CN-121982820-A - Intelligent firefighting intelligent monitoring and early warning system with multiple scene adaptations

CN121982820ACN 121982820 ACN121982820 ACN 121982820ACN-121982820-A

Abstract

The application discloses a multi-scene-adaptive intelligent firefighting monitoring and early warning system which comprises a scene feature recognition module, a multi-mode sensing edge computing network, a heterogeneous sensor node group deployed in a monitoring area, a dynamic threshold self-adaptive engine, a space-time fusion analysis center, a three-dimensional visualization early warning platform and a hierarchical linkage response controller, wherein the heterogeneous sensor node group comprises temperature, smoke, gas, images, sounds and thermal imaging sensors, each node is provided with an edge computing unit to realize local preprocessing and feature extraction of original data, and the cooperative sensing among nodes is realized through a self-organizing network. According to the application, the scene characteristic recognition module is used for fusing the building BIM model and the real-time environment sensor data, the system can automatically recognize the type of the monitored scene, the spatial structure characteristics and the risk level, and a differential monitoring strategy is provided for different building spaces, so that the problem of 'one-cut' of the traditional system is solved.

Inventors

NIU LIJUAN
MA ZHEXUAN

Assignees

中环低碳节能技术(北京)有限公司
中数信科信息科技(北京)有限公司

Dates

Publication Date: 20260505
Application Date: 20260130

Claims (10)

1. An intelligent fire control intelligent monitoring early warning system of many scene adaptations, its characterized in that includes: the scene feature recognition module automatically recognizes the type of the monitored scene, the spatial structure feature and the risk level through the data fusion of the building BIM model and the environmental sensor; The multi-mode sensing edge computing network is deployed in a heterogeneous sensor node group of a monitoring area and comprises temperature, smoke, gas, image, sound and thermal imaging sensors, each node is provided with an edge computing unit, local preprocessing and feature extraction of original data are realized, and cooperative sensing among the nodes is realized through a self-organizing network; The dynamic threshold self-adaptive engine dynamically adjusts the early warning threshold and weight of various sensors based on scene characteristics, historical data and a deep learning model, and builds a scene-specific multi-dimensional early warning index system; The space-time fusion analysis center adopts an improved space-time diagram convolution network and LSTM hybrid architecture to perform space-time correlation analysis on the multi-source heterogeneous monitoring data and identify a potential fire risk mode; The three-dimensional visualization early warning platform is integrated with BIM+GIS technology to construct a building holographic digital model, so that real-time risk situation visualization, smoke spreading simulation and personnel evacuation path optimization are realized; and the grading linkage response controller automatically generates a grading response strategy according to the risk assessment result and coordinates equipment linkage work such as fire-fighting facilities, alarm devices, smoke discharging systems, emergency lighting and the like.
2. The intelligent fire-fighting monitoring and early-warning system with multiple scene adaptation according to claim 1 is characterized in that the scene characteristic recognition module comprises a data preprocessing unit, a data processing unit and a data processing unit, wherein the data preprocessing unit is used for receiving building information model data and real-time sensor data and carrying out unified formatting processing; The feature extraction unit is used for performing space-time alignment and fusion processing on the real-time sensor data and the BIM static data based on a weighted D-S evidence theory algorithm to generate scene feature vectors; the scene classification and identification unit adopts an improved random forest classifier, inputs the feature vector to output the scene type identifier and the risk level, and a built-in self-learning mechanism continuously optimizes the classification decision tree.
3. The intelligent fire protection intelligent monitoring and early warning system with multiple scene adaptation according to claim 1, wherein the multi-mode sensing edge computing network comprises: the sensing terminal node consists of a plurality of sensors, and is arranged according to the building fireproof partition gridding, and the acquisition frequency is 1-5Hz; An edge calculation unit that performs moving average filtering, flame feature extraction, MFCC feature extraction, and adaptive temperature threshold segmentation; The self-organizing network adopts the combination of the ZigBee3.0 protocol and the LoRaWAN protocol, supports the self-discovery and self-repair functions of the nodes, and has the network reconstruction time less than 15 seconds.
4. The intelligent monitoring and early warning system for multi-scene adaptation according to claim 1, wherein the dynamic threshold adaptation engine comprises a data preprocessing unit, a data processing unit and a data processing unit, wherein the data preprocessing unit is used for receiving scene type identifiers and risk grades output by a scene characteristic recognition module and acquiring historical data of the same period in a historical early warning database; the deep learning prediction unit adopts a three-layer LSTM network structure, wherein the number of input layer nodes is 31, the number of hidden layer nodes is 64, the number of output layer nodes is 12, and the initial threshold adjustment coefficients respectively correspond to 12 types of sensors; And the parameter optimization unit adopts an online learning mechanism, re-evaluates the threshold performance after every 24 hours or every 100 early warning events and triggers the parameter retraining process.
5. The intelligent monitoring and early warning system for multi-scene adaptation according to claim 1 is characterized in that the space-time fusion analysis center comprises a data standardization module, a data analysis module and a data analysis module, wherein the data standardization module is used for carrying out unified processing on multi-source heterogeneous monitoring data and extracting feature vectors; the space relation modeling module is used for constructing a building space topological graph, and an improved space-time graph convolution network is adopted, wherein the building space topological graph comprises 3 ST-blocks, and each ST-block comprises 1 graph convolution layer and 1 time convolution layer; the time sequence analysis module adopts a double-layer LSTM network, the number of hidden units is 128, the activation functions of an input gate, a forgetting gate and an output gate are sigmoid, and the unit state update function is tanh; The fusion decision module adopts a gating fusion mechanism to calculate the fusion weight of the ST-GCN output characteristic and the LSTM output characteristic, and generates fusion characteristics; the risk mode identification module comprises two parallel branches, wherein the fully-connected network outputs the fire probability of each risk area, and the regression network predicts the development speed and the spreading direction of the fire.
6. The intelligent fire protection intelligent monitoring and early warning system with multiple scene adaptation according to claim 1, wherein the three-dimensional visual early warning platform comprises: the BIM-GIS data fusion engine is used for converting the building information model into three-dimensional tile data through a converter; A three-dimensional scene rendering engine, which is custom developed and developed based on Unity 3D; The real-time risk visualization module adopts a thermodynamic diagram superposition technology and maps the risk index output by the space-time fusion analysis center into RGB color values; The flue gas dynamics simulation module adopts an improved area model and field model mixing algorithm to solve a Navier-Stokes equation set and an energy conservation equation, and calculates a flue gas temperature field, a concentration field and a visibility field; intelligent evacuation path planning module, based on improved A-algorithm, introduces dynamic weight factors The path cost function is defined as 。
7. The intelligent fire protection intelligent monitoring and early warning system of multiple scene adaptation according to claim 1, wherein the hierarchical coordinated response controller comprises: The risk level judging unit is used for receiving the comprehensive fire risk index CFRI output by the space-time fusion analysis center and dividing four-level risk levels according to a preset threshold value; The strategy generation unit is used for matching a corresponding response strategy set from the strategy library according to the risk level, the building type and the personnel density; the equipment control execution unit is connected with the fire-fighting equipment control system through a double-protocol communication interface; the state feedback monitoring unit is used for collecting state feedback signals of each execution device in real time and constructing a device state matrix ; And the exception handling unit starts the standby control channel when detecting that the device execution is overtime or feedback is abnormal.
8. The intelligent multi-scene adaptive fire protection monitoring and early warning system according to claim 7, wherein the hierarchical response strategy is specifically: The first-level response starts the low-frequency mode of the audible and visual alarm and monitors the tracking of the camera; The second-level response is added to start the floor broadcasting system, open the smoke discharging valve of the corresponding fireproof partition and close the air conditioning system; the three-stage response further starts the emergency lighting system, releases the fireproof roller shutter and starts a dynamic guiding mode of the evacuation indication system; The four-stage response fully starts the spraying system, cuts off the non-fire-fighting power supply and starts the positive pressure air supply system.
9. The intelligent fire protection intelligent monitoring and early warning system with multiple scene adaptation according to claim 1, further comprising a system health self-diagnosis module for periodically performing functional checks on the sensor nodes, the communication links and the analysis engine, generating a diagnosis report and automatically repairing common faults.
10. The intelligent monitoring and early warning system for multi-scene adaptation, which is disclosed by claim 1, is characterized by a two-stage early warning mechanism of an edge layer and a central layer, wherein the two-stage early warning mechanism of the system is characterized in that an edge node immediately triggers local early warning when detecting local abnormality, and a central server triggers global early warning after comprehensively analyzing multi-source data, and the two-stage early warning works cooperatively.

Description

Intelligent firefighting intelligent monitoring and early warning system with multiple scene adaptations Technical Field The application belongs to the technical field of fire safety, and particularly relates to a multi-scene adaptive intelligent fire-fighting monitoring and early-warning system. Background With the acceleration of the urban process and the continuous improvement of the complexity of the building, fire safety has become a core issue of modern building management. The traditional fire-fighting monitoring and early warning system is mainly based on a single type sensor and a fixed threshold judgment mechanism, and has the obvious problem of insufficient scene adaptability. In the prior art, a fire early warning system generally adopts a single parameter threshold triggering mechanism such as temperature, smoke and the like, and cannot be adaptively adjusted according to different building types, spatial structures, material characteristics and using functions, so that the false alarm rate and the false alarm rate are high in complex and changeable environments. Although a part of AI technology is introduced into the intelligent fire protection system in the market, multiple challenges still exist, namely firstly, the multisource heterogeneous sensor data lacks effective space-time correlation analysis, complex characteristic modes of early fires are difficult to identify, secondly, a fixed threshold early warning mechanism cannot adapt to different requirements of different environmental conditions (such as special scenes of a kitchen, a warehouse, a data center and the like), furthermore, the existing system lacks deep understanding of structural features of a building space, a risk source cannot be accurately positioned and a fire spreading path cannot be predicted, and furthermore, an emergency response mechanism usually adopts an 'all or nothing' control strategy, and cannot realize hierarchical linkage of equipment according to risk grades, so that unnecessary property loss and personnel panic are caused. In large-scale complex buildings, the traditional 'one-cut' fire-fighting monitoring strategy is difficult to meet the requirement of fine management due to various spatial structures and different functional areas. Meanwhile, the existing system often adopts a centralized processing mode on a data processing architecture, and the early warning delay is caused by lack of edge computing capability, so that the time window requirement of early quick response of fire can not be met. In the aspect of visualization, the conventional system mostly adopts a two-dimensional plan view or a simple chart to display risk information, so that the development situation of fire in a three-dimensional space cannot be intuitively presented, and the emergency decision efficiency is affected. Disclosure of Invention The application provides a multi-scene adaptive intelligent fire-fighting intelligent monitoring and early warning system, and aims to solve the problems that complex characteristic modes of early fires are difficult to identify and different environmental conditions cannot be met in the prior art. An intelligent fire-fighting intelligent monitoring and early-warning system with multiple scene adaptation comprises: the scene feature recognition module automatically recognizes the type of the monitored scene, the spatial structure feature and the risk level through the data fusion of the building BIM model and the environmental sensor; The multi-mode sensing edge computing network is deployed in a heterogeneous sensor node group of a monitoring area and comprises temperature, smoke, gas, image, sound and thermal imaging sensors, each node is provided with an edge computing unit, local preprocessing and feature extraction of original data are realized, and cooperative sensing among the nodes is realized through a self-organizing network; The dynamic threshold self-adaptive engine dynamically adjusts the early warning threshold and weight of various sensors based on scene characteristics, historical data and a deep learning model, and builds a scene-specific multi-dimensional early warning index system; The space-time fusion analysis center adopts an improved space-time diagram convolution network and LSTM hybrid architecture to perform space-time correlation analysis on the multi-source heterogeneous monitoring data and identify a potential fire risk mode; The three-dimensional visualization early warning platform is integrated with BIM+GIS technology to construct a building holographic digital model, so that real-time risk situation visualization, smoke spreading simulation and personnel evacuation path optimization are realized; and the grading linkage response controller automatically generates a grading response strategy according to the risk assessment result and coordinates equipment linkage work such as fire-fighting facilities, alarm devices, smoke discharging systems, emergency lighting