Flexible Management Research Of Metallurgical Production Supply Chain Based On Artificial Neural Network

Steel industry occupies an important position in national economic development. Traditional steel enterprises have been responded slowly to the changes in market demand, the allocation of resources is low, the high cost of maintaining the supply chain and other short comings. Therefore, more accurate forecasts of the demand for raw materials of metallurgical products can make steel companies capture the market trends early and provide the basis for co-ordinating arrangements for the production. Iron ore is the main raw material for the steel industry; its demand forecast is a variety of factors and a complex nonlinear problem. Now, at home and abroad, there are many demand forecasting methods, such as qualitative analysis, comparison adjustment method, and expert adjustment method. But the approaches above have larger prediction errors, more subjective elements and other defects. So, this article chose BP neural network to establish forecast model.There are many factors affect the demand for iron ore, crude steel production there, steel at current prices, fixed asset investment, consumer spending, domestic iron ore production, iron ore price rate of change of international agreements, GDP(Gross Domestic Product), national import the amount of iron ore, etc.To select the main factors from the many factors as the input vector for the predictive model, thereby reducing the complexity of the model, this paper introduces principal component analysis. With three-layer network topology, the model use test method to determine the number of network layers and nodes. And Quasi-Newton algorithm was used to improve traditional BP neural network. So that the convergence speed of BP network while accelerating the convergence precision is improved. At the same time, combining genetic algorithms with BP neural network, using genetic algorithm’s global search to optimize BP network initial weights, effectively overcomes the deficiencies and defects of BP algorithm, and Iron ore demand forecast model was established.Experimental results show that this method achieves a better prediction; it can make the steel enterprises timely response to market changes and customer demand, adjust production and sales plans, Avoid or share the associated inventory risk. To achieve flexible management purposes.