一种基于改进教学优化的微弱信号检测方法

为了快速准确地检测混沌背景中的微弱信号,提高网络泛化能力,文中利用改进教学优化算法优化贝叶斯回声状态网络的模型参数,提出了一种改进教学优化的混沌背景中微弱信号检测方法。通过建立混沌序列单步预测模型,分析预测误差的幅值,检测混沌背景中微弱瞬态信号和周期信号。对Lorenz系统和实测的海杂波数据进行实验研究,验证预测模型的有效性,结果表明,贝叶斯回声状态网络模型的预测结果比支持向量机和径向基神经网络模型的均方根误差降低了2个数量级,缩短了预测时间,提高了预测精度和预测效率,能快速有效地检测混沌背景中微弱信号,且具有更低的门限。

A Weak Signal Detection Method Based on Improved Teaching Learning-based Optimization

In order to detect the weak signal in chaotic background accurately and quickly, a Bayesian echo state network method based on improved teaching learning based optimization algorithm is proposed. To improve the generalization ability of the network, Bayesian theory is combined with echo state network. Since the parameters of echo state network is susceptible affected by subjective factors and the method based on direct echo state network ignores the uniqueness of the data, the improved teaching learning based optimization algorithm is used to optimize the parameters of the Bayesian echo state network model. The algorithm includes three stages: teaching, learning and feedback. Poor students in the algorithm can not only learn from the students but also feedback with the teacher at all times, so that it becomes a good student as soon as possible, the algorithm''s global search ability and the speed of searching are improved. By establishing the chaos sequence single step prediction model, the amplitude of the prediction error is analyzed to detect the weak transient signal and the periodic signal in the chaotic background. The experimental results show that the root mean square error of the Bayesian echo state network model are reduced by two orders of magnitude than those of the support vector machine and the radial basis function neural network model. Also, the prediction time is shortened, the prediction accuracy and prediction efficiency is improved. The method can quickly and effectively detect weak signal from the chaotic background and has a lower threshold.