Implementation of multi-Kalman filter to detect runaway situations and recover control

This work should concentrate on developing a tool able to detect runaway conditions in a batch reactor and act on the reactor to recover a safer condition. Parallel Kalman filters are implemented, corresponding to some possible models that potentially can describe the system (i.e. parallel or consecutive reactions). Acting over each one of the filters, the Bayes' theorem allows determining the probability of success of each filter and, finally, estimating the divergence of the system. If the divergence were positive, a runaway situation is detected and prevention and protection procedures should be activated dangerous situations. Using the divergence criterion, a runaway situation is detected at very early stages, giving time enough to decide and apply the most conventient actions.