Urban underground comprehensive pipe gallery electrical fire risk assessment method
By using sensors in underground comprehensive pipeline corridors to collect parameters and build a deep neural network model, the electrical fire risk is evaluated in real time, and the problem of untimely monitoring of electrical fires in the existing technology is solved, and the timely detection of fire risks in underground comprehensive pipeline corridors is achieved. and early warning improve the accuracy of fire safety management.
Patent Information
- Application Number
- CN202411204139.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-17
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing fire protection industry is prone to inaccurate alarms in electrical fire monitoring, resulting in frequent fire accidents, especially in narrow and closed spaces of underground comprehensive pipeline corridors. Fire accidents are prone to cause major losses and lack effective fire risk assessment and early warning mechanisms.
Sensors are used to collect the total volatile organic compounds, methane, temperature and other parameters in the underground comprehensive pipeline corridor, and an electrical fire risk assessment model is constructed through a deep neural network to evaluate the fire risk level in real time and provide timely risk warnings.
Real-time assessment of electrical fire risks in underground comprehensive pipeline corridors is achieved, and risks before or early stages of fire are discovered in a timely manner, which improves the accuracy of fire safety management and early warning efficiency, and reduces the possibility of accidents.
Smart Images
Figure CN119151291A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of urban underground comprehensive pipe gallery safety, and in particular relates to an electrical fire risk assessment method for an urban underground comprehensive pipe gallery. Background Art
[0002] At present, the fire protection industry's monitoring of electrical fires is mostly limited to monitoring and alarming when a fire occurs. It is easy for the alarm to be untimely and the best time to control the fire to be missed, causing irreparable losses to people's lives and property.
[0003] Especially for the narrow and confined space of underground utility corridors, due to the special structure and closed space, once a fire accident occurs, it is very easy to cause serious accidents with mass casualties, which poses a new challenge to fire safety. There are a large number of cables in the narrow and long space of underground utility corridors. In recent years, electrical fire accidents caused by cable failures have occurred frequently. According to statistics, electrical fires account for 34%-38% of all fire accidents in my country.
[0004] Therefore, how to assess and warn of fire risks before a fire occurs or in the early stages of a fire is of great significance to the safe operation of underground utility corridor structures and early fire fighting. Summary of the invention
[0005] The purpose of the present invention is to provide a method for assessing the electrical fire risk of an urban underground comprehensive pipe gallery, which can realize real-time assessment of the electrical fire risk of an urban underground comprehensive pipe gallery, so as to facilitate timely discovery of risks before or in the early stages of a fire.
[0006] The technical solution provided by the present invention is:
[0007] A method for assessing the electrical fire risk of an urban underground utility corridor comprises the following steps:
[0008] Step 1: According to the sampling period, the total volatile organic compound concentration, methane concentration and temperature in the underground comprehensive pipeline corridor are collected through sensors;
[0009] The total volatile organic compound concentration C VOCs , methane concentration The temperature T in the underground utility tunnel is respectively related to the standard total volatile organic compound concentration C′ VOCs , Standard methane concentration Compare with the standard temperature T′ in the underground utility corridor;
[0010] If satisfied: The electrical fire risk level in the underground pipe gallery is judged to be level 0;
[0011] If satisfied: The CO sensor, CO2 sensor and ozone sensor are turned on at the same time; and the CO concentration, CO2 concentration and ozone concentration in the underground integrated pipe gallery are collected according to the sampling period;
[0012] Step 2: According to the total volatile organic compound concentration C VOCs , methane concentration The temperature T, CO concentration, CO2 concentration and ozone concentration in the underground utility corridor are used to judge the electrical fire risk of the underground utility corridor, and the electrical fire risk level of the underground utility corridor is divided into level 1, level 2, level 3 and level 4;
[0013] Among them, level 0 means there is no risk of electrical fire, and levels 1, 2, 3 and 4 all mean there is a risk of electrical fire, and the risk of electrical fire increases from level 1 to level 4.
[0014] Preferably, the volatile organic compounds include: methyl chloride, dichloroethylene, chloroform, carbon tetrachloride and trichloroethylene.
[0015] Preferably, in step 2, the method for judging the electrical fire risk of the underground utility corridor is:
[0016] Constructing a deep neural network and training the deep neural network to obtain an underground utility corridor electrical fire risk assessment model;
[0017] The total volatile organic compound concentration C VOCs , methane concentration and the temperature T and CO concentration C in the underground utility corridor CO , CO2 concentration and ozone concentration After normalization processing, the underground utility corridor electrical fire risk assessment model is input, and the underground utility corridor electrical fire risk assessment model outputs the underground utility corridor electrical fire risk level.
[0018] Preferably, the total volatile organic compound concentration C VOCs , methane concentration and the temperature T and CO concentration C in the underground utility corridor CO , CO2 concentration and ozone concentration The formula for normalization is:
[0019]
[0020] Among them, x i represents the measured value of each parameter, x i′ represents the standard value of each parameter; i=1, 2, 3…, 6 represent the total volatile organic compound concentration, methane concentration, temperature, CO concentration, CO2 concentration and ozone concentration respectively.
[0021] Preferably, the standard temperature in the underground integrated pipe gallery is set to 20°C to 25°C; the standard values of other parameters are the actual measured values of other parameters at the standard temperature.
[0022] Preferably, the number of hidden layers of the deep neural network is 5.
[0023] Preferably, the number of neuron nodes in each hidden layer is the same.
[0024] Preferably, the number of neuron nodes in the hidden layer is determined by the following formula:
[0025]
[0026] Among them, N hid is the number of hidden layer neuron nodes, N s is the number of samples in the training set, N in is the number of neuron nodes in the input layer, N out is the number of neuron nodes in the output layer, α is the adjustment variable, and α represents any integer between 2 and 10.
[0027] Preferably, the sampling periods of total volatile organic compound concentration, methane concentration, temperature in the underground integrated pipe gallery, CO concentration, CO2 concentration and ozone concentration are all set to 1 minute.
[0028] The beneficial effects of the present invention are:
[0029] The method for assessing the electrical fire risk of an urban underground comprehensive pipe gallery provided by the present invention can realize real-time assessment of the electrical fire risk of an urban underground comprehensive pipe gallery, so as to facilitate timely discovery of risks before or in the early stages of a fire. DETAILED DESCRIPTION
[0030] The present invention is described in further detail below so that those skilled in the art can implement it according to the description.
[0031] The present invention provides a method for assessing the electrical fire risk of an urban underground comprehensive pipe gallery, and the specific implementation process is as follows.
[0032] 1. Preliminary assessment of electrical fire risks in urban underground utility corridors
[0033] According to the sampling period, sensors are used to collect the total volatile organic compound concentration, methane concentration and temperature in the underground comprehensive pipeline corridor.
[0034] The volatile organic compounds include: methyl chloride, dichloroethylene, chloroform, carbon tetrachloride and trichloroethylene. In one embodiment, the total volatile organic compound and indoor air quality sensor ZMOD4410 produced by Renesas Semiconductor is used to detect the total volatile organic compound concentration in the underground utility tunnel.
[0035] The total volatile organic compound concentration C VOCs , methane concentration The temperature T in the underground utility tunnel is respectively related to the standard total volatile organic compound concentration C′ VOCs , Standard methane concentration Compare it with the standard temperature T′ in the underground comprehensive pipeline corridor.
[0036] If satisfied: The electrical fire risk level in the underground pipe gallery is judged to be level 0, which means there is no risk of electrical fire in the underground integrated pipe gallery. At this time, the total volatile organic compound concentration C is continuously detected according to the set sampling period. VOCs , methane concentration And the temperature T in the underground integrated pipeline corridor and conduct a risk assessment.
[0037] If satisfied: This means that there is a risk of electrical fire in the underground utility corridor. At this time, the CO sensor, CO2 sensor and ozone sensor are turned on at the same time. And according to the sampling period, the CO concentration, CO2 concentration and ozone concentration in the underground utility corridor are collected.
[0038] The standard temperature in the underground utility corridor is set at 20℃~25℃; the standard total volatile organic compound concentration C′ VOCs , Standard methane concentration They are the concentration values detected by the total volatile organic compound sensor and the methane sensor at standard temperature respectively.
[0039] Here, the result of adding the ratio of the measured parameters to the standard parameters is used for judgment, which can avoid the situation where a sensor fails and causes a low detection value, so as to avoid missing risk situations as much as possible.
[0040] 2. Further judge the risk of electrical fire in underground utility corridors and classify the risk level of fire in underground utility corridors
[0041] In this embodiment, a deep neural network is used as a network model to evaluate the electrical fire risk of underground utility corridors. The specific process is as follows.
[0042] A deep neural network is constructed and trained to obtain an underground utility corridor electrical fire risk assessment model.
[0043] A deep neural network is a fully connected neural network model with deep learning technology. Neurons are its most basic processing structural unit. A neuron model consists of an input signal (x n ), weight (w n ), adder and activation function (f(·)), and each neuron is a multi-input and single-output information processing unit, and its mathematical expression is:
[0044]
[0045] Among them, w n Represents input x n The contribution of bias b to the output is to adjust the input of the activation function.
[0046] Neurons read information from neurons in the previous layer through one or more inputs, and each input is assigned a certain weight for information weighting. The weighted linear combination information is output after passing through an activation function. The weight parameters and bias values of the model are obtained through learning. They can control the intensity of input data transmitted to neurons. The activation function performs activation processing inside the neuron, and the final output result becomes the input of the next layer of the network.
[0047] The activation function performs nonlinear operations on the weighted sum of the input signals, which makes the multi-layer structure of the network more practical. The activation function is very important, as it helps the network learn arbitrary mapping relationships between complex input data and output.
[0048] Cross entropy is used as the loss function. Cross entropy is a widely used loss function in classification problems and is defined as follows:
[0049]
[0050] Among them, p and q represent the target and output probability distributions respectively, n represents the vector dimension, and x i Table vector value in the i-th dimension.
[0051] Cross entropy describes the distance between two probability distributions, but the output of the neural network may not conform to the probability distribution (the sum of the probabilities in each dimension is not 1). In order to effectively process the output of the neural network into a probability distribution, Softmax regression (normalized exponential function) is used to solve:
[0052]
[0053] Among them, y j Represents the output of the neural network.
[0054] The deep neural network model constructed in this embodiment is supervised learning with labeled data. The input vectors {X1, X2, X3, X4, X5, X6} are parameters (total volatile organic compound concentration C VOCs , methane concentration and the temperature T and CO concentration C in the underground utility corridor CO , CO2 concentration and ozone concentration ) is the value obtained by normalization.
[0055] Among them, the normalization formula is:
[0056]
[0057] Among them, x i represents the measured value of each parameter, x i ′ represents the standard value of each parameter; i=1, 2, 3…, 6 represent the total volatile organic compound concentration, methane concentration, temperature, CO concentration, CO2 concentration and ozone concentration respectively.
[0058] The output vector {y1, y2, y3, y4} respectively indicates that the risk of electrical fire is level 1, level 2, level 3 and level 4. Among them, level 0 represents no risk of electrical fire, and levels 1, 2, 3 and 4 all represent the risk of electrical fire, and the risk of electrical fire increases from level 1 to level 4.
[0059] Since the risk of electrical fire is non-numeric, the overheating level is encoded into a 4-dimensional array through one-hot encoding; when the risk level is judged to be level n, the number in the nth dimension is 1, and the numbers in the remaining dimensions are 0. For example, if the risk level is level 1, it will be represented as [1,0,0,0] in the deep neural network operation; if the risk level is level 1, it will be represented as [0,1,0,0] in the deep neural network operation.
[0060] In this embodiment, the number of hidden layers of the deep neural network is 5, and the number of neuron nodes in each hidden layer is the same.
[0061] As a preferred embodiment, the number of neuron nodes in the hidden layer is determined by the following formula:
[0062]
[0063] Among them, N hid is the number of hidden layer neuron nodes, N s is the number of samples in the training set, N in is the number of neuron nodes in the input layer, N out is the number of neuron nodes in the output layer, α is the adjustment variable, and α represents any integer between 2 and 10.
[0064] The training set was constructed by experimental means. The cables used in the underground pipe gallery were heated in a closed space to simulate the actual heating conditions of the cables in the underground pipe gallery. The degree of fire at different heating levels was collected, and the critical point of open fire was set at level 4; the risks of other levels were gradually reduced.
[0065] Since the loss value of the model decreases faster and the accuracy rate improves more significantly when the Adam algorithm is used for parameter optimization, the Adam algorithm is used to train the deep neural network in this embodiment.
[0066] After the training is completed, the underground utility corridor electrical fire risk assessment model is obtained, which can be applied to the electrical fire risk assessment of urban underground utility corridors.
[0067] According to the sampling cycle, the total volatile organic compound concentration, methane concentration, temperature, CO concentration, CO2 concentration and ozone concentration in the underground utility corridor are collected and normalized and input into the underground utility corridor electrical fire risk assessment model. In this embodiment, the sampling cycle is set to 1 minute. The underground utility corridor electrical fire risk assessment model outputs the assessed risk level in real time, thereby realizing the real-time risk assessment of underground utility corridor electrical fires, so as to facilitate the timely discovery of risks before or in the early stage of a fire, and take different countermeasures according to the corresponding risk levels.
[0068] The present invention adopts a two-stage risk assessment method. In the initial judgment, fewer parameters are used, which can improve the operation speed of the system and the judgment speed; in the further judgment, more parameters are used, which can improve the accuracy of the judgment result and avoid misjudgment.
[0069] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the embodiments shown and described herein.
Claims
1. A method for assessing the electrical fire risk of an urban underground utility corridor, characterized in that: The steps include: Step 1: According to the sampling period, the total volatile organic compound concentration, methane concentration and temperature in the underground comprehensive pipeline corridor are collected through sensors; The total volatile organic compound concentration C VOCs , methane concentration The temperature T in the underground utility tunnel is respectively related to the standard total volatile organic compound concentration C′ VOCs , Standard methane concentration Compare with the standard temperature T′ in the underground utility corridor; If satisfied: The electrical fire risk level in the underground pipe gallery is judged to be level 0; If satisfied: The CO sensor, CO2 sensor and ozone sensor are turned on at the same time; and the CO concentration, CO2 concentration and ozone concentration in the underground integrated pipe gallery are collected according to the sampling period; Step 2: According to the total volatile organic compound concentration, methane concentration, temperature in the underground utility corridor, CO concentration, CO2 concentration and ozone concentration, the electrical fire risk of the underground utility corridor is judged, and the electrical fire risk level of the underground utility corridor is divided into Level 1, Level 2, Level 3 and Level 4; Among them, level 0 means there is no risk of electrical fire, and levels 1, 2, 3 and 4 all mean there is a risk of electrical fire, and the risk of electrical fire increases from level 1 to level 4.
2. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 1, characterized in that: The volatile organic compounds include: methyl chloride, dichloroethylene, chloroform, carbon tetrachloride and trichloroethylene.
3. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 2 is characterized in that: In step 2, the method for judging the electrical fire risk of the underground utility corridor is: Constructing a deep neural network and training the deep neural network to obtain an underground utility corridor electrical fire risk assessment model; The total volatile organic compound concentration C VOCs , methane concentration and the temperature T and CO concentration C in the underground utility corridor CO , CO2 concentration and ozone concentration After normalization processing, the underground utility corridor electrical fire risk assessment model is input, and the underground utility corridor electrical fire risk assessment model outputs the underground utility corridor electrical fire risk level.
4. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 3 is characterized in that: The total volatile organic compound concentration C VOCs , methane concentration and the temperature T and CO concentration C in the underground utility corridor CO , CO2 concentration and ozone concentration The formula for normalization is: Among them, x i represents the measured value of each parameter, x i ′ represents the standard value of each parameter; i=1, 2, 3…, 6 represent the total volatile organic compound concentration, methane concentration, temperature, CO concentration, CO2 concentration and ozone concentration respectively.
5. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 4 is characterized in that: The standard temperature in the underground comprehensive pipeline corridor is set at 20℃~25℃; the standard values of other parameters are the actual measured values of other parameters at the standard temperature.
6. The method for assessing electrical fire risk in an urban underground utility corridor according to any one of claims 3 to 5, characterized in that: The number of hidden layers of the deep neural network is 5.
7. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 6 is characterized in that: The number of neuron nodes in each hidden layer is the same.
8. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 7, characterized in that: The number of neuron nodes in the hidden layer is determined by the following formula: Among them, N hid is the number of hidden layer neuron nodes, N s is the number of samples in the training set, N in is the number of neuron nodes in the input layer, N out is the number of neuron nodes in the output layer, α is the adjustment variable, and α represents any integer between 2 and 10.
9. The method for assessing electrical fire risk in an urban underground utility corridor according to claim 8, characterized in that: The sampling period of total volatile organic compound concentration, methane concentration, temperature in the underground integrated pipeline gallery, CO concentration, CO2 concentration and ozone concentration was set to 1 min.