Recontruction of the following papers on the topic of steady state detection / identification methods:
[1] J. D. Kelly and J. D. Hedengren, “A steady-state detection (SSD) algorithm to detect non-stationary drifts in processes,” J. Process Control, vol. 23, no. 3, pp. 326–331, 2013.
A closer probability to 1 means the more likely it is in a steady state condition.
[2] R. R. Rhinehart, “Automated steady and transient state identification in noisy processes,” in Proceedings of the American Control Conference, 2013, no. June 2013, pp. 4477–4493.
The figure below is without data windowing applied, each data point is processed immediately. It does not look so good. Perhaps, it needs more tuning for the lambda values.
The figure below is with data windowing applied. It looks nice ;-)
For this method, the lower R-value means the more likely it is in a steady state condition.
The plant that is used:
Continously stirred tank reactor (CSTR):