Urban pipe gallery fire risk assessment system based on big data driving
By deploying microbial fuel cells and multi-layer neural network models in urban pipeline corridors and integrating traditional monitoring data, the problem of insufficient data integration and analysis capabilities in urban pipeline fire risk assessment is solved, and efficient and intelligent fire risk assessment and prediction are achieved.
Patent Information
- Application Number
- CN202510225744.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-27
- Publication Date
- 2025-06-13
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the current fire risk assessment of urban pipeline corridors, data integration and analysis capabilities are insufficient, traditional methods cannot fully reflect the actual situation in the pipeline corridor, lack intelligent evaluation methods, and evaluation efficiency is low and costly.
A fire risk assessment system for urban pipeline corridors is adopted based on big data, including microbial energy monitoring module, fire risk characteristic analysis module and fire risk multi-domain assessment module. Through the integration of microbial fuel cell monitoring data and traditional monitoring equipment data, a fire risk assessment model is built using dark matter algorithms and multi-layer neural networks to achieve comprehensive utilization and accurate prediction of multi-domain data.
It improves the efficiency and accuracy of fire risk assessment, provides comprehensive and intelligent evaluation methods, reduces assessment costs, and enhances the ability to predict fire risks in the pipeline corridor.
Smart Images
Figure CN120146572A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of urban utility tunnel safety, and particularly to an urban utility tunnel fire risk assessment system driven by big data. Background Art
[0002] In the current urban utility tunnel fire risk assessment, the following problems still exist: insufficient data integration and analysis capabilities, mainly including: traditional fire risk assessments often rely only on a limited number of monitoring devices, such as temperature sensors, smoke sensors, etc. The data provided by these devices has limited dimensions and may not comprehensively reflect the actual situation inside the utility tunnel; poor adaptability to complex environments. The environment of urban utility tunnels is complex and changeable, and various factors such as temperature, humidity, and types of organic substances may affect the fire risk. Previous methods cannot accurately reflect the interactions and influences among various factors inside the utility tunnel; lack of intelligent fire risk assessment means. Subjective judgment differences may exist in the manual assessment process, resulting in inconsistencies and uncertainties in the assessment results. Traditional assessment means may only assess the current fire risk and are difficult to predict future risks; low efficiency of fire risk assessment. Previous assessment means require a large amount of human, material, and financial resources, resulting in high assessment costs and low efficiency. Or due to simple models, fixed parameters, etc., the assessment results may be inaccurate and cannot truly reflect the fire risk situation inside the utility tunnel. Therefore, a series of targeted methods are urgently needed to address these problems to improve the efficiency and level of urban utility tunnel fire risk assessment. Summary of the Invention
[0003] The purpose of the present invention is to solve the problems in the background art and propose an urban utility tunnel fire risk assessment system driven by big data.
[0004] To achieve the above purpose, the present invention adopts the following technical solutions: An urban utility tunnel fire risk assessment system driven by big data includes: a microbial energy monitoring module, a fire risk characteristic analysis module, and a fire risk multi-domain assessment module; Microbial energy monitoring module: Select and deploy microbial fuel cells according to the environmental characteristics of urban utility tunnels, and evenly distribute them at key positions to ensure full coverage and no interference with each other; allocate data acquisition components to the microbial fuel cells and establish a data transmission network combining wired and wireless; preprocess and preliminarily analyze the data collected from the microbial fuel cells as the basis for in-depth analysis; among them, the microbial energy monitoring module includes a microbial fuel cell selection and deployment unit, a data acquisition and data transmission establishment unit, and a data preprocessing and preliminary analysis unit; Fire risk characteristic analysis module: Integrate the data from preprocessing and preliminary analysis with the data from traditional monitoring devices to create a dataset of feature vectors containing multi-dimensional information; Use the dark matter algorithm to calculate the non-linear associations between feature vectors, mine the association patterns highly correlated with fire risk, and classify and quantify fire risk characteristics; Among them, the fire risk characteristic analysis module includes a data integration and feature vector construction unit, and a fire risk characteristic mining unit. Fire risk multi-domain assessment module: Integrate and process the data from different analysis domains, that is, define an analysis domain including a three-dimensional physical space analysis domain, a frequency analysis domain of microbial fuel cell data, and an information entropy analysis domain, and map them into a risk assessment model to comprehensively evaluate the fire risk in the utility tunnel; Use a multi-layer neural network structure to construct a fire risk assessment model, and through training and optimization, enable the model to accurately predict the fire risk in each area of the utility tunnel; Among them, the fire risk multi-domain assessment module includes an analysis domain definition and data mapping unit, and a risk assessment model construction unit.
[0005] Furthermore, the process of the microbial fuel cell selection and deployment unit for selecting and deploying microbial fuel cells according to the environmental characteristics of the urban utility tunnel includes: Select the corresponding type of microbial fuel cell according to the environmental characteristics of the urban utility tunnel; Among them, the environmental characteristics of the utility tunnel include the types of organic substances present in the utility tunnel, the temperature range, and the humidity conditions. Deploy them in the utility tunnel according to the layout rules, that is, evenly distribute microbial fuel cells at the key positions in the utility tunnel, and the distance between each microbial fuel cell is determined according to the length and width of the utility tunnel, the environmental complexity, the change of the monitoring radius, and the calculation of the temperature and humidity fluctuations in the utility tunnel; Among them, the key positions in the utility tunnel include areas with dense cables, near flammable storage points, and around ventilation openings.
[0006] Furthermore, the process of the data acquisition and data transmission establishment unit for allocating data acquisition components to microbial fuel cells and establishing a data transmission network combining wired and wireless includes: Allocate data acquisition components to each microbial fuel cell for real-time acquisition of microbial fuel cell data. Use a transmission method combining wired and wireless to transmit the acquired data to the data preprocessing and preliminary analysis unit.
[0007] Furthermore, the process of the data preprocessing and preliminary analysis unit for preprocessing and preliminary analysis of the acquired microbial fuel cell data includes: Preprocess the acquired microbial fuel cell data, that is, perform data cleaning and normalization processing. By performing local weighted regression analysis and trend fluctuation analysis on the preprocessed data, the stability of the environment inside the utility tunnel is initially judged. If it is found that the fluctuations in the data of the microbial fuel cell exceed the preset threshold, it indicates that abnormal changes have occurred in environmental factors, and the preliminary analysis results obtained will be marked and stored.
[0008] Furthermore, the data integration and feature vector construction unit is used to integrate the data from preprocessing and preliminary analysis with the data of traditional monitoring devices. The process of creating a feature vector dataset containing multi-dimensional information includes: Obtain the data preprocessed and preliminarily analyzed by the data preprocessing and preliminary analysis unit; Obtain the data monitored by traditional monitoring devices inside the utility tunnel; Integrate the preprocessed and preliminarily analyzed data with the data of other traditional monitoring devices inside the utility tunnel to jointly form a dataset; Construct feature vectors from the integrated dataset, and each feature vector contains information in multiple dimensions; among them, other traditional monitoring devices include temperature sensors and smoke sensors, and the information in multiple dimensions includes the voltage and current of the microbial fuel cell, the readings of the temperature sensor, and the readings of the smoke sensor.
[0009] Furthermore, the fire risk feature mining unit is used to calculate the non-linear associations between feature vectors using the dark matter algorithm, mine the association patterns highly correlated with fire risk, and the process of classifying and quantifying fire risk features includes: Establish an association function on the entire feature vector dataset for global search to find existing hidden association patterns; According to the pre-set fire risk-related indicators, screen out the association patterns highly correlated with fire risk; For the screened association patterns, further analyze their performance in the time and space scales to obtain fire risk features; Classify and quantify the mined fire risk features, and assign weights and thresholds according to their influence degree on fire risk.
[0010] Furthermore, the analysis domain definition and data mapping unit is used to integrate and process data from different analysis domains and map them into the risk assessment model. The process includes: Clarify each analysis domain in the fire risk assessment model, including the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain: On the three-dimensional physical space analysis domain, use three-dimensional physical space coordinate data to represent the position of the utility tunnel; In the frequency analysis domain, spectral analysis is performed on the monitoring data of the microbial fuel cell to convert the time-domain data into frequency-domain data, and the amplitude and phase information of different frequency components are calculated, that is, the time series data of the monitoring data of the microbial fuel cell is subjected to a fast Fourier transform to obtain the distribution of the monitoring data of the microbial fuel cell at different frequencies; In the information entropy analysis domain, the information entropy of the microbial fuel cell data is calculated and analyzed; The data in each analysis domain is mapped into the risk assessment model.
[0011] Furthermore, the process of using the multi-layer neural network structure to construct the fire risk assessment model by the risk assessment model construction unit includes: The multi-layer neural network structure includes three main parts: an input layer, a hidden layer, and an output layer; Among them, the input layer receives data from different analysis domains, including the three-dimensional spatial coordinates of the pipe gallery, the frequency and information entropy information of the microbial fuel cell data; The hidden layer performs a non-linear transformation on the input data through the connection weights and activation functions between neurons; The output layer outputs the fire risk assessment result, which represents the risk degree of fire occurrence in each area in the pipe gallery in the form of probability, and the output value is between 0 and 1, where 0 indicates no fire risk and 1 indicates that a fire is about to occur; It can be understood that a large amount of historical data (including data during normal operation of the pipe gallery and data on past fires or fire hazards) is collected as a training set, and these data are input into the constructed model according to the analysis domain definition and the analysis domain mapping method of the data mapping unit. By adjusting the parameters of the model (such as the connection weights between neurons), the output result of the model is made to match the actual fire risk situation as closely as possible; during the training process, methods such as cross-validation are used to prevent overfitting of the model. The training set is divided into multiple subsets, and the subsets are used alternately as the validation set to monitor the performance of the model on the validation set. According to the validation results, the structure and parameters of the model are adjusted, such as increasing or decreasing the number of neurons in the hidden layer, adjusting the activation function, etc., to improve the generalization ability of the model; the model is continuously updated and optimized. As new data accumulates, the model is retrained regularly to enable the model to adapt to changes in the pipe gallery environment and the increasing requirements for fire risk assessment.
[0012] Compared with the existing technologies, the advantages of the urban pipe gallery fire risk assessment system based on big data drive provided by the present invention are as follows: 1. By selecting and deploying microbial fuel cells according to the environmental characteristics of the urban pipe gallery, data acquisition components are allocated to the microbial fuel cells, and the collected microbial fuel cell data is preprocessed and preliminarily analyzed, the present invention efficiently utilizes environmental resources, monitors environmental changes in real time, and reduces maintenance costs; 2. By integrating the data from preprocessing and preliminary analysis with the data from traditional monitoring devices, the present invention creates a feature vector dataset containing multi-dimensional information, calculates the non-linear associations between feature vectors using the dark matter algorithm, mines the association patterns highly relevant to fire risks, classifies and quantifies the fire risk features, providing a comprehensive and accurate data basis for fire risk analysis for subsequent precise risk assessment. 3. By integrating and processing the data from different analysis domains, that is, defining an analysis domain including three-dimensional physical space, a frequency analysis domain of microbial fuel cell data, and an information entropy analysis domain, and mapping them into a risk assessment model to comprehensively evaluate the fire risks in the utility tunnel; meanwhile, a multi-layer neural network structure is adopted to construct a fire risk assessment model; this multi-domain assessment method can more comprehensively reflect the fire risk situation in the utility tunnel, helping management personnel take preventive measures in a timely manner and reducing the possibility of fire occurrence. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 It is a module diagram of the urban utility tunnel fire risk assessment system driven by big data proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0014] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0015] Refer to Figure 1 , for the urban utility tunnel fire risk assessment system driven by big data, the system includes a microbial energy monitoring module, a fire risk feature analysis module, and a fire risk multi-domain assessment module; Microbial energy monitoring module: Select and deploy microbial fuel cells according to the environmental characteristics of the urban utility tunnel, evenly distribute them at key positions to ensure full coverage and no interference with each other; allocate data acquisition components for the microbial fuel cells and establish a data transmission network combining wired and wireless; preprocess and preliminarily analyze the collected microbial fuel cell data as the basis for in-depth analysis; among them, the microbial energy monitoring module includes a microbial fuel cell selection and deployment unit, a data acquisition and data transmission establishment unit, and a data preprocessing and preliminary analysis unit; Fire risk characteristic analysis module: Integrate the data from preprocessing and preliminary analysis with the data from traditional monitoring devices to create a dataset of feature vectors containing multi-dimensional information; Use the dark matter algorithm to calculate the non-linear associations between feature vectors, mine the association patterns highly correlated with fire risk, classify and quantify the fire risk characteristics, and assign weights and thresholds; Among them, the fire risk characteristic analysis module includes a data integration and feature vector construction unit, and a fire risk characteristic mining unit. Fire risk multi-domain assessment module: Integrate and process the data from different analysis domains, that is, define the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain, and map them into the risk assessment model to comprehensively evaluate the fire risk in the utility tunnel; Use a multi-layer neural network structure to construct a fire risk assessment model, and through training and optimization, enable the model to accurately predict the fire risk in each area of the utility tunnel; Among them, the fire risk multi-domain assessment module includes an analysis domain definition and data mapping unit, and a risk assessment model construction unit.
[0016] The microbial energy monitoring module selects and deploys microbial fuel cells according to the environmental characteristics of the urban utility tunnel; Allocate data acquisition components for the microbial fuel cells and establish a data transmission network combining wired and wireless; The steps for preprocessing and preliminary analysis of the collected microbial fuel cell data include: Step 101: According to the environmental characteristics of the urban utility tunnel (considering factors such as the types of organic substances, temperature range, and humidity conditions that may exist in the tunnel), select the corresponding type of microbial fuel cell (select a microbial fuel cell with higher power generation efficiency and stability. For example, for a utility tunnel area with relatively low organic content but high humidity, select a microbial fuel cell model with low substrate requirements and resistance to humid environments). Step 102: Deploy in the utility tunnel according to the layout rules, that is, evenly distribute the microbial fuel cells at key positions in the tunnel (such as near cable-intensive areas, flammable material storage points, and around ventilation openings), and the spacing between each microbial fuel cell is determined according to the length and width of the tunnel, environmental complexity, changes in the monitoring radius, and calculations of temperature and humidity fluctuations in the tunnel to ensure that all areas in the tunnel can be comprehensively covered and there is no interference in data collection between them. It can be understood that the microbial fuel cell, that is, MFC, is a device that directly converts the chemical energy in organic substances into electrical energy using microorganisms, and has multiple application potentials such as wastewater treatment, environmental purification, and power supply for low-power devices. Step 103: Allocate data acquisition components for each microbial fuel cell to obtain microbial fuel cell data in real time. It is understandable that the data acquisition component features high resolution and fast response, ensuring that minute changes in microbial fuel cell data can be captured. Among them, microbial fuel cell data includes, but is not limited to, electrical parameters (i.e., the output voltage and current of the microbial fuel cell), microbiological parameters (i.e., the microbial community structure and microbial activity in the microbial fuel cell), physical parameters (i.e., temperature and pressure during the operation of the microbial fuel cell), etc.; Step 104: Transmit the collected data to the data preprocessing and preliminary analysis unit by using a combination of wired and wireless transmission methods. Specifically, the combination of wired and wireless transmission methods is as follows: Lay an optical fiber network in the pipe gallery as the main transmission trunk line. In areas where wiring is difficult, use a low-power and high-reliability wireless sensor network for data transmission to ensure the real-time and integrity of the data; Step 105: Preprocess the collected microbial fuel cell data, that is, perform data cleaning and normalization processing; Step 106: Conduct local weighted regression analysis and trend fluctuation analysis on the preprocessed data to preliminarily judge the stability of the environment in the pipe gallery. If it is found that the fluctuation of the microbial fuel cell data exceeds the preset threshold, it indicates that abnormal changes have occurred in environmental factors, which may be related to fire risks. Mark and store the obtained preliminary analysis results as the basis for subsequent in-depth analysis; In Steps 101 - 106, taking electrical parameters (i.e., the output voltage and current of the microbial fuel cell) as an example: G11: Deployment of microbial fuel cells Obtain the length and width of the pipe gallery, and mark the length and width of the pipe gallery as ; Assume that under ideal conditions, the effective monitoring radius of each microbial fuel cell is , To ensure full coverage and no interference, the deployment spacing of the microbial fuel cells satisfies: , where is the environmental complexity coefficient, which is a dimensionless coefficient comprehensively considering factors such as the equipment layout and channel complexity in the pipe gallery, and its value range is , The larger the value, the more complex the environment; is the correction coefficient, and , which is used to consider factors such as redundancy and interference avoidance in the actual layout; is the number of microbial fuel cells; is the standard deviation of the monitoring radius of the microbial fuel cell; is the average value of the monitoring radius of the microbial fuel cell; is the temperature fluctuation range in the pipe gallery; is the average temperature in the pipe gallery; is the average humidity in the pipe gallery; G12, Electrical Parameter Acquisition and Transmission: Obtain the output voltage and current of the microbial fuel cell, and label the output voltage and current of the microbial fuel cell as , where represents time; The resolution of the data acquisition component for voltage and current is respectively , and satisfies: , In the formula, the resolution is set to be within one percent of the minimum value of the acquired electrical parameters to ensure that small changes can be captured; If the reliability of fiber optic network transmission is , and the reliability of wireless sensor network transmission is , since a combination of wired and wireless methods is adopted, the reliability of the entire data transmission is expressed as: , where represents the transmission link stability index; represents the proportion of data transmitted through the fiber optic network in the total transmitted data, and ; G13, Data Preprocessing and Preliminary Analysis: Obtain the voltage sequence data of the microbial fuel cell collected by the data acquisition component , the current sequence data , and the time sequence , where represents the number of voltage or current data points collected; Use the standard deviation method to detect outliers in the voltage sequence data of the microbial fuel cell: Set the average value of the voltage data as , and the standard deviation of the voltage data as ; If any voltage data satisfies , then it is considered that is a voltage outlier and is removed from the voltage data sequence, where is the index of the voltage or current data of the microbial fuel cell, and ; The outlier detection of the current sequence data is the same as above, and current outliers are obtained; Use the linear normalization method to obtain the normalized voltage data and current data of the microbial fuel cell: , , where respectively represent the minimum and maximum voltages of the microbial fuel cell, respectively represent the minimum and maximum currents of the microbial fuel cell; Perform locally weighted regression on the voltage data of the microbial fuel cell to fit the changing trend of the voltage of the microbial fuel cell over time: Set the voltage fitting function : , where, is the order of the polynomial in the voltage fitting function , is the term number index of the polynomial in the voltage fitting function , and (for example, when , this term is , representing the constant term, when , this term is , representing the first-order term; in the process of calculating the locally weighted regression, by determining the coefficients corresponding to different values to construct a function that fits the changing trend of the voltage over time); among them, is the coefficient that changes over time; For each voltage data point of the microbial fuel cell, calculate its weight using the tricube function: , where, is the bandwidth parameter, which affects the smoothness of the regression curve. If the bandwidth parameter is smaller, it will cause the regression curve to be closer to the original voltage data points, but may increase the risk of noise and overfitting; if the bandwidth parameter is larger, the regression curve will be smoother. At the same time, the bandwidth parameter determines the shape and range of the tricube function; Solve by the least squares method to make the smallest; Calculate the fluctuation index of the voltage data trend of the microbial fuel cell: , where, represents the index in the time series. In the trend fluctuation analysis, ranges from 1 to for summing the differences in the voltage change rates at adjacent time points to obtain the fluctuation index of the voltage data trend; among them, , ; At the same time, calculate the deviation degree of the voltage data of the microbial fuel cell relative to the trend : ; Using the same method as for the voltage data of the microbial fuel cell, obtain the trend fluctuation index of the current data of the microbial fuel cell and the degree of deviation relative to the trend ; Define the environmental stability index : , where, is the weight factor, and (for example ), which is used to adjust the relative importance of the voltage and current data of the microbial fuel cell in the determination of environmental stability; When the value of the environmental stability index is close to 1, it is determined that the environment is relatively stable; when the value of the environmental stability index is far from 1, it is determined that there may be abnormal changes in the environment, which may be related to the fire risk.
[0017] The fire risk characteristic analysis module integrates the data from preprocessing and preliminary analysis with the data from traditional monitoring devices, creating a feature vector data set containing multi-dimensional information; the steps of using the dark matter algorithm to calculate the non-linear associations between feature vectors, mining the association patterns highly related to the fire risk, classifying and quantifying the fire risk characteristics, and assigning weights and thresholds include: Step 201: Integrate the data from preprocessing and preliminary analysis with the data from other traditional monitoring devices in the pipe gallery (such as temperature sensors, smoke sensors, etc.) to jointly form a data set that covers various environmental information in the pipe gallery; Step 202: Construct feature vectors from the integrated data set, where each feature vector contains information in multiple dimensions, such as the voltage and current of the microbial fuel cell, the readings of the temperature sensor, the readings of the smoke sensor, etc., ensuring that the feature vectors can comprehensively reflect the environmental state in the pipe gallery; Step 203: Establish an association function on the entire feature vector data set for global search to find the existing hidden association patterns; Step 204: According to the pre-set fire risk-related indicators, screen out the association patterns highly related to the fire risk; among them, the pre-set fire risk-related indicators are important bases for screening the association patterns highly related to the fire risk, aiming to reflect the potential relationship between the environmental state in the pipe gallery and the fire risk, including but not limited to environmental parameter categories (such as temperature, smoke concentration, harmful gas concentration), equipment operation status categories (such as electrical equipment parameters, fire protection equipment status); Step 205: For the selected correlation patterns, further analyze their performance at different time and spatial scales to obtain fire risk characteristics. For example, some correlation patterns may be particularly evident in specific areas of the pipe gallery within a few hours before a fire. These correlation patterns that are closely related to fire risk under specific spatio-temporal conditions are defined as fire risk characteristics; Step 206: Classify and quantify the excavated fire risk characteristics, and assign different weights and thresholds according to their impact on fire risk, so as to accurately reflect their importance in the subsequent risk assessment model. It can be understood that in the dark matter algorithm, hidden relationships are found by calculating the complex correlation function between feature vectors, and this correlation function can capture the non-linear and weak correlations between data. For example, it may find a subtle relationship between the current change of a microbial fuel cell and the smoke concentration at a certain distant location, and this relationship may start to appear before a fire occurs; In Steps 201 - 206, for example: G21: Data integration and feature vector creation: Set the data obtained through preprocessing and preliminary analysis by the data preprocessing and preliminary analysis unit as the set , where represents the th data after being processed and analyzed by the data preprocessing and preliminary analysis unit, and , is the data index after being processed and analyzed by the data preprocessing and preliminary analysis unit, is the number of data after being processed and analyzed by the data preprocessing and preliminary analysis unit; Set the data monitored by traditional monitoring devices (such as temperature sensors, smoke sensors, etc.) in the pipe gallery as the set , where represents the th data monitored by traditional monitoring devices in the pipe gallery, and , is the data index monitored by traditional monitoring devices in the pipe gallery, is the number of data monitored by traditional monitoring devices in the pipe gallery; Integrate the data obtained through preprocessing and preliminary analysis by the data preprocessing and preliminary analysis unit and the data monitored by traditional monitoring devices in the pipe gallery to obtain the set : , where represents the th data after integration, and , is the data index after integration, is the number of data after integration, and ; Based on the integrated dataset, construct feature vectors: In the formula, represents different feature vector indices, is the dimension of the feature vector; G22, the correlation function in the dark matter algorithm: The correlation function is defined as: , in the formula, is the weight parameter of the correlation function, representing the importance of the relationship between dimension and dimension , and ; is a non - linear function used to capture the non - linear relationship between , for example ; G23. Use the dark matter algorithm for in - depth mining: Perform a global search on the entire feature vector dataset to find correlation patterns, where is the quantity of the entire feature vector data; Assume there is a correlation pattern , and the probability of its existence is related to the correlation function: , in the formula, is an indicator function. If there is a correlation pattern between , then , otherwise ; The preset set of fire risk - related indicators is: , where is the quantity of fire risk - related indicators, is the index of fire risk - related indicators, and ; Establish the correlation function between the correlation pattern and the fire risk as: , in the formula, is the weight coefficient of the correlation function , representing the importance of the fire risk - related indicator in the correlation assessment; is a function used to measure the relationship between the fire risk - related indicator and the correlation pattern. For example, if is the smoke concentration threshold, which can be the probability that the smoke concentration exceeds the threshold when the correlation pattern exists; Analyze the performance at different spatio-temporal scales, and label the time scale and space scale as ; For the correlation pattern , define its performance function at the spatio-temporal scale as: , where is a function describing the performance of the correlation pattern in time and space. For example, if the correlation pattern is related to the temperature change, then is the relationship function between the temperature change rate in time and space and the correlation pattern; respectively represent the integral variables of ; Obtain the fire risk characteristics based on the analysis results of the spatio-temporal scale, and organize the fire risk characteristics to obtain the fire risk characteristic set , where is the number of fire risk characteristics, is the fire risk characteristic index, represents the th fire risk characteristic; Set the classification function , whose value range is the category set , where is the total number of categories. For example, represents the risk characteristic category related to temperature, represents the risk characteristic category related to smoke, etc.; Assume that a classification method based on physical meaning is adopted. For the fire risk characteristic , if is mainly related to temperature, for example, it is the performance of the correlation pattern between the temperature change rate in a certain area and other factors at a specific spatio-temporal scale, then ; Define the fire risk monitoring function as: , where is the weight factor of the fire risk characteristic. The determination of the weight factor is based on the relative importance of the fire risk characteristic to the fire risk. For example, , where are the influencing factors in space and time respectively, is the fire risk characteristic 's influence range in space (for example, if is the temperature correlation pattern of a certain local area, with a small influence range, then the value of is relatively small; if it is the average temperature correlation pattern of the entire utility tunnel, then is the fire risk characteristic in terms of the early warning ability in time (for example, if is the correlation pattern that starts to change several hours before a fire occurs, then the value of is relatively large; if it is the correlation pattern that becomes obvious only when a fire occurs, then is the fire risk characteristic of the danger threshold, and the determination of the threshold is based on historical data, safety standards, and the specific situation of the utility tunnel; is the indicator function for fire risk monitoring. If the fire risk characteristic meets the danger threshold and the classification function , for example, is the smoke concentration, is the danger threshold of the smoke concentration, is the risk characteristic category related to the smoke concentration, then there is , otherwise .
[0018] The multi-domain fire risk assessment module integrates and processes data from different analysis domains, that is, it defines the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain, and maps them into the risk assessment model; the steps of constructing a fire risk assessment model using a multi-layer neural network structure include: Step 301, clarify each analysis domain in the fire risk assessment model, including the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain; Step 302, on the three-dimensional physical space analysis domain, use three-dimensional physical space coordinate data for the position representation of the utility tunnel; Step 303, on the frequency analysis domain, perform spectrum analysis on the microbial fuel cell monitoring data, convert the time-domain data into frequency-domain data, calculate the amplitude and phase information of different frequency components, that is, perform a fast Fourier transform on the time-series data of the microbial fuel cell monitoring data to obtain the distribution of the microbial fuel cell monitoring data at different frequencies, and the changes in different frequencies correspond to different fire risk states; Step 304: Calculate the information entropy of the microbial fuel cell data in the information entropy analysis domain. It can be understood that the information entropy reflects the uncertainty and chaos degree of the data. For the environmental data in the pipe gallery, the high information entropy area may mean the existence of more unknown and complex fire risk factors. Step 305: Map the data of each analysis domain into the risk assessment model. Step 306: The multi-layer neural network structure includes three main parts: the input layer, the hidden layer, and the output layer. Among them, the input layer receives data from different analysis domains, including the three-dimensional space coordinates of the pipe gallery, the frequency and information entropy information of the microbial fuel cell data. The hidden layer performs a non-linear transformation on the input data through the connection weights and activation functions between neurons. It can be understood that multiple neuron nodes are set in the hidden layer, and each node is connected to the neurons of the previous layer through weights and biases to learn the complex relationships between the data of different analysis domains. The output layer outputs the fire risk assessment result, which represents the risk degree of fire occurrence in each area of the pipe gallery in the form of probability, and the output value is between 0 and 1. 0 indicates no fire risk, and 1 indicates that a fire is about to occur. Among them, because the output layer outputs a numerical value representing the fire risk, there is only one neuron in the output layer here. It can be understood that a large amount of historical data (including the data during the normal operation of the pipe gallery and the data of past fires or fire hazards) is collected as the training set, and these data are input into the constructed model according to the analysis domain definition and the analysis domain mapping method of the data mapping unit. By adjusting the parameters of the model (such as the connection weights between neurons), the output result of the model is made to match the actual fire risk situation as much as possible. During the training process, methods such as cross-validation are used to prevent the model from overfitting. The training set is divided into multiple subsets, and the subsets are used alternately as the validation set to monitor the performance of the model on the validation set. According to the validation results, the structure and parameters of the model are adjusted, such as increasing or decreasing the number of neurons in the hidden layer, adjusting the activation function, etc., to improve the generalization ability of the model. The model is continuously updated and optimized. As new data accumulates, the model is retrained regularly to enable the model to adapt to the changes in the pipe gallery environment and the improvement of the fire risk assessment requirements. In Steps 301 - 306, for example: G31: Set the pipe gallery space , and discretize it into monitoring points. For the th monitoring point , its position in the pipe gallery is represented by the three-dimensional coordinates . In the formula, respectively represent in The coordinate value in the represents the monitoring point index, represents the number of monitoring points, and ; Define the discrete time series of the microbial fuel cell monitoring data as , where is the index in the time domain, is the length of the time series; Perform discrete Fourier transform on the discrete time series of the microbial fuel cell monitoring data to obtain the frequency domain sequence , where the specific analysis formula is , in the formula, is the frequency domain, is the imaginary unit, satisfying ; is the index in the frequency domain, representing the frequency; Obtain the frequency domain sequence , and calculate its amplitude and phase : , , in the formula, are respectively the real part and the imaginary part of; Define the frequency weight function to reflect the importance of different frequency components to the fire risk, there is , in the formula, is the center frequency, is the bandwidth parameter, is the frequency weight subscript; The function that maps the frequency information to the risk assessment model is defined as: , in the formula, ; G32. Obtain the analysis results of the risk monitoring function corresponding to each monitoring point in the pipe gallery space and organize them into a data set , where is different data types for each monitoring point; For each data point , calculate its probability distribution , is the data point value index, and , is the number of possible values; Assume the data point takes the value , then there is: , where represents a data point takes values as the degree of; Calculate the information entropy using the formula: ; Set the information entropy mapping function , that is , where is the central entropy value, is the scaling parameter of the entropy; Map the information entropy of each data point into the risk assessment model; G33. For the input layer, define the input vector of the input layer, where is the number of input vectors, is the matrix transpose symbol; Based on the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain, obtain the th monitoring point vector : ; For the hidden layer, define that the hidden layer has layers, and the th layer has neurons; among them, represents the number of layers, represents the layer index; Set the weight matrix from the th layer to the th layer as , and its element is , representing the connection weight from the th neuron in the th layer to the th neuron in the th layer; among them, represents the neuron index, and ; Set the bias vector of the th layer as ; The input vector of the neurons in the th layer: , where is the output vector of the th layer (when then ); Using the hyperbolic tangent function , then the output vector of the th layer is : ; For the output layer, set the weight matrix of the output layer to be , and the bias vector to be ; Then the input vector of the output layer is : ; The output vector of the output layer is : , where, represents the risk probability of a fire occurring in the corresponding area inside the utility tunnel, and , where 0 represents no fire risk and 1 represents that a fire is about to occur.
[0019] In the embodiments of the present invention, by selecting and deploying microbial fuel cells according to the environmental characteristics of urban utility tunnels (such as types of organic substances, temperature, humidity, etc.), it is ensured that the equipment operates efficiently and stably in the utility tunnel environment. By evenly distributing microbial fuel cells at key positions, full coverage is achieved to ensure no monitoring blind spots. Through a data transmission network combining wired and wireless, it is ensured that the collected data is transmitted to the analysis module in a timely and accurate manner, providing a reliable basis for fire risk assessment. By using microbial fuel cells as the energy source for monitoring equipment, sustainable energy utilization is realized, and the operation cost is reduced. By integrating the preprocessed and preliminarily analyzed data with the data of traditional monitoring equipment, a feature vector data set containing multi-dimensional information is created, providing rich information for mining fire risk characteristics. By using the dark matter algorithm to calculate the non-linear correlation between feature vectors, correlation patterns highly related to fire risk are mined, improving the accuracy and efficiency of fire risk identification. Classifying and quantifying the mined fire risk characteristics provides a quantitative basis for risk assessment. By integrating and processing data from the three-dimensional physical space analysis domain, the microbial fuel cell data frequency analysis domain, and the information entropy analysis domain, comprehensive utilization of multi-source data is realized, improving the comprehensiveness and accuracy of risk assessment. By adopting a multi-layer neural network structure to construct a fire risk assessment model and learning the complex relationships between data in different analysis domains, accurate prediction of fire risk is realized. In summary, the embodiments of the present invention solve the problem of poor fire risk assessment effect in current urban utility tunnels. In actual situations, more data and context information may be required to make specific decisions and optimization plans.
[0020] In addition, the formulas involved above are all calculated by removing the dimension and taking their numerical values. They are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The proportionality coefficients in the formula and various preset thresholds in the analysis process are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data. The magnitude of the proportionality coefficient is a specific value obtained for quantifying each parameter, which is convenient for subsequent comparison. Regarding the magnitude of the proportionality coefficient, it depends on the amount of sample data and the processing coefficients initially set by those skilled in the art for each group of sample data, as long as the proportional relationship between the parameters and the quantified values is not affected.
[0021] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key points of each embodiment are the differences from other embodiments. In particular, for the device embodiments, since they are basically based on the method embodiments, they are described relatively simply, and reference can be made to the corresponding parts of the method embodiments for the relevant content.
[0022] For the convenience of description, the above device is described by dividing it into various units according to functions. Of course, when implementing the present application, the functions of each unit can be realized in one or more software and / or hardware.
[0023] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0024] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one or more flows or multiple flows and / or blocks Figure 1 one or more blocks or multiple blocks.
[0025] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the procedures Figure 1 procedure or procedures and / or blocks Figure 1 block or blocks.
[0026] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the procedures Figure 1 procedure or procedures and / or blocks Figure 1 block or blocks.
[0027] Second: In the drawings of the disclosed embodiments of the present invention, only the structures related to the disclosed embodiments are involved. For other structures, reference may be made to the general design. Without conflict, the same embodiment and different embodiments of the present invention may be combined with each other. Finally: The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered by the protection scope of the present invention.
Claims
1. The urban pipe gallery fire risk assessment system driven by big data is characterized by: It includes microbial energy monitoring module, fire risk characteristic analysis module and fire risk multi-domain assessment module; Microbial energy monitoring module: select and deploy microbial fuel cells according to the characteristics of the urban pipeline corridor environment; allocate data acquisition components to microbial fuel cells, and establish a data transmission network combining wired and wireless; pre-process and preliminarily analyze the collected microbial fuel cell data; the microbial energy monitoring module includes a microbial fuel cell selection and deployment unit, a data acquisition and data transmission establishment unit, and a data pre-processing and preliminary analysis unit; Fire risk feature analysis module: Integrate pre-processed and preliminary analysis data with data from traditional monitoring equipment to create a feature vector data set containing multi-dimensional information; use the dark matter algorithm to calculate the nonlinear correlation between feature vectors, mine the correlation patterns that are highly correlated with fire risk, and classify and quantify fire risk features; the fire risk feature analysis module includes a data integration and feature vector construction unit and a fire risk feature mining unit; Fire risk multi-domain assessment module: Integrates and processes data from different analysis domains, that is, defines a three-dimensional physical space analysis domain, a frequency analysis domain of microbial fuel cell data, and an information entropy analysis domain, and maps them to the risk assessment model; uses a multi-layer neural network structure to build a fire risk assessment model; among them, the fire risk multi-domain assessment module includes an analysis domain definition and data mapping unit, and a risk assessment model construction unit.
2. The urban pipe gallery fire risk assessment system driven by big data according to claim 1 is characterized in that: The process of selecting and deploying microbial fuel cells according to the characteristics of the urban pipeline corridor environment includes: According to the environmental characteristics of the urban pipeline corridor, the corresponding type of microbial fuel cell is selected; the environmental characteristics of the pipeline corridor include the types of organic matter, temperature range and humidity conditions in the pipeline corridor; The deployment is carried out according to the layout rules in the tunnel, that is, the microbial fuel cells are evenly distributed at the key positions of the tunnel, and the spacing between each microbial fuel cell is determined according to the length and width of the tunnel, the complexity of the environment, the change of the monitoring radius, and the calculation of the temperature and humidity fluctuations in the tunnel; among them, the key positions of the tunnel include cable-dense areas, near the storage points of flammable materials, and around the ventilation holes.
3. The urban pipe gallery fire risk assessment system driven by big data according to claim 1 is characterized in that: The data acquisition and data transmission establishment unit is used to allocate data acquisition components to the microbial fuel cell and establish a data transmission network combining wired and wireless transmission, and the process includes: Allocating a data acquisition component to each microbial fuel cell for real-time acquisition of microbial fuel cell data; The collected data is transmitted to the data preprocessing and preliminary analysis unit by using a combination of wired and wireless transmission methods.
4. The urban pipe gallery fire risk assessment system based on big data drive according to claim 1 is characterized in that: The data preprocessing and preliminary analysis unit is used to preprocess and preliminarily analyze the collected microbial fuel cell data, including the following process: Preprocessing the collected microbial fuel cell data, i.e. data cleaning and normalization; By performing local weighted regression analysis and trend fluctuation analysis on the preprocessed data, the stability of the environment in the pipeline corridor can be preliminarily judged. If the fluctuation of the microbial fuel cell data is found to exceed the preset threshold, it indicates that the environmental factors have changed abnormally. The preliminary analysis results obtained will be marked and stored.
5. The urban pipe gallery fire risk assessment system based on big data drive according to claim 1 is characterized in that: The data integration and feature vector construction unit is used to integrate the pre-processed and preliminarily analyzed data with the data of traditional monitoring equipment. The process of creating a feature vector data set containing multi-dimensional information includes: Obtain data preprocessing and preliminary analysis unit preprocesses and conducts preliminary analysis of data; Obtain data monitored by traditional monitoring equipment in the pipeline corridor; Integrate the pre-processed and initially analyzed data with the data from other traditional monitoring equipment in the tunnel to form a complete data set; Feature vectors are constructed from the integrated data set, and each feature vector contains information of multiple dimensions; among them, other traditional monitoring equipment includes temperature sensors and smoke sensors, and the multiple dimensions of information include the voltage and current of the microbial fuel cell, the readings of the temperature sensor, and the readings of the smoke sensor.
6. The urban pipe gallery fire risk assessment system based on big data drive according to claim 1 is characterized in that: The fire risk feature mining unit is used to calculate the nonlinear correlation between feature vectors using the dark matter algorithm, and to mine the correlation patterns that are highly correlated with fire risk. The process of classifying and quantifying fire risk features includes: Establish a correlation function on the entire feature vector data set to conduct a global search to find the existing hidden correlation patterns; According to the pre-set fire risk related indicators, the correlation patterns that are highly correlated with fire risk are screened out; For the selected correlation patterns, further analyze their performance in time and space scales to obtain fire risk characteristics; The mined fire risk features are classified and quantified, and weights and thresholds are assigned according to their impact on fire risk.
7. The urban pipe gallery fire risk assessment system driven by big data according to claim 1 is characterized in that: The process of the analysis domain definition and data mapping unit for integrating and processing data from different analysis domains and mapping them to the risk assessment model includes: Clarify the various analysis domains in the fire risk assessment model, including the three-dimensional physical space analysis domain, the frequency analysis domain of microbial fuel cell data, and the information entropy analysis domain: In the 3D physical space analysis domain, the location of the pipeline corridor is represented using 3D physical space coordinate data; In the frequency analysis domain, spectrum analysis is performed on the microbial fuel cell monitoring data, the time domain data is converted into frequency domain data, and the amplitude and phase information of different frequency components are calculated, that is, the time series data of the microbial fuel cell monitoring data is fast Fourier transformed to obtain the distribution of the microbial fuel cell monitoring data at different frequencies; In the information entropy analysis domain, the information entropy of microbial fuel cell data is calculated and analyzed; Map the data from each analysis domain into the risk assessment model.
8. The urban pipe gallery fire risk assessment system based on big data drive according to claim 1 is characterized by: The process of the risk assessment model building unit for building a fire risk assessment model using a multi-layer neural network structure includes: The multi-layer neural network structure consists of three main parts: input layer, hidden layer and output layer; Among them, the input layer receives data from different analysis domains, including the three-dimensional spatial coordinates of the pipeline gallery, the frequency and information entropy information of the microbial fuel cell data; The hidden layer performs nonlinear transformation on the input data through the connection weights and activation functions between neurons; The output layer outputs the fire risk assessment results in the form of probability to express the risk level of fire in each area of the tunnel, and the output value is between 0 and 1, where 0 means no fire risk and 1 means a fire is about to occur.