CN-122024390-A - Park fire-fighting early warning method and device, early warning equipment and storage medium

CN122024390ACN 122024390 ACN122024390 ACN 122024390ACN-122024390-A

Abstract

The invention relates to a park fire-fighting early-warning method, a device, early-warning equipment and a storage medium, belonging to the technical field of fire-fighting safety management, wherein the park fire-fighting early-warning method comprises the steps of acquiring temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of a plurality of monitoring areas; determining fire fighting abnormal states, fire fighting abnormal type probability distribution and state confidence coefficient of the plurality of monitoring areas based on the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the plurality of monitoring areas, wherein the fire fighting abnormal types comprise electric overheat, chemical leakage, equipment friction and pipe network leakage, determining fire hazard risk levels, risk tracing results and fire fighting facility matching states of the park based on the fire fighting abnormal states, the fire fighting abnormal type probability distribution and the state confidence coefficient of the plurality of monitoring areas, and generating a fire fighting early warning treatment strategy based on the fire hazard levels of the park. The method effectively improves the accuracy of the fire early warning of the park.

Inventors

MA YANJUAN
DONG ZHIYONG
Zhang Diehu
QIU LIN

Assignees

武汉理工光科股份有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. A park fire-fighting early warning method is characterized by comprising the following steps: Acquiring temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of a plurality of monitoring areas, wherein the temperature data is acquired based on a distributed temperature measuring fiber core, and the gas concentration data, the water pressure data and the water pipe vibration acceleration data are acquired based on a fiber bragg grating array; determining abnormal fire states, abnormal fire type probability distribution and state confidence of the plurality of monitoring areas based on the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the plurality of monitoring areas, wherein the abnormal fire types comprise electric overheat, chemical leakage, equipment friction and pipe network leakage; Based on the abnormal fire-fighting states, abnormal fire-fighting type probability distribution and state confidence of the monitoring areas, determining fire risk levels, risk tracing results and fire-fighting facility matching states of the parks, and generating a fire-fighting early-warning processing strategy based on the fire risk levels of the parks.
2. The method of claim 1, wherein determining the fire anomaly status, the fire anomaly type probability distribution, and the status confidence of the plurality of monitored areas based on the preprocessed temperature data, the preprocessed gas concentration data, the preprocessed water pressure data, and the preprocessed water pipe vibration acceleration data of the plurality of monitored areas comprises: The temperature data, the gas concentration data, the water pressure data and the water pipe vibration acceleration data of the plurality of monitoring areas after pretreatment are used as the input of a fire fighting abnormal reasoning model, the fire fighting abnormal state, the fire fighting abnormal type probability distribution and the state confidence coefficient of the plurality of monitoring areas output by the fire fighting abnormal reasoning model are obtained, the fire fighting abnormal reasoning model is obtained by training the improved MobileNetV model by taking the park historical fire fighting abnormal data as a sample, the fully-connected layer of the improved MobileNetV model is replaced by a depth separable convolution layer, and a self-adaptive threshold adjusting module is arranged behind the feature fusion layer.
3. The method of claim 1, wherein determining the fire risk level, the risk traceability result, and the fire facility matching status for the campus based on the fire anomaly status, the fire anomaly type probability distribution, and the status confidence of the plurality of monitoring areas comprises: Based on the abnormal fire fighting state, abnormal fire fighting type probability distribution and state confidence of the monitoring areas, and the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the monitoring areas, determining fire risk level, risk tracing result and fire fighting facility matching state of the park.
4. The method of fire early warning in a campus of claim 3, wherein determining the fire risk level, the risk tracing result, and the fire facility matching state in the campus based on the fire abnormal state, the fire abnormal type probability distribution, and the state confidence of the plurality of monitoring areas, and the temperature data, the gas concentration data, the water pressure data, and the water pipe vibration acceleration data of the plurality of monitoring areas after the preprocessing, comprises: The fire control abnormal state, fire control abnormal type probability distribution and state confidence of the monitoring areas, and temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the preprocessed monitoring areas are used as inputs of a fire disaster reasoning model, so that fire disaster risk level, risk tracing result and fire control facility matching state of a park output by the fire disaster reasoning model are obtained, the fire control abnormal reasoning model is obtained by training an improved transducer model by taking park historical fire disaster data as samples, and the improved transducer model comprises 4 layers of encoders and 4 multiple attention layers.
5. The method of campus fire early warning according to claim 1, further comprising: and generating a park fire protection early warning state display diagram based on the fire risk level, the risk tracing result and the fire protection facility matching state of the park and the three-dimensional GIS map of the park.
6. The method of campus fire early warning according to claim 1, further comprising: and carrying out fault early warning on the waterproof cancellation pump based on the water pressure data and the water pipe vibration acceleration data of the monitoring areas.
7. The method of claim 6, wherein the performing fault warning on the fire fighting pump based on the water pressure data and the water pipe vibration acceleration data of the plurality of monitoring areas comprises: And taking the water pressure data and the water pipe vibration acceleration data of the monitoring areas as the input of a fire pump fault early warning model to obtain fire pump fault early warning information output by the fire pump fault early warning model, wherein the fire pump fault early warning model is obtained by training the LSTM model by taking historical water pressure data and the water pipe vibration acceleration data as samples and taking the fire pump historical fault condition as a label.
8. The utility model provides a garden fire control early warning device which characterized in that includes: The acquisition module is used for acquiring temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of a plurality of monitoring areas, wherein the temperature data is acquired based on a distributed temperature measurement fiber core, and the gas concentration data, the water pressure data and the water pipe vibration acceleration data are acquired based on a fiber bragg grating array; The first determining module is used for determining abnormal fire-fighting states, abnormal fire-fighting type probability distribution and state confidence of the plurality of monitoring areas based on the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the plurality of monitoring areas, wherein the abnormal fire-fighting types comprise electric overheat, chemical leakage, equipment friction and pipe network leakage; The second determining module is used for determining fire risk levels, risk tracing results and fire-fighting facility matching states of the park based on the fire-fighting abnormal states, fire-fighting abnormal type probability distribution and state confidence of the plurality of monitoring areas, and generating a fire-fighting early-warning processing strategy based on the fire risk levels of the park.
9. An early warning device is characterized by comprising a memory and a processor, wherein, The memory is used for storing programs; the processor, coupled to the memory, for executing the program stored in the memory to implement the steps in the campus fire early warning method of any one of the preceding claims 1 to 7.
10. A computer readable storage medium storing a computer readable program or instructions which when executed by a processor is capable of carrying out the steps of the method of any one of claims 1 to 7.

Description

Park fire-fighting early warning method and device, early warning equipment and storage medium Technical Field The invention relates to the technical field of fire safety management, in particular to a park fire early warning method, a park fire early warning device, early warning equipment and a storage medium. Background Currently, fire monitoring systems in parks (including industrial parks, scientific parks, large factories, etc.) are often in a mode of combining traditional point-type smoke detectors, temperature detectors and independent fire pipe network pressure monitoring systems. With the development of optical fiber sensing technology, a system based on distributed optical fiber temperature measurement (Distributed Temperature Sensing, DTS) and fiber grating (Fiber Bragg Grating, FBG) point temperature measurement has started to be applied to specific scenes such as cable tunnels, storage tanks and the like. However, when facing modern gardens with complex structures and various risk sources, the prior art scheme has obvious systematic limitations that the sensing dimension is single, early warning of fire can not be realized, the systems are isolated and run, linkage is delayed, centralized analysis is relied on, the instantaneity and the reliability are insufficient, the intelligent level is insufficient, and the false alarm rate is high. Therefore, how to improve the accuracy of the fire-fighting early-warning scheme in the park and further improve the safety of fire-fighting management in the park becomes a technical problem to be solved urgently. Disclosure of Invention In view of the foregoing, it is necessary to provide a method, a device, an early warning device and a storage medium for early warning of park fire, which are used for solving the problem of insufficient accuracy of the existing park fire early warning scheme. In order to solve the above problems, in a first aspect, the present invention provides a method for early warning of fire in a campus, including: Acquiring temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of a plurality of monitoring areas, wherein the temperature data is acquired based on a distributed temperature measuring fiber core, and the gas concentration data, the water pressure data and the water pipe vibration acceleration data are acquired based on a fiber bragg grating array; determining abnormal fire states, abnormal fire type probability distribution and state confidence of the plurality of monitoring areas based on the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the plurality of monitoring areas, wherein the abnormal fire types comprise electric overheat, chemical leakage, equipment friction and pipe network leakage; Based on the abnormal fire-fighting states, abnormal fire-fighting type probability distribution and state confidence of the monitoring areas, determining fire risk levels, risk tracing results and fire-fighting facility matching states of the parks, and generating a fire-fighting early-warning processing strategy based on the fire risk levels of the parks. In one possible implementation manner, the determining the abnormal fire state, abnormal fire type probability distribution and state confidence of the plurality of monitoring areas based on the preprocessed temperature data, gas concentration data, water pressure data and water pipe vibration acceleration data of the plurality of monitoring areas includes: The temperature data, the gas concentration data, the water pressure data and the water pipe vibration acceleration data of the plurality of monitoring areas after pretreatment are used as the input of a fire fighting abnormal reasoning model, the fire fighting abnormal state, the fire fighting abnormal type probability distribution and the state confidence coefficient of the plurality of monitoring areas output by the fire fighting abnormal reasoning model are obtained, the fire fighting abnormal reasoning model is obtained by training the improved MobileNetV model by taking the park historical fire fighting abnormal data as a sample, the fully-connected layer of the improved MobileNetV model is replaced by a depth separable convolution layer, and a self-adaptive threshold adjusting module is arranged behind the feature fusion layer. In one possible implementation manner, the determining the fire risk level, the risk tracing result and the fire protection facility matching state of the park based on the fire protection abnormal states, the fire protection abnormal type probability distribution and the state confidence of the plurality of monitoring areas includes: Based on the abnormal fire fighting state, abnormal fire fighting type probability distribution and state confidence of the monitoring areas, and the preprocessed temperature data, gas concentration data, water pressure data and water pipe vib