Research On Data Mining Algorithm Based On Early Warning Of Cotton Storage Quality

For many years in the past, the cotton is treated as an important quartermaster materials and textile materials. Cotton should be stored after harvest until it be used. However, due to the environmental factors affecting during the storage process, the quality of cotton would affecting would lead to a decline. So the research on early-warning system based on cotton quality can not only avoid quality declining after storage by rational management of cotton storage environmental conditions, but also predict the quality of cotton after long-term storage. It has important meaning to our country agriculture and military. At the same time, with the development of agricultural informationization management, data related to agriculture has become more and more, early-warning based on cotton storage must become one of the important research direction of cotton storage informationization. In view of the research on cotton storage quality is near blank, applied the mature data mining technology to the research on early-warning system based on cotton storage quality can determine the warning factor and establish the early warning model correctly, objectively and comprehensively. And it can provide the correct information to control the storage environment effectively, to reach the goal of slowing down the decline in the quality of cotton in the stored procedure.According to the data in the cotton storage warehouse, this paper mainly using data mining technology to complete the preprocessing of classification of the cotton warehousing environmental data, mining the predictors of cotton storage quality factors, building the early-warning model of cotton storage quality and other related work. It can ultimately control the cotton storage environment by this early-warning system. It can also provide warning to those cotton warehouses who do not meet environmental requirements of the cotton storage, so that it can avoid a serious decline in quality of cotton.The main contents of this paper include:1. As the BIRCH algorithm for non-spherical clusters cannot accurate clustering, this paper combining with the data characteristics of the cotton warehouse environment factor and proposed an improved algorithm. The improved algorithm using representative thoughts instead of BIRCH algorithm to radius or diameter distance to control the clustering and this improved algorithm based on representative points, which can be called BURE algorithm. The optimized algorithm can be applied to non-spherical cluster in order to ensure the accuracy.2.This paper, according to the FP growth algorithm in FP-tree constructed lookup prefix sharing characteristics, combined with the practical application of cotton warehousing quality early warning, proposes improved New_FP-growth algorithm. The new algorithm reduces the FP tree*construction time and improve efficiency. The more data and transaction repeat rate is high, the more efficiency is improved. it is especially suitable for shared prefix or repeated transaction database.3. Data preprocessing is carried out by using the improved BURE algorithm in this paper. The cotton storage environment data in outlier removal after discretization and clustering. The converted data is suitable for this experiment. The Data after cleaning and transformation can greatly improves the efficiency of association analysis of processed data.4. By using the improved New_FP-growth algorithm for mining association rules in the text, It can use the discrete data clustering algorithm treated as a predictor of the establishment of early warning system. In this experiment, the FP growth algorithm is the basic algorithm, because of the cotton warehousing environment of large amount of data, frequent itemsets public more etc. Then use the improved algorithm to obtain early warning factor storage quality of cotton.5. Be excavated by the improved FP algorithm rules as predictors of growth. It can construct the cotton storage quality early-warning model, mine the warning information and then early warning the storage that may not be suitable for cotton library environment, in order to ensure the cotton warehousing environment is suitable.