基于强化学习的分布式安全预警系统研究

Research of Distributed Safety Warning System Based on Reinforcement Learning

文中主要针对危险环境,设计了基于强化学习的分布式安全预警系统。其中,系统硬件选用FPGA作为边缘控制器构建分布式危险响应端,软件结合了多传感器数据融合技术和基于模型的强化学习算法进行环境参数预测与危险判断。首先,利用FPGA连接RL78/G15传感器阵列作为边缘控制器,进行环境参数的准确采集和高速并行运算;其次,应用基于支持度的算法对多组传感器阵列的检测值进行数据融合;最后,改进基于模型的强化学习算法,结合蒙特卡洛树搜索进行数据预测以及危险判断。仿真结果显示,系统有效地减轻了计算任务,同时提高了准确性。

The paper primarily focuses on hazardous environments and presents a design of the distributed safety warning sys-tem based on the reinforcement learning method.The system comprises hardware components,with FPGA serving as the edge con-troller for distributed hazard response,and software components that integrate multi-sensor data fusion and reinforcement learning algorithms for environmental parameter prediction and hazard detection.Firstly,an RL78/G15 sensor array is employed as an edge controller on FPGA for accurate data acquisition and high-speed parallel computation.Secondly,a support-based algorithm is ap-plied to fuse the detection values from multiple sensor arrays.Finally,an improved model-based reinforcement learning algorithm is developed,incorporating Monte Carlo tree search(MCTs)for data prediction and hazard detection.This approach effectively reduces computational tasks while enhancing accuracy.