CAJ | 학술논문

针对传统方法单独采用 BP 神经网络算法易陷入局部极值的问题，提出了遗传算法优化 BP 神经网络，并将其应用于 MIMO -OFDM系统信号检测中。该方法将遗传算法与神经网络相结合，用遗传算法优化神经网络初始值，使 BP 网络快速收敛到最优解，避免了由初始值的随机选取而带来的检测误码。仿真结果表明所提出的方法在误码率方面有比较好的性能。
침대전통방법단독채용 BP 신경망락산법역함입국부겁치적문제，제출료유전산법우화 BP 신경망락，병장기응용우 MIMO -OFDM계통신호검측중。해방법장유전산법여신경망락상결합，용유전산법우화신경망락초시치，사 BP 망락쾌속수렴도최우해，피면료유초시치적수궤선취이대래적검측오마。방진결과표명소제출적방법재오마솔방면유비교호적성능。
In view of the problem of easily trapping in local minima caused by the traditional method which using BP neural network algorithm alone,the genetic algorithm is put forward to optimize the BP neural network,and applied to signal detection in MIMO -OFDMsystem.The method,combining genetic algorithm and neural network,optimizes neural network initial values by genetic algorithm for the BP network conver-gence to the optimal solution quickly in order to avoid test error by the initial value of the random selection. The simulation results show that the presented method has a better performance in bit error rate.