2006 Fiscal Year Final Research Report Summary
Research on the model analysis technique in energy system planning for city area
| Project/Area Number |
16560352
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
System engineering
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
SAEKI Osamu Osaka University, Graduate School of Engineering, Assistant Professor, 大学院工学研究科, 助手 (20252596)
|
|---|
| Project Period (FY) |
2004 – 2006
|
| Keywords | Optimization / Energy systems / Model analysis / Interior point methods |
|---|
| Research Abstract |
The main results that were obtained by this research are follows :
1. Study of the algorithm
It extended to the technique of having applied the interior point method to the branch-and-cut method. It was shown that it is possible to reduce the number of relaxed problems generated by the proposal algorithm. On the other hand, the target energy system optimization problem has the special structure. It accelerates with the block iterative method using the structure. It had turned out that numerical instability occurs by the conventional research. The improvement technique was proposed, and it was confirmed that the computation time was reduced. Moreover, the relation between the shape of the optimal plane and the number of generated subproblems was shown by numerical simulation.
2. Application to large-scale problems
In recent years, advance of grid computing technology is remarkable. It is considered that parallelization is effective for further improvement in the speed of large-scale problems. Therefore, a study about parallelization on an electric power and energy system problem that has the block structure was carried out. The technique using multiple computers based on a proposal algorithm was proposed, and it was shown that the proposal technique is effective. Moreover, it was confirmed that the reduction of computational time by parallelization became large as the number of blocks increased.
3. Study of modeling
There is essentially flexibility in the formulization for the same problem. Even if it originally formulizes the variable that should be an integer as a continuous variable, it is shown that the solution that satisfies the original constraints may be obtained. And the relationship between the definition of variables and the computational time was shown.
|
