الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 198 |  |  |
| | | from | 198 | | |
Abstract
Due to the steady growth of load demand in power systems at a rate of 7%per annul which means that power consumed may be doubled every ten years, new generation sites and perhapsa new voltage level are to be considered.
The computer-aided methods of visualizing new circuits in a network context are neededu using the different utilities in transmission system planning. This problem is solved before using many traditional techniques,heuristic programming techniques and mathematical optimization techniques.
In this thesis, the utilized traditional tequniques are heuristic tequnique and the mathematical programming optimization tecnique, which are represented in the transportaion technique and the de linear programming optimization technique. In traditional technique, the system was planned by minimizing the capital and operating costs of the system taking into account the associated constraints with the applied model,
Artificial neural network technique is proposed as a new and an accureate method to solve the planning problem as the potential benefits of nural networks extend beyond the high computation rate and because of the computation burden and consuming time when utilizing the traditional techniques. An artificial neural netwoek with back-propagation training is utilized with two different techniques, a Fortran program algorithm and the Matlab programming technique.
These techniques are applied and trained on a set of input/target patterns that obtained from the dc linear programming technique using the Fortran program. The input of the neural network is the set of input patterns to the dc linear programming technique and the target of the neural network is the set of output patterns resulted from the dc linear programming technique.
The techniques are applied again and trained on this set of input/target patterna using the Matlab program.
Also, an application of the artificial neural network with back-probagation training was utilize using the Fortran program on a set of input/target patterns that pbtained by the dc linear programming with the sensitivity matrix technique in order to plan a reliable system. The input of the neural network is the same set of input patterns of the dc linear programming technique, the target of the linear programming with the sensitivity matrix technique.
These techniques are applied on a six-bus system with twenty-one total lines data are indicated in the systm configuration.
Satisfactory results are obtained. Comparison betwen rsults obtained from the traditional anf the suggestedn techniques is made.