基于量子粒子群优化算法的压缩感知数据重构方法  

Perceptual Data Reconstruction for Compressed Sensing Based on Quantum Behaved Particle Swarm Optimization

作　　者：刘洲洲[1] 李艳平[2] 

机构地区：[1]西安航空学院,西安710077 [2]菏泽学院计算机与信息工程系,山东菏泽274015

出　　处：《传感技术学报》2015年第6期836-841,共6页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(No.61103242)

摘　　要：针对传感器监测对象特点,将压缩感知理论应用于数据压缩过程以降低通信能耗,并根据现有压缩感知数据重构算法存在的重构精度受稀疏度影响较大的缺点,在分析了压缩感知数据重构原理后,提出了将原始信号按固定长度进行分帧处理以减少算法解空间的数量,并将量子理论中的编码方式应用于粒子群优化算法,提出了基于量子粒子群优化算法的压缩感知数据重构方法 QP-CSDR。算法根据传感器监测对象特点,从统计学角度出发对粒子群优化算法中的粒子初始位置及粒子群更新方式加以改进,以提高数据重构精度。仿真实验结果表明,在稀疏度小于50的条件下,QP-CSDR算法相对已有算法在重构精度方面性能提升20%-40%,该算法已应用于微地震及音频监测系统中,经实际检验算法在保证数据精度的前提下延长系统寿命2倍-4倍左右。According to wireless sensor network monitoring object features, the compressed sensing theory is applied to data compression to reduce the communication energy. Considering that reconstruction accuracy of existing data reconstruction in compressed sensing can be easily influenced by sparsity, after analysis of compressed sensing data reconstruction principle, with sub-frame processing the original signal in fixed length to reduce the solution space, and applying quantum theory encoding in Particle Swarm Optimization, Compressed Sensing Data Reconstruction that based on Quantum-behaved Particle Swarm Optimization appears. According to wireless sensor network monitoring object features, this algorithm improves the accuracy of the data reconstruction by improving particle initial position and update mode in Particle Swarm Optimization from Statistics. Simulation results show that under conditions of sparsity less than 50, QP-CSDR gets 20%- 40% performance improvement on Reconstruction accuracy comparing to existing algorithms. Now the algorithm has been applied to micro-earthquakes and audio monitoring system, and in actual inspection, the actual system life is extended about 2-4 times with assurance data accuracy.

关 键 词：量子理论 粒子群优化算法 压缩感知 数据重构 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]