A study of using back-propagation neural network in joint procurement

In the era with low profit margin, the procurement cost is regarded as a key factor that influences the business operations of small and medium enterprises (SMEs). Joint procurement is a feasible method that would reduce the procurement cost effectively. This study is intended to use the back-propagation neural network and take the maximization of profit as the objective function in the situation of joint procurement. This study first trained the back-propagation neural network by the history data of enterprise's production and sales, in order to figure out the optimal weight value and bias of the back-propagation neural network for the case company to estimate the procurement quantity during different periods. It then validated the accuracy and reliability of this neural network based on late production and sales data of the enterprise, as well as the optimal weight value and bias of neural network, in order to design an optimal back-propagation neural network algorithm of procurement cycle and quantity. Finally, it simulated the business scenario and substituted the predicted procurement quantity of different cycles of all enterprises into the objective function of maximum profit, in order to find out the optimal procurement cycle and procurement quantity. Therefore, this study confirmed that the joint procurement is an effective method for reducing the procurement cost on the theoretical basis of relevant researches. The results of this study provide a new strategy to help SMEs finding their preferred cycle of joint procurement and procurement quantity, so as to reduce the procurement cost and improve their competitiveness through the joint procurement.