Although intelligent machine learning techniques have been used for input-output modeling of many different manufacturing processes, these techniques map directly from the input process parameters to the outputs and do not take into consideration any partial knowledge available about the mechanisms and physics of the process. In this paper, a new approach is presented for taking advantage of the partial knowledge available about the mechanisms of the process and embedding it into the neural network structure. To validate the proposed approach, it is used to create a forward prediction model for the process of electrochemical micro-machining (μ-ECM). The prediction accuracy of the proposed approach is compared to the prediction accuracy of pure neural structure models with different structures and the results show that the Neural Network (NN) models with embedded knowledge have better prediction accuracy over pure NN models.

Rolling element bearing is a critical component in many mechanical systems in view of its critical functionality. One of the major issues industries face today is the failure of bearings, which results in catastrophic consequences. Although various prognostic approaches were proposed for the degradation of bearings, the incapability of adaptation of those models yields inaccurate predictions under different running conditions of the bearings. To address this issue, this paper proposes a prognostic algorithm using the variable forgetting factor recursive least-square (VFF-RLS) combined with an auto-regressive and movingaverage (ARMA) model. The structure and parameters of ARMA model were initially determined using the vibrational data of the bearing without significant defect presented. During the bearing degradation process, the ARMA model makes predictions of the future degradation trend. Once the future acquired signal becomes available, the error between the acquired and predicted vibrational signal is calculated. The VFF-RLS algorithm uses the calculated error, correlation matrix and other parameters to update the coefficients of the ARMA model. In addition, the VFF-RLS algorithm updates the forgetting factor during each iteration to achieve faster convergence and reduced error. The updated ARMA model makes new predictions and the adaptive process continues. To demonstrate the applicability of adaptive prognosis methodology, the accuracy of the prediction of the proposed model is tested using experimental and simulated data in comparison with an auto-regressive integrated moving average (ARIMA) model without adaptation. Results show accurate predictions of the vibrational signal and degradation trend of the bearings over the ARIMA model.