Data-driven approach for time-delay estimation of industrial processes

An accurate estimate of time delay is of crucial importance for control tasks such as performance assessment and controller design. In this paper, a novel data-driven approach to time-delay estimation is developed for a process subject to industrial background disturbances, in which the closed-loop output data under the routine operating conditions is only required. The practical solutions are proposed to estimate the time delay based on the estimated impulse response of the closed loop that is estimated online by utilizing the output data. Without relying on system identification and any prior knowledge of the process as well, the time delay is estimated directly for a large time delayed process; while for a small time delayed process, the time delay is estimated by means of the stationarilized filter, the pre-filter and the loop filter. The effectiveness of the proposed approach is validated by various numerical and industrial examples including a distillation column, a petroleum refinery heating furnace and a ceramic dryer.