模糊强化学习的改进GEAR传感网络均衡算法  

Improved GEAR Sensor Network Equalization Based on the Fuzzy Reinforcement Learning Algorithm

作　　者：安子强[1] 

机构地区：[1]新乡学院计算机与信息工程学院,河南省新乡市453000

出　　处：《计算机仿真》2013年第6期279-283,共5页Computer Simulation

基　　金：河南省政府决策研究招标课题(2011B637)

摘　　要：在网络节点分配均衡优化中,需要依靠地理位置与能量开销等先验知识才能进行调度,但传统GEAR节点分配算法中,传感节点的有效性不强,降低了通信效率。在分析GEAR传感节点分配算法基础上,提出模糊强化Q学习的改进GEAR传感网络均衡算法。运用模糊神经网络对强化学习中的Q值进行逼近,把Q值与无线传感节点分配过程相结合,根据模糊Q值对传递节点进行选择。提高通信效率,通过仿真结果验证了改进方法的通信能量消耗曲线趋势要低于传统GEAR算法,能有效避开网络拥塞区域,通信效率得到了显著的提高。The traditional GEAR node distribution algorithm needs to rely on the geographical position, energy costs and prior knowledge for the effectiveness of the scheduling, a single sensor node which is not strong can reduce the communication efficiency. Based on the analysis of the traditional GEAR sensor node distribution algorithm, an improved GEAR sensor network equalization algorithm was proposed based on fuzzy strengthen Q learning. The fuzzy neural network was used to approach the Q value of reinforcement learning, the Q value and wireless sensor node dis- tribution process were combined, and the transfer node was selected according to the combination of fuzzy Q value. The simulation results show that this method is lower than the traditional GEAR algorithm in term of communication energy consumption curve trend can effectively avoid network congestion area, and the communication efficiency has been improved significantly.

关 键 词：模糊神经网络 强化学习 算法 

分 类 号：TP116[自动化与计算机技术—控制理论与控制工程]