基于BP神经网络的多类运动想象脑电信号识别系统研究

Research on The Recognition System of Multi-Class Motor Imagery EEG Signals Based on BP Neural Network

脑电信号识别指令多设定为单目标,导致系统误识率高,为此研究基于反向传播(Back Propagation,BP)神经网络的多类运动想象脑电信号识别系统。通过电极帽与脑电采集板的构建与搭接,完成系统硬件的设计;进行BP神经网络多目标识别指令集群的接入,同时利用分段BP神经网络功能识别模块增强系统的识别控制能力,完成系统软件的设计。最终的测试结果表明:对上述系统进行3个阶段,5组测试之后,系统在第3阶段的脑电波误识率明显控制在2%以下,系统的误识别率较低,具有较高的识别应用价值。

The EEG signal recognition instructions are mostly set to single target,which leads to high false recognition rate of the system.To this end,we study the multi-class motion imagination EEG signal recognition system based on Back Propagation(BP)neural network.The system hardware design is completed through the construction and lap of electrode cap and EEG acquisition board;the access of BP neural network multi-target recognition command cluster is carried out,and the recognition control of the system is increased by using segmented BP neural network function recognition module to complete the system software design.The final test results show that:after conducting 3 stages and 5 groups of tests on the above system,the false recognition rate of EEG waves in stage 3 of the system is obviously controlled below 2%,indicating that the false recognition rate of the system is lower and has stronger recognition application value..