铁路路基状态检测中探地雷达数据并行处理

Research on Parallel Processing Method for Ground Penetrating Radar Data in State Detection of Railway Subgrade

为快速处理探地雷达检测铁路路基状态所产生的大量检测数据,缩短检测报告的生成周期,采用并行计算技术设计并构建适用于探地雷达数据解析的并行处理平台,利用计算机集群处理解析探地雷达数据;基于服务器计算能力的动态探地雷达数据任务负载均衡算法,对用户提交的探地雷达数据解析任务统一调度分发。采用实际的铁路路基状态检测雷达数据对构建的并行处理平台进行实验,分析雷达数据并行处理的准确性、时间消耗、并行化加速比和系统可扩展性等指标。结果表明:在8个节点的集群并行处理平台上进行探地雷达数据的处理效率比用单机版软件提高553%,处理时间比基于Hadoop的探地雷达数据并行处理方法缩短50%以上。

In order to improve the processing speed of large amount of ground penetrating radar （GPR） data produced by railway subgrade state detection, shorten the generation cycle of test report, parallel computing technology was introduced to design and construct the parallel processing platform for GPR data analysis. Computer cluster was used to process and analyze GPR data. Dynamic load balance algorithm based on server computing ability was proposed to schedule and distribute the GPR process task submitted by users. Experiments were carried through on the constructed parallel processing platform with the actual GPR data of railway subgrade condition detection. The accuracy, time consumption, parallel speedup and system scalability of the GPR data parallel processing platform were analyzed. Results show that the GPR data processing efficiency of 8-node cluster parallel processing platform is increased by 553% compared with the single version software. The processing time is reduced by more than 50% compared with the GPR data processing method based on Hadoop.