Neural Network Research And Its Application In Supply Chain Inventory Control

Along with the intensifying market competition, the supply chain management thinking is being widely used, and the inventory management is an important part of the supply chain. The inventory control is a key factor than it can reduce inventory cost and improve customer satisfaction and the valid inventory management can bring huge economic value. As the development of information technique, the application of intelligence technique in the supply chain management becomes a trend. So this technology for supply chain inventory control is a hotspot of current research.The artificial neural network is one of computer intelligence technologies. The Back Propagation neural network is one of the most popular neural networks currently. But there are inevitably problems of the back propagation neural network that is easy to get into partial least extremum, slowly constringency speed, long training time and so on. With the question that Back Propagation algorithm convergence rate is slow, this article proposed one kind of tendency auto-adapted adjustment study advanced of parameter version algorithm, during the learning process, the parameters and the erroneous restraining threshold were dynamically adjusted, the improved algorithm was applied to the XOR problem, The experiments proved that the algorithm can automatically adjust parameter values, and can reduce turbulence and accelerate convergence speed. With the question that the Back Propagation algorithm is easy to get into partial least extremum, this article proposed to use genetic algorithm to optimize the initial weights. Genetic algorithm is a stochastic optimization technique, which can find the global optimal solution. So combined genetic algorithm with BP neural network will facilitate the realization of complementary advantages, and it can ensure overall optimization characteristics of genetic algorithms as well as it can speed up the convergence rate.The paper selected parts of a car company——bearing as the research object. The paper analysis the factors affecting the parts inventory and designed three-layer BP neural network model. A new self-adaptable genetic algorithm was put forward to improve genetic operators and compared the fitness value between two generations, then it selected the appropriate crossover rate and mutation rate to ensure the excellent individual into the next generation, at the same time, it avoided the situation that the largest population fitness value of individual cross-rate and mutation rate were zero. Lastly, combined the improved algorithm with BP network, and established inventory control model, and carried on the simulation experiment under the Matlab environment. The experiments proved that the error of the text result obtained from the model is within the scope of the requirements. The improvement self-adaptation genetic algorithm avoids sinking into partial minimum and it enhances the convergence rate of the network, and it improves the network study performance. Therefore, improved algorithm proposed in this paper is effective and has a certain practical significance.