Machine learning based online fault prognostics for nonstationary industrial process via degradation feature extraction and temporal smoothness analysis

Fault degradation prognostic, which estimates the time before a failure occurs and process breakdowns, has been recognized as a key component in maintenance strategies nowadays. Fault degradation processes are, in general,slowly varying and can be modeled by autoregressive models. However, industrial processes always show typical nonstationary nature, which may bring two challenges: how to capture fault degradation information and how to model nonstationary processes. To address the critical issues, a novel fault degradation modeling and online fault prognostic strategy is developed in this paper. First, a fault degradation-oriented slow feature analysis(FDSFA) algorithm is proposed to extract fault degradation directions along which candidate fault degradation features are extracted. The trend ability assessment is then applied to select major fault degradation features. Second, a key fault degradation factor(KFDF) is calculated to characterize the fault degradation tendency by combining major fault degradation features and their stability weighting factors. After that, a time-varying regression model with temporal smoothness regularization is established considering nonstationary characteristics. On the basis of updating strategy, an online fault prognostic model is further developed by analyzing and modeling the prediction errors. The performance of the proposed method is illustrated with a real industrial process.

Parameter Optimization Method for Gaussian Mixture Model with Data Evolution

To learn from evolutionary experimental data points effectively,an evolutionary Gaussian mixture model based on constraint consistency(EGMM)is proposed and the corresponding method of parameter optimization is presented.Here,the Gaussian mixture model(GMM)is adopted to describe the data points,and the differences between the posterior probabilities of pairwise points under the current parameters are introduced to measure the temporal smoothness.Then,parameter optimization of EGMM can be realized by evolutionary clustering.Compared with most of the existing data analysis methods by evolutionary clustering,both the whole features and individual differences of data points are considered in the clustering framework of EGMM.It decreases the algorithm sensitivity to noises and increases the robustness of evaluated parameters.Experimental result shows that the clustering sequence really reflects the shift of data distribution,and the proposed algorithm can provide better clustering quality and temporal smoothness.

Road model prediction based unstructured road detection

Vision-based road detection is an important research topic in different areas of computer vision such as the autonomous navigation of mobile robots.In outdoor unstructured environments such as villages and deserts,the roads are usually not well-paved and have variant colors or texture distributions.Traditional region- or edge-based approaches,however,are effective only in specific environments,and most of them have weak adaptability to varying road types and appearances.In this paper we describe a novel top-down based hybrid algorithm which properly combines both region and edge cues from the images.The main difference between our proposed algorithm and previous ones is that,before road detection,an off-line scene classifier is efficiently learned by both low- and high-level image cues to predict the unstructured road model.This scene classification can be considered a decision process which guides the selection of the optimal solution from region- or edge-based approaches to detect the road.Moreover,a temporal smoothing mechanism is incorporated,which further makes both model prediction and region classification more stable.Experimental results demonstrate that compared with traditional region- and edge-based algorithms,our algorithm is more robust in detecting the road areas with diverse road types and varying appearances in unstructured conditions.