基于狮群优化极限学习机的数据融合算法

Data aggregation algorithm based on extreme learning machine with lion swarm optimization

为降低异构无线传感器网络中冗余数据传输数量,设计一种基于狮群算法改进的极限学习机数据融合算法。针对草莓种植园实时监测数据的时间相关性,将双簇首分簇机制与LSO-ELM时间序列预测模型结合,对初始数据序列进行三次指数平滑和归一化的预处理,使用输出的预测值代替真实值,避免冗余数据的传输。仿真结果表明,双簇首机制可以均衡网络的通信负担,算法能够有效地降低网络中冗余数据的发送,保证数据准确性,延长网络生命周期。

To reduce the number of redundant data transmission in heterogeneous wireless sensor networks, a lion swarm optimization-extreme learning machine for data aggregation algorithm was designed. Aiming at the time correlation of real-time moni-toring data of strawberry plantation, the dual cluster head clustering mechanism of heterogeneous wireless sensor network was combined with the time series prediction model of LSO-ELM. The pretreatment of the initial data series was carried out for cubic exponential smoothing method and normalization, and the predicted value was used to replace the real value, so as to avoid the transmission of redundant data. The simulation results show that dual cluster head mechanism can balance the communication burden of network, and the algorithm can reduce the transmission of redundant data in the network effectively, ensure the accuracy of data and extend the network life cycle.