NLP在智能消防接处警系统中的应用研究

Research on the application of NLP in intelligent alarm receiving and handling system

随着城市工业化、现代化进程的持续加快,城市规模扩张、工程建设量急剧增加导致火灾事故频发,对消防救援队伍的响应速度、救援资源调度分配、现场救援作业、科学施救、重点单位预案等综合能力提出更高要求,使信息化手段全面融合消防业务管理和实战应用。为适应城市消防指挥中心的接处警工作要求,针对当前大多数接处警系统信息录入、力量调配效率偏低等问题,提出一种基于ALBERT-BiLSTM-CRF的预训练模型,在接处警系统中对警情要素进行提取,在自行构建的警情语料库中进行实验,基于ALBERT-BiLSTM-CRF的消防警情要素实体识别模型,即将自然语言处理技术运用于消防火灾警情接处警系统中,所得到的警情文本实体识别的F1值为81.660%,该领域需要对语音信息转文本信息后进行快速提取,验证了该模型在保证时间开销和识别准确率的条件下可以提高接警录入效率,并辅助消防人员快速作出救援决策。

With the continuous acceleration of urban industrialization and modernization,the expansion of urban scale and the sharp increase of project construction have led to frequent fire accidents. Higher requirements have been put forward for the response speed of fire rescue teams,the dispatch and allocation of rescue resources,on-site rescue operations,scientific rescue,plans of key units and other comprehensive capabilities,so that the information means are fully integrated with fire business management and practical application. In order to meet the requirements of the urban fire command center for alarm reception and handling,a pre training model based on ALBERT-BiLSTM-CRF is proposed to solve the problems such as the low efficiency of information input and force allocation in most of the current alarm reception and handling systems. The alarm elements are extracted in the alarm reception and handling system,and experiments are conducted in the self-developed alarm corpus. The entity recognition model of fire alarm elements based on ALBERT-BiLSTM-CRF is proposed,Namely,the natural language processing technology is applied to the fire alarm reception and handling system,and the F1 value of the alarm text entity recognition is 81.660%. In this field,it is necessary to quickly extract the voice information after transforming it into text information,which verifies that the model can improve the efficiency of alarm reception and input under the conditions of ensuring time cost and recognition accuracy,and help firefighters make rescue decisions quickly.