Efficient Hidden Danger Prediction for Safety Supervision System: An Advanced Neural Network Learning Method

Abstract

Hidden danger prediction plays an important role in safety production and safety supervision. To improve the hidden danger prediction accuracy of tertiary industries in some small-medium cities, this paper utilizes extreme learning machine (ELM) algorithm to study the impact of relevant management index on the trend of hidden danger, and conduct hidden danger prediction. ELM is a novel learning algorithm for single hidden layer feedforward neural network (NN) with fast learning speed and good generalization performance. We use the nationwide enterprise hidden danger data to conduct the prediction experiment, and the comparisons between traditional NN learning method and ELM demonstrate the effectiveness and superiority of our method.