A number of approximate methods for optimal control synthesis are reviewed and tested on a highly non‐linear continuous‐stirred‐tank‐reactor example. It was found that parameterizing a feed‐back controller and searching for the optimal coefficients was computationally as fast as the standard conjugate gradient approach. Ancilliary advantages of this approach make it an attractive means of optimal control synthesis.

We consider the bilateral contract satisfaction problem arising from electrical power networks due to the proposed deregulation of the electric utility industry in the USA. Given a network § and a set (or multiset) of pairs of vertices (contracts) with associated demand functions that reflect the amount of flow which needs to be sent between the pairs of vertices, the goal is to find the maximum number of simultaneously satisfiable contracts. Four different algorithms for bilateral contract satisfaction problems are considered: (i) SMALLEST-FIRST HEURISTIC (ii) LARGEST-FIRST HEURISTIC (iii) RANDOM-ORDER HEURISTIC and (iv) ILP-RANDOMIZED ROUNDING: an integer linear programming based approximation algorithm with proven performance guarantee in restricted cases. The main focus of the paper is to study how the heuristics performed in fairly realistic settings. For this purpose we used an approximate electrical power network from Colorado.From our preliminary analysis, it appears that although the first three heuristic algorithms do not have a worst-case guarantee, they outperform the theoretically better ILP-RANDOMIZED ROUNDING algorithm. We also tested the algorithm on four types of scenarios that are likely to occur in a deregulated marketplace. The empirical results show that the networks that are adequate in regulated marketplace might not be adequate for satisfying all the bilateral contracts in a deregulated industry. Our studies indicate that simple clearing mechanisms that are currently in use in many of the power markets are computationally fast and near-optimal in their use of network capacity; also, a linear programming upper bound on the quality of any solution, turns out to be a very good bound in practice.Basic and Applied Simulation Sciences, (D-2), and Computer and Computational Science (CCS-3),

No abstract

No abstract