Wearable WSN Data Fusion Method Based On Neural Network

Wearable wireless sensor network nodes are generally powered by battery, The energy is limited, at the same time, there are a large amount of redundant data and conflicted data. Data fusion technology can effectively reduce the amount of communication data during the transfer phase, It can help sensor nodes to reduce energy consumption and prolong sensor network lifetime. In this paper, Mainly research data fusion method that based on neural networks and fuzzy neural networks.In order to solve network convergence oscillation problem, BY improving the convergence rate of learning algorithm, we propose a momentum-adaptive learning rate backpropagation learning algorithm (A Momentum Adaptive Learning Rate Algorithm, MALRBPA). To solve the two-dimensional XOR problem as the background. The improved learning algorithm, standard learning algorithm and traditional improved learning algorithm convergence rate was validated. The neural network of using MALRBPA learning algorithm convergence rate has been significantly improved. Through combine the improved BP neural network and wireless sensor network clustering routing protocol (LEACH-F protocol), an BP neural network data fusion algorithm (A Back Propagation Neural-Network Data Fusion Algorithm, BPNDFA) was proposed, within the cluster of sensor networks, the raw data collected by data fitting, To feature data using CDMA-coded and send them to the sink node, in order to accelerate Neural network learning and training convergence speed, reduce the amount of communication data of node. Simulation results show that compared with the traditional BP neural network data fusion algorithm, BPNDFA algorithm reduces the energy consumption of nodes and prolong the network life cycle. With good accuracy, validity and timeliness in monitoring human characteristic parameters.Finally, the use of TS-fuzzy inference system (Takagi-Sugeno) structure, Training fuzzy neural network with MALRBPA algorithms, fuzzy rules, while introducing wireless sensor network cluster routing technology, presents a data fusion based on fuzzy neural network algorithm (a Fuzzy Neural networks data fusion algorithm, FNNDFA), simulation results show that, compared with traditional data fusion algorithm based on fuzzy neural network, FNNDFA network algorithm has a higher rate of convergence and accuracy of prediction, that could reduce the wearable sensor node energy consumption, extend the life of the network, this data fusion algorithm is suit for parameter indexs that are not determined values but a blur range in wearable wireless sensor networks...