论文部分内容阅读
针对海浪水压场的短时平稳性,采用局域支持向量机进行预测滤波。首先分析了海浪水压场信号的混沌特性,并根据其混沌特性,提取训练空间中与当前相点的行为特征密切相关的最近邻点作为训练样本对支持向量机进行训练,减少了训练样本的数目,节省了网络学习时间,从而可实时对网络参数进行更新,使支持向量机能够跟随海浪的变化。实际计算表明这种算法能够以较快的学习速度和较高准确度实现海浪预测,能够克服由于海浪的短时平稳性所带来的随时间的增长预测精度下降的问题。
Aiming at short-term stability of wave pressure field, local support vector machine is used to predict and filter. Firstly, the chaos characteristics of wave pressure signals are analyzed. Based on its chaotic characteristics, the nearest neighbors in the training space that are closely related to the behavior characteristics of the current point are extracted as training samples to reduce the training samples The number of network learning time is saved, so that the network parameters can be updated in real time so that the support vector machine can follow the change of the sea wave. The actual calculation shows that this algorithm can realize the wave prediction with faster learning speed and higher accuracy, and can overcome the problem of the decline of the prediction accuracy over time caused by the short-term stability of the wave.