Clustering-based System Identification of Output Error Moving Average Systems with Time Delays

The presence of time-varying delays introduces uncertainty of system controllers and the development of identification models. Hence, an unsupervised clustering algorithm for identifying output error moving average systems with time delays is proposed. The auxiliary model idea addresses estimation issues where the information vector contains unknown inner variables and unmeasurable noise. The set of time delays is constructed and the true values are identified through a self-organizing mapping approach. A probability density function is formulated to correlate time delays with weights, enabling the simultaneous identification of model parameters and time delays within the expectation-maximization framework. A simulation example illustrates the effectiveness of the proposed algorithm.