Online Fault Prediction Research Articles

Disjunctive fault prediction of decentralized discrete event systems: Verification, predictor design and K-copredictability

This paper deals with the problem of fault predictability of discrete event systems (DES) modeled by automata. The predictability property is approached in a disjunctive decentralized architecture, where the verdict that a fault will inevitably occur is given when at least one local agent is sure about its future occurrence. We adapt previous results on diagnosability verification to develop two new strategies to verify disjunctive fault predictability (called copredictability): the first one is based on a diagnoser-like test automaton, and the second one on by using verifiers. Necessary and sufficient conditions for a regular language to be copredictable are provided, and also a strategy for designing local fault predictor systems, to be used in online fault prediction. We also address the problem of verifying if all fault occurrences can be predicted at least K events prior to their occurrences (K-copredictability) and provide both a necessary and sufficient conditions for a regular language to be K-copredictable and an automaton-based test to verify if a regular language is K-copredictable.

A Novel Fault Prediction Method of Wind Turbine Gearbox Based on Pair-Copula Construction and BP Neural Network

Among the various maintenance technologies of wind turbines, online fault prediction technology is a kind of more cost-effective and reliable method. It may also be the most promising method for wind turbines with potential mechanical faults. SCADA data-based online condition monitoring technology has become a hot spot in current researches. Therefore, a novel fault prediction method based on the Pair-Copula model is proposed in this study. First, the conditional mutual information method is introduced to screen out useful variables from a number of variables. Then aiming at the limitation that the conventional Copula model can only deal with two-dimensional variables, the Pair-Copula model is introduced. In addition, the complexity of the prediction model and the dimension of the input variables are greatly reduced by the Pair-Copula model. So, the BP neural network is selected to complete the prediction model. A combined model based on BP neural network and Pair-Copula model is proposed. In order to solve the problem that the conventional Pair-Copula model cannot process real-time data which must be required in fault prediction, a kind of improved Pair-Copula model combined with the kernel density estimation is used to calculate the real-time data. Finally, the proposed method is validated with real data from a 1.5 MW wind turbine, and the effectiveness is confirmed.

Fault Prediction via Symptom Pattern Extraction Using the Discretized State Vectors of Multisensor Signals

Fault prediction and early degradation detection have received considerable attention in many engineering disciplines. Fault symptoms can be identified by abnormal values or unusual trends in the monitored sensor signals over a certain period prior to fault occurrence. However, how to extract abnormal pattern, particularly those with conditional relations among multiple sensor signals, remains unclear. Pattern extraction is further difficult particularly when there is no gradient relationship between measurements and operational states due to highly scattered data and unclear boundaries for distinguishing operational states. Additionally, defining the time period for symptom periods is challenging. To resolve these issues, we define the terms symptom pattern and symptom period, and then present a symptom pattern extraction method that collects all evidence of potential fault occurrence from multiple sensor signals. We postulate that, given time markers of fault occurrences, a symptom period precedes the occurrence of a fault. Symptom patterns are defined as either only found in the symptom periods or not found in the given time series, but similar to fault patterns. We further discuss an iterative search procedure for determining the length of symptom periods and propose a severity assessment method for symptom patterns. Finally, we apply the symptom pattern extraction and severity assessment methods to an online fault prediction procedure. By assessing the total severity of patterns in the monitoring window, early warning decision can be made. The procedure is tested in the early detection of abnormal cylinder temperature in a marine diesel engine and automotive gasoline engine knocking.

Fault Prediction for Nonlinear System Using Sliding ARMA Combined with Online LS-SVR

A robust online fault prediction method which combines sliding autoregressive moving average (ARMA) modeling with online least squares support vector regression (LS-SVR) compensation is presented for unknown nonlinear system. At first, we design an online LS-SVR algorithm for nonlinear time series prediction. Based on this, a combined time series prediction method is developed for nonlinear system prediction. The sliding ARMA model is used to approximate the nonlinear time series; meanwhile, the online LS-SVR is added to compensate for the nonlinear modeling error with external disturbance. As a result, the one-step-ahead prediction of the nonlinear time series is achieved and it can be extended ton-step-ahead prediction. The result of then-step-ahead prediction is then used to judge the fault based on an abnormity estimation algorithm only using normal data of system. Accordingly, the online fault prediction is implemented with less amount of calculation. Finally, the proposed method is applied to fault prediction of model-unknown fighter F-16. The experimental results show that the method can predict the fault of nonlinear system not only accurately but also quickly.

Research on early Fault Prediction of Wind Turbine Gearbox

Fault of gearbox is one of the significant causes which lead to high cost of wind farm, so early fault prediction of gearbox is meaningful for ensuring reliable running and reducing maintenance costs. With condition monitoring data, the relation between gearbox temperature and potential faults was researched and a new method for online fault prediction of wind turbine gearbox was presented. First, the temperature prediction model for normal behavior of gearbox was built up by non-linear regression analysis. Then, a detecting function which can indicate the deviation between actual running state and prediction state of gearbox was introduced. The condition of gearbox could be monitored by comparing the real-time value of detecting function with the chosen threshold. Theoretical analysis and simulation results demonstrated that this method could predict the abnormality of gearbox in time, and it can be applied to monitor the running condition of gearbox.

Fault Prediction for Nonlinear Time Series Based on Temporal Pattern Estimation

An on-line Least Square Support Vector Regression (LS-SVR) algorithm is given in this paper. Based on this algorithm the method of on-line fault prediction by temporal pattern estimation is proposed. The method needs neither the model to approximate the true system nor the fault training data and primary knowledge. It can study and predict while system’s running, and it is believed with fast speed, fewer amounts of calculation and better real-time capability. The result of simulation on CSTR proved the efficiency.

A New Real-Time Reliability Prediction Method for Dynamic Systems Based on On-Line Fault Prediction

While a specific system is in use, its reliability will decrease gradually after the infant mortality period because of the components' degradation, or external attacks. Thus, reliability is a natural characteristic of a system's health, and can be used for condition monitoring & predictive maintenance. This paper introduces a new real-time reliability prediction method for dynamic systems which incorporates an on-line fault prediction algorithm. The factors that may reduce a system's reliability are modeled as an additive fault input to the system, and the fault is assumed to be varying linearly with time, approximately. The time-varying fault is roughly estimated based on a modified particle filtering algorithm at first. Then, as a time series, the fault estimate sequence is smoothed, and predicted by an exponential smoothing method. Mathematical analysis shows that the effects of the system, and measurement noises on the fault estimates are greatly reduced by exponential smoothing, which indicates that the comparatively high accuracy of the fault estimates & predictions is guaranteed. Based on the particle filtering & fault prediction results, the whole system's predictive reliability is computed through a Monte Carlo simulation strategy. The effectiveness of the proposed real-time reliability prediction method is validated by a computer simulation of a three-vessel water tank system.