Abstract: The accurate control of the temperature of work pieces is a nonlinear uncertain complicated control problem in vacuum annealing furnace. In order to control the work piece temperature accurately, the optimization model for accurately controlling work piece temperature is proposed by the data gathered from the scene. The model is set up based on wavelet neural networks, and an adaptive immune genetic algorithm is used to optimize the number of elements in hidden layer, weights, dilation and translation parameters of wavelet neural networks. Simulation and experiment results show that the model based on the method in this paper is better than the model based on BP neural networks. And, it improves the training rate of networks and steady state precision of system, and obtains a system that has good characteristics of real time and robustness.