Use of Stochastic Algorithms to Solve Optimization Problems

In recent times, stochastic methods have become frequently used for solving different types of optimization problems [4.54—4.59]. If we consider here, for a steady state process analysis, the optimization problem given schematically in Fig. 4.14, we can wonder where the place of stochastic methods is in such a process. The answer to this question is limited to each particular case where we identify a normal type distribution for a fraction or for all the independent variables of the process pc = pCj]). When we use a stochastic algorithm to solve an optimization problem, we note that stochastic involvement can be considered in [4.59]  

The success of this computation method depends strongly on the dimension of the computation field which is considered here with the values of s ax i max-Indeed, when the values of i ax s ax greater than 2 10 and 10 respectively, using this method can be problematic because of the size of the computation volume. It is important to notice that this method works without the preparations considered in the gradient optimizing procedures (see Section 3.5.5). 

This procedure can easily be transformed to identify the parameters of a process as is shown in Fig. 4.15. 

---