Search

CN-121982865-A - Safety and extinction integrated risk early warning management and control platform based on big data intelligent analysis

CN121982865ACN 121982865 ACN121982865 ACN 121982865ACN-121982865-A

Abstract

The invention relates to the technical field of early warning management and control, in particular to an safety and security integrated risk early warning management and control platform based on big data intelligent analysis. The system comprises a data acquisition unit, an abnormality analysis unit, a path planning unit, an emergency response unit and an emergency response unit, wherein the data acquisition unit is used for acquiring environmental state information and equipment operation state information of a target area through different sensors, the abnormality analysis unit is used for establishing a temperature prediction model according to the environmental state information and the equipment operation state data and utilizing a cyclic neural network, the path planning unit is used for planning an optimal evacuation path through an algorithm and transmitting the optimal evacuation path to the emergency response unit, and the emergency response unit is used for triggering an emergency response program when an early warning signal is received. The platform can generate accurate prediction results, help management personnel to more scientifically make emergency plans and maintenance plans, and can make effective early warning in a short period, also consider long-term accumulation effects, reduce false alarm and missing report, and improve early warning accuracy and response speed.

Inventors

  • LIU QUANFENG
  • LIU JIE
  • WANG ZHANFU
  • ZOU PINGHAI

Assignees

  • 云南边锋信息技术有限公司

Dates

Publication Date
20260505
Application Date
20260205

Claims (10)

  1. 1. Safety and extinction integrated risk early warning management and control platform based on big data intelligent analysis, its characterized in that includes: The system comprises a data acquisition unit (1), a path planning unit (3) and a monitoring unit, wherein the data acquisition unit (1) is used for acquiring environmental state information and equipment operation state information of a target area through different sensors, transmitting the acquired environmental state information and equipment operation state information to the anomaly analysis unit (2), acquiring personnel flow information through the monitoring unit and transmitting the acquired personnel flow information to the path planning unit (3); The abnormality analysis unit (2) is used for analyzing the collected environmental state information and equipment operation state information by utilizing big data and a machine learning algorithm, establishing a temperature prediction model by utilizing a cyclic neural network according to the environmental state information and the equipment operation state data, introducing smoke concentration influence factors and a multi-head attention mechanism to optimize in the process of establishing the temperature prediction model, generating an abnormality signal once abnormality occurs, and transmitting the abnormality signal to the path planning unit (3) and the emergency response unit (4); The path planning unit (3) is used for extracting key data from the received personnel flow information, planning an optimal evacuation path by using an algorithm based on the real-time key data when an abnormal signal of the abnormal analysis unit (2) is received, and transmitting the optimal evacuation path to the emergency response unit (4); The emergency response unit (4) is used for triggering an emergency response program when receiving the early warning signal, and the emergency response program comprises the steps of spraying water to extinguish a fire, broadcasting a notification personnel evacuation path and sending alarm information to a fire department.
  2. 2. The security and elimination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 1, wherein the abnormality analysis unit (2) specifically comprises the following steps: S1, cleaning the collected environmental state information and equipment running state information, removing abnormal values and missing values, and normalizing the data to the same scale; S2, constructing a multidimensional time sequence input vector, wherein the input vector is formed by splicing time sequence temperature data of target equipment and other external variable data of a time sequence in a time dimension, the spliced multidimensional time sequence input vector is processed by using an LSTM network, dynamic characteristics of each variable along with time change and coupling relations among the variables are captured, a multi-head attention mechanism is introduced to capture the characteristics in different subspaces, and finally, a context vector generated by the attention mechanism is converted into a final temperature predicted value; S3, calculating a second-order residual error of the temperature value predicted by the model, estimating the mean value and the variance of the second-order residual error to determine parameters of normal distribution, and calculating a cumulative probability distribution function according to the normal distribution parameters of the second-order residual error, wherein the cumulative probability distribution function is used for determining the position of a certain temperature value in distribution; and S4, setting early warning quantiles according to the cumulative probability distribution, calculating a corresponding temperature value as an early warning threshold according to the set quantiles, and defining fire risks through the cumulative probability distribution and the early warning quantiles.
  3. 3. The security and elimination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 2, wherein the specific steps of S2 are as follows: the input feature vector of the building model is formed by splicing main input data and external variable data, wherein the main input data is time sequence temperature data , wherein, Is the first The temperature at each time point, the external variable including ambient temperature And a current load Wherein, the method comprises the steps of, Is the first Ambient temperature at each time point; Is the first Current loading at each time point; ; Wherein, the For the current time step Is hidden in the first layer; for the previous time step Is hidden in the first layer; For the current time step Is set at a temperature of (2); For the current time step Ambient temperature of (2); For the current time step Is used for the current load of the (c), ; Attention introducing mechanism: ; ; ; Wherein, the Is a time step Is a weight of attention of (2); Is a time step Is a concentration score of (2); Is the length of the time series; Is a time step Is a concentration score of (2); ; Is a weight vector in the attention mechanism; transpose of weight vectors in the attention mechanism; Is a weight matrix; Is a bias term; is a context vector; ; Wherein, the Is the first Predicted temperatures at the respective time points; A weight matrix for the full connection layer; is a bias term for the fully connected layer.
  4. 4. The safety and decontamination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 3 is characterized in that in the abnormality analysis unit (2), smoke concentration influence factors and a multi-head attention mechanism are introduced to optimize in the process of establishing a temperature prediction model, and the optimization is specifically as follows: ; Wherein, the Is the output voltage of the smoke sensor; Is a sensitivity constant; Is smoke concentration; is a bias voltage; the external variable increases the output voltage of the smoke sensor ; ; Wherein, the For the optimized current time step Is hidden in the first layer; For the previous time step after optimization Is hidden in the first layer; For the current time step Is a smoke sensor of (a) is set to the output voltage of (a); attention introducing mechanism: ; ; ; ; Wherein, the Is the first The attention heads being in time steps Is a weight of attention of (2); Is the first The attention heads being in time steps Is a concentration score of (2); Is a time step Is a concentration score of (2); Is the first Transpose of the weight vectors of the individual attention heads; Is the first A weight matrix of the individual attention heads; Is the first Bias terms for the individual attention heads; Is the first Context vectors for the individual attention heads; Is the sum of the context vectors of all the attention heads; is the number of attention heads; ; ; Wherein, the To the optimized first Predicted temperature at each time point.
  5. 5. The security and elimination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 4, wherein the step S3 is specifically as follows: Calculating a second-order residual of the temperature value predicted by the model: ; ; Wherein, the To be in a time step When the actual temperature is different from the model predicted temperature; To be in a time step Is set, the actual temperature of (a); To be in a time step Model predicted temperature of (2); Is that I.e. the second order residual; ; ; Wherein, the Is the mean value of the second order residual error; Is the length of the time step; Is the variance of the second order residual; ; Therefore, the normal distribution parameters of the second order residuals are: ; Representing second order residual error Obeying normal distribution, the average value is Variance is ; ; Wherein, the Is a cumulative probability distribution function; for a specific temperature value for which the cumulative probability is to be calculated; is a cumulative distribution function of a standard normal distribution.
  6. 6. The security and elimination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 5, wherein in the step S4, early warning quantiles are set according to the cumulative probability distribution, and corresponding temperature values are calculated as early warning threshold values according to the set quantiles, specifically as follows: Selecting quantiles Finding the corresponding quantile by looking up a standard normal distribution table Z value of (2) Then the Z value corresponding to the fractional number is used And parameters of normal distribution And Calculating an early warning threshold value : ; Wherein, the Is an early warning threshold.
  7. 7. The security and elimination integrated risk early warning management and control platform based on big data intelligent analysis according to claim 6 is characterized in that in S4, fire risks are defined through cumulative probability distribution and early warning positioning, and the security and elimination integrated risk early warning management and control platform concretely comprises the following steps: S41, selecting early warning quantiles, and calculating corresponding early warning thresholds according to the selected quantiles; wherein the early warning score comprises Dividing the number of digits, Number of digits Dividing the number of bits; For the following Quantiles: ; Wherein, the Is that Quantile early warning threshold; Is that Dividing the number of bits; For the following Quantiles: ; Wherein, the Is that Quantile early warning threshold; Is that Dividing the number of bits; For the following Quantiles: ; Wherein, the Is that Quantile early warning threshold; Is that Dividing the number of bits; s42, defining different fire risk levels according to the calculated early warning threshold value, and triggering early warning of corresponding levels: if the actual temperature is If the fire risk is low, the early warning is not triggered; If it is Triggering low-risk early warning when the fire risk is a medium fire risk; If it is Triggering medium risk early warning if the fire is a high-level fire risk; If it is And triggering high-risk early warning for extremely high fire risks.
  8. 8. The safety and security integrated risk early warning management and control platform based on big data intelligent analysis according to claim 7 is characterized in that key data in the path planning unit (3) comprise the position, the density and the moving direction of personnel.
  9. 9. The safety and decontamination integrated risk early warning management and control platform based on big data intelligent analysis of claim 8 is characterized in that an optimal evacuation path is planned by an algorithm based on real-time key data in the path planning unit (3), and the method is specifically as follows: s5, dividing the target into grids to generate a grid map, wherein each grid represents a node, the connection between the nodes represents a feasible path, the non-passable area is marked in the grid map, and the grid map is updated according to real-time key data; s6, calculating the number of people in each grid, generating a personnel density map, and predicting the moving direction and speed according to the historical position data of the people; s7, setting an abnormal signal generation place as a starting point, setting an end point as a safety exit, selecting a path which has the shortest distance and avoids a high-density area and an obstacle from the shortest path distance between each node and the end point, adjusting the path according to real-time data, and finally transmitting the planned evacuation path to an emergency response unit (4).
  10. 10. The safety and security integrated risk early warning management and control platform based on big data intelligent analysis according to claim 9 is characterized in that the emergency response unit (4) comprises an early warning model optimization module, and the early warning model optimization module can continuously optimize early warning speed according to historical data and real-time feedback information, specifically as follows; the early warning model optimization module performs reinforcement learning by utilizing SARSA, and the updating rule is as follows: ; Wherein, the To be in a time step State at time Take action downwards A kind of electronic device The value of the sum of the values, The value represents the slave state Take action The expected cumulative return thereafter; is the learning rate; Is a discount factor; To be in a time step State at time Take action downwards A kind of electronic device A value; To be in a time step A state at that time; To be in a time step A state at that time; To be in a time step Action taken at that time; To be in a time step Action taken at that time; To be in a time step Instant rewards obtained at that time.

Description

Safety and extinction integrated risk early warning management and control platform based on big data intelligent analysis Technical Field The invention relates to the technical field of early warning management and control, in particular to an safety and security integrated risk early warning management and control platform based on big data intelligent analysis. Background With the acceleration of the urbanization process and the increasing population density, public safety issues in hospitals are becoming increasingly important. The security management of hospital buildings and public places is subject to higher demands. In recent years, various safety accidents frequently happen, particularly fire disasters pose serious threat to the life and property safety of people, and more advanced early warning and management and control means are urgently needed. Meanwhile, the evolution of internet of things (IoT) technology has enabled a large number of sensors and devices to be deployed within hospital buildings and areas, collecting environmental data in real-time. Advances in big data and cloud computing technology have enabled the storage, processing, and analysis of vast amounts of data. Artificial intelligence technology, especially machine learning and deep learning, can extract valuable information from a large amount of data, and perform intelligent analysis and decision support; The existing system enables various sensors, video monitoring equipment, data analysis platforms and the like to cooperate with each other through the internet of things (IoT), real-time monitoring, early warning and disposal of potential safety hazards and fire events are achieved, in order to meet the requirements of different industries and scenes, the existing security system is generally in a modularized design, the system is customized and expanded according to actual requirements, and various potential safety hazards and fire events including fire smoke, electrical fires, abnormal personnel flow and the like can be monitored in real time. However, the prior art has difficulty in integrating various safety monitoring and fire-fighting early warning devices, so that the overall integration level of the system is not high, the stability and reliability of the system can be influenced, the accuracy and timeliness of early warning are reduced, and the intelligent level of part of the system is still limited although the artificial intelligent technology is applied to an safety and fire-fighting integrated system. The method is mainly characterized in that the method is in the aspects of accuracy and robustness of an algorithm model and the adaptability of the system to complex scenes, so that an safety and security integrated risk early warning management and control platform based on big data intelligent analysis is designed. Disclosure of Invention The invention aims to provide an safety and decontamination integrated risk early warning management and control platform based on big data intelligent analysis, so as to solve the problems that the overall integration level of a system provided in the background technology is not high, the stability and reliability of the system are affected, the accuracy and timeliness of early warning are reduced, and although an artificial intelligence technology is applied to the safety and decontamination integrated system, the intelligent level of part of the system is still limited, and the problems are mainly expressed in the aspects of the accuracy and the robustness of an algorithm model and the adaptability of the system to complex scenes. In order to achieve the above purpose, the present invention aims to provide an security and elimination integrated risk early warning management and control platform based on big data intelligent analysis, comprising: the data acquisition unit is used for acquiring the environmental state information and the equipment running state information of the target area through different sensors, transmitting the acquired environmental state information and the acquired equipment running state information to the anomaly analysis unit, acquiring the personnel flow information through the monitoring equipment and transmitting the acquired personnel flow information to the path planning unit; The abnormality analysis unit is used for analyzing the acquired environmental state information and equipment operation state information by utilizing big data and a machine learning algorithm, establishing a temperature prediction model by utilizing a cyclic neural network according to the environmental state information and the equipment operation state data, introducing smoke concentration influence factors and a multi-head attention mechanism to optimize in the process of establishing the temperature prediction model, generating an abnormality signal once abnormality occurs, and transmitting the abnormality signal to the path planning unit and the emergency respons