Masakazu Shirakawa

Multi-objective Model Predictive Control

For many real-world problems, true function form cannot be given a prior. Consequently, for example, engineering design requires experiments and/or numerical simulations to evaluate objective and constraint functions as function in terms of design variables. However, those experiments/simulations are computationally expensive. For alleviating this burden of function evaluations, model predictive optimization (or surrogate modelbased optimization depending on literatures) methods have been used extensively in recent years. As a result, constructing good surrogate model with as few function evaluations as possible is essential to finding an optimal solution for problems. This research considers model predictive optimization problems under a dynamic environment with multiple objectives. Some techniques using machine learning such as support vector regression or radial basis function networks are applied to the generation of surrogate model. Although they are effective for model prediction, their prediction abilities may become worse due to a long prediction period. In order to develop accurate and stable prediction in model predictive optimization under a dynamic environment with multiple objectives, we propose computational intelligence methods with predetermined model, and investigate its effectiveness through numerical examples.

INTELLIGENT START-UP SCHEDULE OPTIMIZATION SYSTEM FOR A THERMAL POWER PLANT

Abstract This paper proposes an intelligent start-up schedule optimization system for a thermal power plant. This system consists of a dynamic simulation, a neural network, and an interactive multi-objective programming technique. The features of this system are as follows. (1) The start-up schedule can be optimized based on multi-objective evaluation and (2) the optimal start-up schedule can be determined with a reasonable computing time and calculation accuracy through interaction between human beings and computers.

Start-Up Schedule Optimizing System of a Combined Cycle Power Plant

Abstract This paper presents a method to determine the optimal operational parameters of a combined cycle power plant. The proposed method combines a dynamic simulation with an optimization calculation based on the nonlinear programming method. The reason for the optimization of the plant start-up scheduling is being able to reduce the start-up time by keeping the thermal stresses in the thick parts of the heat recovery steam generator and in the steam turbine under their allowable values.

Multi-objective Model Predictive Control Using Computational Intelligence

When function forms in mathematical models can not be given explicitly in terms of design variables, the values of functions are usually given by numerical/ real experiments. Since those experiments are often expensive, it is important to develop techniques for finding a solution with as less number of experiments as possible. To this end, the model predictive optimization methods aim to find an optimal solution in parallel with predicting the function forms in mathematical models. Successive approximate optimization or metamodeling are of the same terminology. So far, several kinds of methods have been developed for this purpose. Among them, response surface method, design of experiments, Kriging method, active learning methods and methods using computational intelligence are well known. However, the subject of those methods is mainly static optimization. For dynamic optimization problems, the model predictive control has been developed along a similar idea to the above. This chapter discusses multi-objective model predictive control problems and proposes a method using computational intelligence such as support vector regression.