Comparison of neural network algorithms based on gas qualitative analysis

Abstract

For the problem of gas qualitatively identify in the field of gas detection, this paper is based on the multi-sensor and pattern recognition of neural network, the uniform change voltage of the sensor output is simulated by the gradient descent algorithm, the additional momentum algorithm and the LM algorithm of neural network, then compare the three simulation results of the three algorithms, the result proves that the LM algorithm is the optimal algorithm of the data simulation in this paper, in the range of allowable error, completed the gas qualitative identification.