ABSTRACT
This chapter focuses on adopting neural network with simulated annealing method, the model of landslide deformation prediction in long-term was set up, which using simulated annealing method to overcome the disadvantage of BP neural network, furthermore, by using dynamic forecasting technique to reduce the influence of the prophase displacement, it can get better forecast precision. In the model of simulated annealing with neural work, the network structure is the same as BP neural network, but use the simulated annealing method to search the weight and critical point threshold instead of the opposite propagation errors in BP neural work method. When cutting the accumulated data in certain time, training is carried out by using the BP neural network with simulated annealing method, and, after the training is finished, the prediction for deformation can be executed. The chapter discusses Tianhuangpin ‘‘3.29 landslide’’ deformation in long-term was predicted by using the model, the result is coincident to the real condition of the slope.
