Machine prognostics based on sparse representation model

The prognostic technologies for machines refer to the estimation of machines’ remaining useful life using monitoring data from sensors. Different from traditional maintenance strategies, this maintenance strategy can reduce downtime, maintenance costs and critical risks. Given these advantages, an increasing number of prognostic models are introduced. Data driven methods such as neural networks and Bayesian approaches are used widely in machine prognostics. However, the sequential information and inherent relationships among historical data are rarely considered in these models. So, the estimations are usually not accurate enough. In our paper, we take a novel methodology to estimate the remaining useful life: first, we adopt sparse representation model to extract the inherent relationships of training samples and measure the similarities between testing samples and training samples, and then a weight is given to every training sample to note its similarity to the testing sample. When all testing samples are measured, a hierarchical Hough voting process utilizing the sequential information of monitoring data is carried out to evaluate the remaining useful life. The industry experiment has proven the effectiveness of our approach.