Application of Neural Network Algorithm Based on PCA-BP in Earthquake Early Warning of Buildings

This paper established an efficient and practical earthquake damage prediction method for buildings by combining principal component analysis (PCA) with neural network. To avoid BP neural network from being caught in a local minimum, according to the features of PCA algorithm, this paper combined the two to form PCA-BP mixed model and trained the network through the initial weight of PCA-optimized neural network. Based on a collection of large quantities of earthquake damage data of buildings, this model was introduced in earthquake early warning for buildings. The results show that this method can predict earthquake for buildings in an effective and accurate way.