Neural Networks Based Sensor Design and Evaluation

This paper mainly focuses on the modelling of the frequency output and evaluating the performance of the high precision humidity and temperature sensor which is designed for a wireless sensor network based on a kind of neural networks. Firstly, some backgrounds of neural networks and the sensor are described, and then, the two problems in the design of the sensor are given. Secondly, the possibility of solving the two problems using neural networks is given. Thirdly, in order to solve the two problems, a kind of three-layer multilayer perceptrons (MLP) networks is constructed to testify the new modelling method. Finally, experiment results are presented to show the feasibilities and effectiveness of the neural networks method.