Improving Semi-Supervised Learning for Remaining Useful Lifetime Estimation Through Self-Supervision

RUL estimation plays a vital role in effectively scheduling maintenance operations. Unfortunately, it suffers from a severe data imbalance where data from machines near their end of life is rare. Additionally, the data produced by a machine can only be labeled after the machine failed. Both of these points make using data-driven methods for RUL estimation difficult. Semi-Supervised Learning (SSL) can incorporate the unlabeled data produced by machines that did not yet fail into data-driven methods. Previous work on SSL evaluated approaches under unrealistic conditions where the data near failure was still available. Even so, only moderate improvements were made. This paper defines more realistic evaluation conditions and proposes a novel SSL approach based on self-supervised pre-training. The method can outperform two competing approaches from the literature and the supervised baseline on the NASA Commercial Modular Aero-Propulsion System Simulation dataset.

A Nonlinear Prognostic Model Based on the Wiener Process with Three Sources of Uncertainty

Estimation of the remaining useful life (RUL) is an important component of prognostics and health management (PHM). The accuracy of the RUL estimation for complex systems is mainly affected by three sources of uncertainty, i.e., the temporal uncertainty, the product-to-product uncertainty, and measurement errors. To improve PHM and account for the effects of the three sources of uncertainty, a nonlinear prognostic model with three sources of uncertainty is presented here. An approximated analytical expression for the probability density function (PDF) of the RUL is obtained based on the concept of first hitting time (FHT). Model parameters are then obtained by the expectation maximization (EM) algorithm, and the drift parameter is estimated adaptively using a Bayesian procedure. Finally, in order to illustrate the practical applications of the presented approach, a comparative study of real data on fatigue crack propagation is presented. Results demonstrate that our method improves model fit and increases the accuracy of the lifetime estimation.

Lifetime Estimation Based Health Index and Conditional Factor for Underground Cable System

In this paper, a lifetime estimation method for underground cable systems is proposed by combining a health index (HI) and conditional factor (CF). The underground cable system consists of underground cable, joint, termination, manhole, and duct bank. The HI is an indicator to indicate the actual condition of underground cable components and systems whereas the CF is used to indicate different operating stresses of the system under different operating conditions such as percentage loading, electrical stresses, laying structure, environment, etc. The actual technical data as well as historical operating and testing records are applied. The weighting and scoring method with the analytical hierarchy process are used to classify an importance of underground cable components, testing methods, and criteria used in the HI and CF calculation. The annual calculated HIs are plotted to investigate the lifetime trending curve by using a polynomial function. The degradation curve based on calculated CF is estimated by using the Weibull distribution function. Finally, the remaining life of the underground cable system is determined by matching the lifetime trending curve with the degradation curve. Ten practical underground cable systems supplying power in a high voltage power delivery system are evaluated with effective results. The lifetime of the underground cable system can be successfully estimated.

The Prediction of CMDB Propellant lifetime Based on Arrhenius accelerated model, Berthelot’s equation and Multi-step Prout-Tompkins Model

Different models were provided to predict the storage lifetime of propellants more accurately. The stabilizer was recognized as a vital parameter for double based propellants’ storage lifetime estimation. The stabilizer contents of a certain RDX-CMDB propellant were traced during the accelerated aging tests. Based on that, the safe storage lifetime of this propellant were predicted using the Berthelot’s equation, Arrhenius accelerated equation and the advanced kinetic model, respectively. The predicted results were compared and the causes were analysed. It found that the biggest disadvantage of Berthelot’s equation and Arrhenius accelerated equation is that the predicted results are significantly affected by the original data. In details, the minor difference of original data will bring tremendous errors when extrapolating to normal temperature. The general model which can be used to depict complex reactions adopted in AKTS software was preferred compared to reaction order (RO) model and Prout-Tompkins (PT) model.

Remote Microgrids for Energy Access in Indonesia—Part II: PV Microgrids and A Technology Outlook

This paper is the companion paper of Remote Microgrids for Energy Access in Indonesia “Part I: scaling and sustainability challenges and a technology outlook”. This part II investigates the issues of photovoltaic (PV) systems with respect to the planning, design, and operation, and maintenance phases in microgrids in Indonesia. The technology outlooks are also included as PV has an important role in providing electricity in the underdeveloped, isolated, and border areas. The data in this paper are from PV microgrids located in Maluku and North Maluku, which are two provinces where there is barely any grid connection available and thus very dependent on remote microgrids. The data are obtained from interviews with Perusahaan Listrik Negara (PLN) and NZMATES, which are an Indonesian utility company and a program for supporting role for the PV systems in Maluku funded by New Zealand respectively. Common issues with respect to reliability and sustainability are identified based on the provided data. Advanced technologies to increase reliability and sustainability are also presented in this paper as a technology outlook. Among these solutions are online monitoring systems, PV and battery lifetime estimation, load forecasting strategies, and PV inverters technology.