Research on Invalid Detection Data Model of Mine Catalytic Sensors Based on Machine Learning

Abstract

In order to solve the problem of nonlinear failure data output by the catalytic combustible sensor (later referred as a sensor) when working in the mine environment, this article proposes a backpropagation (BP) neural network nonlinear data filtering model based on the Levenberg–Marquardt (L-M) algorithm. The experimental analysis shows that this model has obvious advantages in training speed and error performance compared with the BP neural network model established by a quasi-Newton algorithm and adaptive linear regression (lr) momentum gradient descent algorithm. In terms of generalization ability, this model has better generalization ability than the radial basis function (RBF) feedforward neural network model with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${K}$ </tex-math></inline-formula> -means clustering. Based on the above advantages, the model can effectively filter the sensor failure output data, eliminate the hidden danger of safety production caused by the failure output data, and improve the level of safety production in coal mines.