The automotive industry has been dramatically developed these years. However, the whole process of automotive production chain is directly affected by the accuracy of its production forecasting model, such as safety inventory quantity, out of stock losses, and timely performance. Therefore, to improve the accuracy of production forecasting, this paper uses the Chinese automotive industry as a case study, which has been the largest in the world measured by automobile unit production since 2008. It presents three kinds of combined models based on grey neural network, which are parallel grey neural network, series grey neural network, and inlaid grey neural network, compared the single model GM(1,1) and neural network. The experimental results demonstrate that the combined models are higher forecasting accuracy than the single model.