多传感器信息融合模糊控制模型设计  

Design of Multi⁃Sensor Information Fusion Fuzzy Control Model

作　　者：陈婷婷[1] 赵世忠[2] CHEN Tingting;ZHAO Shizhong(College of Science and Technology,Nanchang University,Jiujiang Jiangxi 332020,China;College of Construction Engineering,Nanchang University,Nanchang Jiangxi 330036,China)

机构地区：[1]南昌大学科学技术学院,江西九江332020 [2]南昌大学建设工程学院,江西南昌330036

出　　处：《传感技术学报》2023年第6期911-915,共5页Chinese Journal of Sensors and Actuators

基　　金：江西省高校人文社会科学研究项目(TJ21201);2021年度江西省教育厅科学技术研究项目(GJJ217812)。

摘　　要：多传感器数据的来源众多,数据时间序列的特征随机性强,难以统一,导致其信息应用范围缩小。提出一种多传感器信息融合的模糊控制模型。运用不同映射模式描述多传感器信息融合状态空间,创建随机时段下测量空间矩阵,获得传感器信息采集时间序列特征。根据信息采集时间序列特征构建二级架构信息融合模型,第一级架构使用模糊控制算法划分输入-输出空间模糊区间,得到模糊规则并计算模糊规则相对信任度,利用模糊规则映射关联聚类信息,剔除传感器冗余数据。在此基础上使用智能粒子滤波法将多传感器信息传输至相应粒子滤波模块,代入遗传算法交叉与变异操作调整粒子权重,通过重采样保存高权值粒子,得到完整的多传感器信息融合结果。仿真结果表明,多传感器信息融合的最大能耗值为110 mJ,信息采集网络延迟为0.75 s,融合时间平均值为4.5 s,信息融合的误差值小于50 m,系统鲁棒性较强。There are many sources of multi⁃sensor data,and the characteristics of data time series are highly random and difficult to unify,which leads to the narrowing of the application range of the corresponding information.A fuzzy control model of multi⁃sensor information fusion is proposed.Different mapping modes are used to describe the state space of multi⁃sensor information fusion,and the measurement space matrix under random time period is created to obtain the time series characteristics of sensor information collection.The information fusion model of the second⁃level architecture is constructed according to the characteristics of the information collection time series.The first⁃level architecture uses the fuzzy control algorithm to divide the input⁃output space fuzzy interval,obtains the fuzzy rules and calculates the relative trust degree of the fuzzy rules,uses the fuzzy rules to map the associated clustering information,and eliminates redundant sen⁃sor data.On this basis,the intelligent particle filter method is used to transmit the multi⁃sensor information to the corresponding particle filter module,and the crossover and mutation operations of the genetic algorithm are used to adjust the particle weight.The high⁃weight particles are saved by resampling,and the complete multi⁃sensor information fusion result is obtained.The simulation results show that the maximum energy consumption of multi⁃sensor information fusion is 110 mJ,the information acquisition network delay is 0.75 s,the average fusion time is 4.5 s,the error value of information fusion is less than 50 m,and the system is robust.

关 键 词：多传感器 信息融合模型 模糊控制 粒子滤波 遗传算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]