LAUSR

To increase reliability, reduce machine tool failure, and shorten maintenance time, serval failure distribution models are discussed based on the cycloidal gears form grinding. A fault data expansion algorithm model based on the radial basis function (RBF) neural network is proposed to accurately evaluate the reliability of cycloidal gears form grinding machines. The model uses a self-organizing clustering learning algorithm to determine the RBF centers and expansion constants of the neural network. It trains the neural network using the learning algorithm’s output and imports the randomly generated cumulative failure distribution function to obtain simulation data. A numerical example of researching failure distribution is presented which shows that the estimated value of mean time between failures (MTBF) is 909.20h, and the estimated value of reliability is 0.4874. Finally, the fault tree analysis-analytic hierarchy process (FTA-AHP) is used to analyze the fault tree of the cycloidal gear grinding machine. The reliability evaluation indexes are obtained by analyzing the corresponding reliability characteristic function, which proves the failure distribution model is valid. It has important guiding significance in evaluating the reliability and performance of large computer numerical control (CNC) form gear grinding machines.