Application of GA-improved wavelet network in temperature compensation of sensor

Precise identification of the temperature compensation of sensor is of significance for the improvement of the precise testing of the system. Neural network has such capacities as self-learning, adaptive and non-linear mapping. However, it is often slow in training speed, and is easy to be trapped in local minimum value. While, genetic algorithm (GA) has very strong, global optimization searching capability but it is insufficient in local searching. This paper has explored the utilization of GA-improved wavelet neural network to obtain the global optimal solution. The measured data under multiple temperature conditions have been referred to so as to carry out effective identification of the temperature compensation model of eddy current sensor. The result shows that this method is quick in operation, high in precision and strong in generality. It has very promising application prospect in the areas such as intelligent sensor modeling and compensation.