论文部分内容阅读
无线传感器网络中多传感器节点在采集数据时,客观因素的本质决定了数据间存在时间相关性和空间相关性.数据预测机制能有效挖掘数据间的时空相关特性,从而减少节点间冗余数据的传输.通过利用粒子群优化算法训练BP神经网络,提出了一种带PSO-BPNN的时空预测算法,根据单预测机制的准确率,将时间预测和空间预测的结果进行加权求和.一方面改善了传统预测算法容易陷入局部最优及过学习的缺点,另一方面有效克服了单预测机制的盲目性,提高了预测准确度.实验结果表明了算法的有效性.
When collecting data, the nature of the objective factors determines the time-dependent and spatial correlation between the data.Multi-sensor nodes in WSN can effectively predict the spatio-temporal correlation between data and reduce the redundant data between nodes Transmission.Through the training of BP neural network using Particle Swarm Optimization algorithm, a PSO-BPNN spatio-temporal prediction algorithm is proposed, based on the accuracy of the single prediction mechanism, the results of time prediction and spatial prediction weighted summation, on the one hand to improve The traditional prediction algorithm is easy to fall into the local optimal and over-learning shortcomings, on the other hand it effectively overcomes the blindness of the single prediction mechanism and improves the prediction accuracy.The experimental results show the effectiveness of the algorithm.