmented early warning levels;

[0036] The alarm module, together with the processing module, issues corresponding alarms based on the corresponding segmented warning levels;

[0037] The cloud platform is used to provide weather data and corresponding segmented thresholds.

[0038] The beneficial effects of this invention are as follows: The fire early warning method and system for distribution boxes based on multiple sensors proposed in this invention determine whether an abnormal point is an accumulation point for the corresponding segmented early warning by cross-validation between environmental features and physical features, thereby reducing the possibility of false alarms and effectively eliminating false alarms; at the same time, the use of a hierarchical accumulation early warning method allows staff to focus their attention on solving important problems, ensuring the safety of fire early warning. Attached Figure Description

[0039] Figure 1 This is a flowchart of the early warning process of the present invention. Detailed Implementation

[0040] The technical solution protected by this invention will be described in detail below with reference to the accompanying drawings.

[0041] This invention proposes a fire early warning method and system for distribution boxes based on multiple sensors. By acquiring the microenvironment and operating status of the distribution box in real time, and then predicting the status data of the distribution box within a certain period of time, the possibility of a fire warning can be determined. This can detect early fire hazards, avoid false alarms and misleading alarms, help staff focus their attention on important events, and improve the quality of supervision and the safety of distribution boxes.

[0042] This invention also proposes a multi-sensor-based fire early warning system for distribution boxes, comprising:

[0043] The data acquisition module includes a pyrolysis particle sensor for acquiring the concentration of pyrolysis particles in the distribution box, an electrical fire sensor for acquiring the characteristics of the distribution box itself, and a temperature sensor for acquiring the temperature of the distribution box and the cable temperature.

[0044] The processing module, connected to the acquisition module, is used to execute the fire early warning method described above;

[0045] A storage module, connected to the processing module, is used to store the accumulated points of the corresponding segmented early warning levels;

[0046] The alarm module, together with the processing module, issues corresponding alarms based on the corresponding segmented warning levels;

[0047] The cloud platform is used to provide weather data and corresponding segmented thresholds.

[0048] Please refer to Figure 1 First, several sensors are deployed inside the distribution box to collect data on the microenvironment and the box's own operation, recording these as environmental and physical characteristics. The collected data is then transmitted to the processing module according to a set sampling frequency. Specifically, the sensors include a pyrolysis particle sensor for collecting the concentration of pyrolysis particles inside the distribution box, an electrical fire sensor for collecting the physical characteristics of the distribution box, and a temperature sensor for collecting the temperature of the distribution box and cables. The environmental characteristics can include both the microenvironment inside the distribution box, such as the concentration of pyrolysis particles and the temperature inside the distribution box, and the macroenvironment surrounding the distribution box, such as the concentration of smog, which can be obtained from cloud-based weather data. The physical characteristics include the temperature of the cables and the current in the distribution box.

[0049] The processing module acquires real-time characteristic values ​​such as pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, as well as their respective segmented thresholds. Based on the segmented thresholds corresponding to the pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and according to the results of cross-validation, it determines the current segmented warning level for each of these parameters and records the cumulative number of points for each segmented warning level. The segmented warning levels include normal level, first warning level, second warning level, third warning level, and alarm level. Based on preset escalation rules and a prediction model, it predicts whether there will be a segmented warning level that reaches the alarm level within a certain period of time; if so, an alarm is issued. The preset escalation rules refer to the rules for escalating the current segmented warning level to the next level. Specifically, the preset upgrade rules include: when the current segmented warning level is non-alarm level, obtaining the cumulative number of points for the current segmented warning level within a certain period of time; obtaining the upgrade cumulative point threshold for the current segmented warning level; when the cumulative number of points for the current segmented warning level within a certain period of time is greater than the upgrade cumulative point threshold for the current segmented warning level, incrementing the cumulative number of points for the next level of the current segmented warning level by 1; and the prediction model can use real data within the cumulative time window as input parameters of the neural network model to predict the changing trend of response data within a certain period of time, and determine the warning level within a certain period of time based on the predicted data, thereby achieving early warning of risks.

[0050] To provide staff with a visual distinction, the first threshold can be considered as the boundary between the normal range and the most basic warning. If the current pyrolysis particle concentration exceeds the first threshold and cross-validation passes, it is recorded as an accumulation point of the first warning level. Blue represents the first warning level, yellow represents the second warning level, orange represents the third warning level, and red represents the alarm level.

[0051] The processing module compares the current collected values ​​with the corresponding thresholds for each segment. When one feature value exceeds its corresponding threshold, cross-validation is initiated, using other feature values ​​for cross-validation to reduce the possibility of false alarms. If cross-validation fails, it is recorded as a special event. If cross-validation succeeds, it should be a warning or warning accumulation point. If the warning point is at level three (orange warning) or level warning (red warning), the warning is initiated directly according to the pre-set method. If the warning point is another registered warning point, the module then compares whether the number of accumulation points for the corresponding segment warning level within a certain period of time is greater than or equal to its corresponding upgrade accumulation point threshold. If the upgrade condition is met, it is recorded as an accumulation point for the next warning level. If the number of accumulation points within a certain period of time does not reach the set threshold, the number of accumulation points for that warning level is cleared, and recording begins again. Simultaneously, the processing module compares in real time whether there are alarm-level segment warnings. That is, when a feature value is greater than the alarm segment threshold and cross-validation succeeds, or when the number of segment accumulation points for level three warning exceeds its corresponding upgrade accumulation point threshold, the corresponding alarm is issued.

[0052] Specifically, the cross-validation steps include: when the current pyrolysis particle concentration exceeds the corresponding segment threshold, obtaining the pyrolysis particle concentration, distribution box current, and cable temperature at the previous time point; if the difference between the current pyrolysis particle concentration and the previous time point exceeds a set value, calculating the difference between the current distribution box current and / or cable temperature and the previous time point; if the difference between the current distribution box current and / or cable temperature and the previous time point exceeds the set value, the cross-validation passes and is recorded as an accumulation point for a corresponding segment warning level; otherwise, the cross-validation fails and is recorded as a special event.

[0053] Furthermore, when the current pyrolysis particle concentration exceeds the corresponding segmented threshold, the process also includes obtaining current weather data from the cloud, including haze concentration; adjusting the preset segmented threshold for pyrolysis particle concentration or calculating the actual pyrolysis particle concentration based on the haze concentration; comparing the current pyrolysis particle concentration with the adjusted segmented threshold or comparing the actual pyrolysis particle concentration with the preset segmented threshold; if the difference exceeds the preset value, then calculating the difference between the current distribution box current and / or cable temperature and the previous time point's distribution box current and cable temperature; if the difference between the current distribution box current and / or cable temperature and the previous time point's distribution box current and cable temperature exceeds the set value, then cross-validation passes and is recorded as an accumulation point for a corresponding segmented warning level; otherwise, cross-validation fails and is recorded as a special event.

[0054] When the current current of the distribution box exceeds the corresponding segment threshold, the current of the distribution box and the pyrolysis particle concentration at the previous time point are obtained. If the difference between the current current of the distribution box and the current of the distribution box at the previous time point exceeds a set value, the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point is calculated. If the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point exceeds a set value, the cross-validation passes and is recorded as an accumulation point of a corresponding segment warning level. Otherwise, the cross-validation fails and is recorded as a special event.

[0055] In one embodiment, the first threshold is the high point of the corresponding normal value obtained by the environmental feature or the ontological feature through a normal model; the second threshold is the 75th percentile value of the corresponding abnormal value obtained by the environmental feature or the ontological feature through a normal model, wherein the normal model is an AI model trained on normal environmental data of the distribution box in a safe operating environment; and the third threshold is the sample mean obtained by the environmental feature or the ontological feature through an anomaly model, wherein the anomaly model is an AI model trained on labeled data from destructive experiments. Preferably, the alarm threshold is the average of the third threshold and the set standard threshold of the alarm.

[0056] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for early warning of fires in distribution boxes based on multiple sensors, characterized in that, include: The real-time environmental and physical characteristics of the distribution box are obtained. The environmental characteristics include the concentration of pyrolysis particles and the ambient temperature of the distribution box. The physical characteristics include the current of the distribution box and the cable temperature. The segmented thresholds corresponding to the pyrolysis particle concentration, the ambient temperature of the distribution box, the current of the distribution box, and the cable temperature are obtained; the segmented thresholds include cumulative thresholds and alarm thresholds; the cumulative thresholds include a first threshold, a second threshold, and a third threshold with progressively increasing levels. Based on the segmented thresholds corresponding to pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and based on the results of cross-validation, determine the segmented warning levels for the current pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature, and record the cumulative number of corresponding segmented warning levels. The segmented early warning levels include normal level, first early warning level, second early warning level, third early warning level, and alarm level; Based on preset upgrade rules and prediction models, it is predicted whether there will be a segmented warning level of alarm level within a certain period of time. If so, an alarm is issued. The preset upgrade rules refer to the rules for upgrading the current segmented warning level to the next level. The cross-validation includes: when the current pyrolysis particle concentration exceeds the corresponding segmented threshold, obtaining the pyrolysis particle concentration, distribution box current, and cable temperature at the previous time point; if the difference between the current pyrolysis particle concentration and the previous time point exceeds a set value, calculating the difference between the current distribution box current and / or cable temperature and the previous time point; if the difference between the current distribution box current and / or cable temperature and the previous time point exceeds a set value, the cross-validation passes and is recorded as an accumulation point for a corresponding segmented warning level; otherwise, the cross-validation fails and is recorded as a special event. The cross-validation rules also include: when the current current of the distribution box exceeds the corresponding segment threshold, the current of the distribution box and the pyrolysis particle concentration at the previous time point are obtained; if the difference between the current current of the distribution box and the current of the distribution box at the previous time point exceeds a set value, the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point is calculated; if the difference between the current pyrolysis particle concentration and the pyrolysis particle concentration at the previous time point exceeds the set value, the cross-validation passes and is recorded as an accumulation point of a corresponding segment warning level; otherwise, the cross-validation fails and is recorded as a special event.

2. The method for early warning of fires in distribution boxes based on multiple sensors according to claim 1, characterized in that, The preset upgrade rules include: when the current segmented warning level is non-alarm level, obtaining the cumulative number of points for the current segmented warning level within a certain period of time; obtaining the upgrade cumulative point threshold for the current segmented warning level; when the cumulative number of points for the current segmented warning level within a certain period of time is greater than the upgrade cumulative point threshold for the current segmented warning level, incrementing the cumulative number of points for the next level of the current segmented warning level by 1; otherwise, clearing the cumulative number of points for the current segmented warning level to zero.

3. The method for early warning of fires in distribution boxes based on multiple sensors according to claim 1, characterized in that, When the current pyrolysis particle concentration exceeds the corresponding segmented threshold, the process also includes obtaining current weather data from the cloud, including haze concentration; adjusting the preset segmented threshold for pyrolysis particle concentration or calculating the actual pyrolysis particle concentration based on the haze concentration; comparing the current pyrolysis particle concentration with the adjusted segmented threshold or comparing the actual pyrolysis particle concentration with the preset segmented threshold; if the difference exceeds the preset value, calculating the difference between the current distribution box current and / or cable temperature and the previous time point's distribution box current and cable temperature; if the difference between the current distribution box current and / or cable temperature and the previous time point's distribution box current and cable temperature exceeds the set value, cross-validation passes and is recorded as an accumulation point for a corresponding segmented warning level; otherwise, cross-validation fails and is recorded as a special event.

4. The method for early warning of fires in distribution boxes based on multiple sensors according to claim 1, characterized in that, The first threshold is the high point value of the corresponding normal value obtained by the normal model of the environmental feature or the ontology feature; the second threshold is the 75th percentile value of the corresponding abnormal value obtained by the normal model of the environmental feature or the ontology feature; the normal model is an AI model that uses normal environmental data of the distribution box in a safe operating environment as the training set; the third threshold is the sampling mean value obtained by the abnormal model of the environmental feature or the ontology feature; the abnormal model is an AI model that uses labeled data of destructive experiments as the training set.

5. The method for early warning of fires in distribution boxes based on multiple sensors according to claim 4, characterized in that, The alarm segment threshold is the average of the third segment threshold and the set standard threshold of the alarm.

6. The method for early warning of fires in distribution boxes based on multiple sensors according to claim 5, characterized in that, When the current pyrolysis particle concentration, distribution box ambient temperature, distribution box current, and cable temperature exceed the corresponding alarm threshold, an alarm will be issued based on the cross-validation results.

7. A fire early warning system for distribution boxes based on multiple sensors, characterized in that, include: The acquisition module includes a pyrolysis particle sensor for acquiring the concentration of pyrolysis particles in the distribution box, an electrical fire sensor for acquiring the characteristics of the distribution box itself, and a temperature sensor for acquiring the temperature of the distribution box and the temperature of the cables; a processing module connected to the acquisition module for executing the fire early warning method as described in any one of claims 1 to 6; and a storage module connected to the processing module for storing the accumulated points of the corresponding segmented early warning levels. An alarm module, connected to the processing module, issues corresponding alarms based on the corresponding segmented warning levels; a cloud platform is used to provide weather data and corresponding segmented thresholds.

Images