摘要
针对常规BPNN(Back Propagation Neural Network)容易陷入局部最优解、收敛速度慢等问题,提出了一种基于小生境思维进化NMEA(Niche Mind Evolutionary Algorithm)及BPNN的传感器数据流异常检测算法(NMEA-BP)。该算法利用NMEA的全局搜索性优化BPNN的参数,获得BPNN的最优权阈值,从而提高异常检测的准确性。为了评估算法的性能,使用因特尔伯克利实验室数据集IBRL(Intel Berkeley Research Lab)及带标记的传感网络数据集LWSNDR(Labeled Wireless Sensor Network Data Repository)完成了仿真实验,并与基于常规BPNN、支持向量机(Support Vector Machine)和极限学习机(Extreme Learning Machine)等3种异常检测算法作对比。仿真实验结果表明,与上述3种算法相比,NMEA-BP算法对各个数据集都具有较高的检测率和较低的误报率,检测率平均达到99.45%,误报率平均仅为1.45%。此外,NMEA-BP异常检测算法的模型训练时间比传统的BPNN异常检测算法平均减少30%以上。
The conventional BP neural network（ BPNN） is prone to be trapped in the local minima,and its training process converges slowly. In this paper,a data stream anomaly detection algorithm（ NMEA-BP） is proposed for sensor networks. The proposed algorithm is based on NMEA and BPNN. The parameters of BPNN are optimized by NMEA＇s global search,and then the optimized weights and thresholds of BPNN can be obtained. Thus the accuracy of anomaly detection algorithm is improved. To evaluate the performance of proposed anomaly detection algorithm,some experiments are carried out with Intel Berkeley dataset（ IBRL） and Labeled dataset（ LWSNDR）,and NMEABP is compared with other algorithms such as BPNN,SVM and ELM. The simulation results show that,NMEA-BP has higher detection rate and lower false alarm rate than the other three algorithms,with an average detection rate of99.45% and an average false positive rate of 1.45%. In addition,NMEA-BP can reduce the training time more than30% compared with the traditional BPNN.
作者
顾晓勇
李光辉
GU Xiaoyong;LI Guanghui(School of computer technology,Jiangnan University,Wuxi Jiangsu 214122,China;Jiangsu High Technology Research Key Laboratory forWireless Sensor Networks,Nanjing 210003,China;Research Center of IoT Technology Application Engineering（MOE）,Wuxi Jiangsu 214122 China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2018年第5期746-752,共7页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金项目(61472368)
江苏省重点研发计划项目(BE2016627)
中央高校基本科研业务费(JUSRP51635B)
无锡市国际科技研发合作项目(CZE02H1706)
关键词
无线传感网络
参数优化
BP神经网络
异常检测
仿真
wireless sensor network
parameter optimization
BP neural network
anomaly detection
simulation
