Neural network-based method for virtual age estimation for predicting remaining lifetime of a device of a given type for predicting optimum maintenance intervals in preventative maintenance systems

摘要

Method has the following steps: monitoring of a preset number of significant parameters of the corresponding device for a training amount of device of the given type, where the parameters represent usage increments; determination of neural network coefficients with a radial base function for modeling the usage increments, that have been determined from the training quantity of devices that have been operated until failure and their virtual age normalized. A formula for the virtual age is derived from the data.