Research On Application Of Artificial Neural Network In Evaluation Of Intrinsically Safe Circuit

At present,there are not any prosperous theory on breaking through the research of the judgement of intrinsically safe circuit by using compute.Artificial neural network is a complicated information processing one made of many processing units. This network has the ability of learning memory and input information trait extracting.In this paper,There ara five factors mainly which will make some affection on the judgement of intrinsically safe circuit were eastablished basing on the analyzing the theory of flame-proof.Getting a great deal of data samples,the neural network models for predicting intrinsically safe circuit was established basing on the five factors mainly, which includes defining the number of network layer, defmingthe number of hidden layer neurons,initializing the weights and selecting the expected error,the models has been tested also.The Back-Propagation Network models for predicting was eastablished on the basis of MATLAB toolbox.In order to optimize BP algorithm ,some effect measures,unitary treatment and variable learn rate algorithm ,were used in the paper . The training process showed that the convergent speed is very fast by using Levenberg-Marquardt algorithm, And the models developed have good fitting capability compared with experiments by simulation.Use artificial neural network ,we can made the success on the judgement of intrinsically safe circuit base on compute.