Wireless Sensor Networks (WSN) consists of large number of sensor nodes. The sensor nodes are battery powered devices, they communicate over a wireless medium and consumes energy during data transmission. In the past years various techniques have been proposed to reduce the battery consumption of wireless sensor networks. The previous approaches are based on the static clustering which follow the rules of LLC algorithm. In this paper, novel technique is been implemented which is based on the Dynamic clustering. The Cluster heads in each cluster have changed according to the network conditions.  Dynamic clustering is being proposed in this paper which is based on the neural networks. The proposed technique is implemented in NS2 and simulation results show that novel technique will reduce network overhead and increase network lifetime.