基于Spark的煤矿安全预警系统

Coal-mine safety warning system based on Spark

为了减少煤矿安全事故,提高井下工作的安全性,文中设计一种煤矿安全预警系统。该系统由数据采集模块、网络模块、大数据处理模块、客户端模块和井下报警模块5部分构成。其中大数据处理模块具有接收实时环境数据和通过预警模型判断是否预警的功能。此模块采用遗传算法对支持向量机进行参数寻优,建立安全预警模型后使用大量样本对该模型进行训练。文中在大数据处理模块上搭建Spark并行计算框架,在此框架上加载已训练的安全预警模型,并采用Spark Streaming接口接收实时数据,对井下环境数据进行并行计算;然后通过安全预警模型进行安全预测,将预测结果实时发送至手机APP、计算机客户端和井下报警模块。实验结果表明,文中方法能够提高煤矿安全预警系统预测的准确性并降低预警的时延。

A coal-mine safety early warning system is designed to reduce coal-mine accidents and improve the safety of underground work. The system is composed of five parts:data acquisition module,network module,big data processing module,client module and underground alarm module. The big data processing module has the functions of receiving real-time environmental data and judging whether makes early warning. In this module,the genetic algorithm is used to conduct the parameter optimization for the support vector machine,and establish the safety early warning model and then train it with large number of samples. Spark parallel computing framework is built on the big data processing module,and the trained safety warning model is loaded on this framework. The Spark Streaming interface is used to receive real-time data,and conduct the parallel computing of the underground environment data. The safety warning model is used to make safety predictions,and the prediction results are sent to mobile phone APP,computer client and underground alarm module in real time. The experimental results show this method can improve the prediction accuracy and reduce the early-warning time delay of coal-mine safety earlywarning system.