Identification of continuous-time hybrid ‘Box-Jenkins’ systems having multiple unknown time delays

For many years, various methods for the identification of parameters of continuous-time models have been available and implemented in widely. However, most methods apply models where the output are contaminated by a white noise or without noise in some others cases, which are unrealistic in most practical applications owing to their associated noise structure. Some other methods neglect the presence of time delays. Then it can be shown that the estimates are not statistically efficient. To cope with this issue, this paper deals with the identification of multi-input single-output continuous-time hybrid ‘Box-Jenkins’ systems having multiple unknown time delays from sampled input/output data. The proposed work presents a based-instrumental variable method for the separable estimation of both process parameters, multiple unknown time delays and the noise model. The effectiveness of the proposed scheme is proven through a numerical example illustrated by Monte Carlo analysis.