A hybrid ARIMA–WNN approach to model vehicle operating behavior and detect unhealthy states

As modern vehicles system becomes increasingly complex, there is an urgent need to develop a framework monitor the behavior and detect unhealthy states appropriately arrange maintenance in order extend vehicle life cycle. Sensors installed are able record huge amount of multiple channel time series data. This paper develops prediction model state detection strategy for monitoring operating by analyzing data channels. Due complexity underlying process data, hybrid ARIMA–WNN proposed combining Auto-Regressive Integrated Moving Average (ARIMA) Wavelet Neural Network (WNN) models. exploits ARIMA capacity predicting unseen patterns without labeled historical WNN’s flexibility different variations Furthermore, threshold-based anomaly developed timely recognize vehicle. To test performance ARIMA–WNN, large scale multiple-channel including three months second-wise records 101 channels used. Results reveal higher accuracy compared with WNN models modeling By estimating proper distribution threshold, method also efficient identifying • Develop based time-series approach operating. Use set validate approach.