基于优化FAEMD-OSELM的WSN流量预测算法研究

Research on WSN traffic prediction algorithm based on optimized FAEMD-OSELM

针对无线传感器网络(WSN)组合流量预测算法自适应性低、计算复杂度高等不足,提出一种基于优化FAEMD-OSELM的WSN流量预测算法。算法利用快速自适应经验模态分解(FAEMD)将信号分解为一系列本征模态函数和一个残余函数,通过设计自适应滤波窗口提高信号分解过程的自适应性。进一步,采用在线贯序极限学习机(OSELM)对信号分量进行训练、预测,运用奇异值分解(SVD)理论优化神经网络的参数和拟合过程,降低计算复杂度。同时,结合样本选择器进一步控制预测误差范围,保证算法的预测精度。实验结果表明,算法在分解效果、耗机时间、预测精度等关键性能指标上具有较为明显的优势。

Aiming at the deficiency of low adaptability and high computational complexity of the combined traffic prediction algorithms for wireless sensor networks(WSN),this paper proposes a WSN traffic prediction algorithm based on optimized FAEMD-OSELM.The algorithm uses fast and adaptive empirical mode decomposition(FAEMD)to decompose the signal into a series of intrinsic mode functions and a residual function,and an adaptive filtering window is designed to improve the adaptability of the signal decomposition process.Furthermore,the online sequential extreme learning machine(OSELM)is used to train and predict the signal components,and singular value decomposition(SVD)theory is used to optimize the parameters and fitting process of the neural network to reduce the computational complexity.Meanwhile,the sample selector is used to further control the prediction error range to ensure the prediction accuracy of the algorithm.The experiment results indicate that the proposed algorithm has obvious advantages in the key performance specifications such as decomposition effect,machine consumption time and prediction accuracy.