Research Of Extension's Utilization In Model's Self_learning

In this article, the author seeks after the model's intelligentizing management in DSS (Decision Support System) introducing the extension theory and extension engineering methods. DSS is developing towards intelligentizing direction, so is its kernel-model. The model's intelligentizing management is incorporated by the auto-establishment, the auto-selection and the self-learning of the model. This article brings forward a new kind of machine-learning method - extension machine method, hi this method, according to the evaluation of the model in the system utilizing extension evaluation method, the model piles up experience in the continually using and obtains the exact knowledge about decision, corrects its parameter and ameliorate the arithmetic of itself, hi this way, the model achieve the aim of perfecting its capability through self-learning.With the background of the developing of the DSS of 4810 factory in LvShun, the author unites the above theory and practice. Through detailed investigating, analysing and developing of the evaluation process of the project, the author set up the project's budgeting model, initialized the model's budgeting quotiety and fixed the quotiety as the model's auto-correcting quotiety. So we could evaluate the project according to the established model, fix the price to the bid, undertake the project, put the project after making the manufacture contract, evaluate the project in our factory and direct the man-hour tickets to every workshop. When the project is completed, we could make our conclusion and calculate the profit margin, fix the function of the profit margin with the factory's satisfaction and the client's satisfaction. And we could figure out the satisfaction of the two side towards this project, use the two sides' satisfaction as weighing condition and fix the power quotiety as the weighing condition utilizing extension evaluation method, calculate the optimized degree of the budgeting model. According to the result, we could set up the relation function of the corrected quantity of the budgeting model's parameter and the model's optimized degree, calculate the parameter's adjusted difference. Then the system could adopt man-machine intercourse method to carry out model self-learning, correct thebudgeting quotiety to piles up experience for the model and increase its optimized degree gradually. All this makes our budgeting process rational, and this improves our enterprise's competition ability.The 4810's DSS has already completed according to the above theory. And now it's running smoothly which validate the rightness and feasibility of the theory.