Design Of Product Forecasting System Architecture In Brand Apparel Planning System

In recent years, the production scale of many domestic apparel companies has been expanding, but the competitiveness compared to international brands is still insufficient. Most apparel companies rely on low price to interest customers. These companies are difficult to survive. For apparel such a fashion product, how to grasp the trend of market is the most important thing for apparel companies, and that can improve their competitiveness. Many international apparel companies have deployed a variety of system to help designers and business decision-makers to make strategy.This paper is based on a 211 cross-disciplinary project. Analyzes the purpose of the apparel planning system, and expounds the design premises and targets. This paper is focus on the collection of data in planning system and establishment of data warehouse. We proposed a set of feasible, high-performance and low-cost implementation plan.The author's main work is summarized as below.1) Designing a system to collect the sale data from different distributors, in order to solve the data heterogeneous problems from different distributors, such as:platforms are not unified, products are inconsistent and so on.2) Building the data warehouse system of apparel planning forecast system through extracting the sales data and enterprise information data. This data warehouse system contains different comprehensive level data warehouse, to adapt to different prediction task that has different data size.3) Analyzing the data warehouse products which we used in apparel planning forecasting system and summarizing several effective optimization strategy on the data warehouse.4) Proposed a forecast model based on Gray relational analysis and BP neural network for apparel production parameters. Through the Gray relational analysis help BP neural network select main factors, makes the forecast model has a better forecast result.Through the experiment in the small size clusters, the system proposed by this paper is Feasible. It meets the basic target of the apparel planning forecast system.