Frequently, the engineer in charge of a process is faced with optimization problems. In fact, this may cover relatively different ideas, such as parameter identification or process optimization. It is known that the reactive feed flow rate profile for a fed-batch reactor and the temperature or pressure profile to be followed for a batch reactor, will have an influence on the yield, the selectivity or the product quality. To optimize production, one must then seek a time profile and perform a dynamic optimization with respect to the manipulated variables, while respecting the constraints of the system such as the bounds on temperature and temperature rise rate, the constraints related to the possible runaway of the reactor. Similarly ,to optimize the conversion in a tubular reactor, one can seek the optimal temperature profile along the reactor. In the latter case, it is a spatial optimization very close to the dynamic optimization where the time is replaced by the abscissa along the reactor. The profile thus determined is calculated in open loop and will be applied as the set point in closed loop, which may lead to deviations between the effective result and the desired result. The direct closed-loop calculation of the profile in the nonlinear case is not studied here; on the contrary, the linear case is treated in linear quadratic control and Gaussian linear quadratic control. In a continuous process, problems of dynamic optimization can also be considered with respect to the process changes from the nominal regime. For