| Power distribution network is the last part of electric power system for power system. It is like a bridge connecting electric power system and users. So safe and stable operation of power distribution network has a direct influence on numbers of users. Scientific programme and investment in power distribution network is the foundation of power distribution network’s development. And the programme’s level and project’s quality is related to the safety reliability and economical efficiency of power supply.Post-evaluation means analysis and evaluation on the purpose, implementation, effectiveness and effect of completed project.To find out the reason for success or failure and draw lessons from the past by evaluation. By building up a timely and effective Information feedback system, to give some advices to improve investment decision-making level and to increase investment returns. So post-evaluation of distribution network project is necessary. But the existing evaluation has some shortages and it is short of objectivity and operability.With a strong ability for mode recognition and data-fitting, Artificial neural systems is be used in the areas of mode recognition, cluster, regression for fitting and optimizing calculation and so on. With good ability of self-study and self-adaption, neural network can adjust autologous properties to environment. When the environment has changed, network can adjust structure parameters automatically and change the mapping conditions to produce corresponding desired output to specific inputs. In consequence, Comparing to expert system in fixed reasoning, artificial neural network has a better adaptability and is closer to the movement of the human brain.To improve the traditional post-evaluation of distribution network, this text summited to use artificial neural network to post-evaluation of distribution network project for network modeling, then to train it with samples selected by experts. With experiences of experts, decision procedure of evaluation will do better in science and handleability. Improving the quality and speed of evaluation, it has a good application prospect. |