Uncertainty Research And Application Of Management Data In Characteristic Food Production

Food is the paramount necessity of the people. Our country government as focused on food production. But recently the management level of food production in China is not high. The enterprises of food production rarely apply analysis of data to improve production management. The management of production of characteristic food in the backward areas is especially poor and it has an adverse effect on food quality and livelihood of people in the backward area. In recent year, characteristic food industry develops rapidly in Guizhou province and trade of characteristic food explodes throughout the world. It generates the huge economic benefit and attracts the local government’s attention. It helps to upgrade local characteristic food enterprises and enhance their management. With the development of Wireless Sensor Network, Internet of Thing, Cloud Computing, these techniques would have large-scale application in the field of food industry and produce abundant production data. These data usually carry uncertainty. For example, the information system of production management in Laoganma is built on the wireless sensor network and has on-site data collection, management and controlling of the whole course of production. Due to low sensing accuracy, limited hardware resources and weak anti-interference of sensor nodes, uncertainty exists in the data collected from the wireless sensor network and it will result in economic losses because uncertainty seriously impacts accuracy of management and control of the system of characteristic food production. Therefore, the research on management and application of uncertain data in characteristic food production would have high theoretical value and practical significance.To solve above problems, the main work that is finished in this paper as follows:(1)Aiming at low accuracy of data measured in outlier detection of production monitoring and measuring system, an outlier detection algorithm based on interval data was proposed in uncertain data stream. In this algorithm, firstly, the interval data introduced was used to express the inaccuracy of measured signal. Then most of the normal data in the sliding windows was pruned according to the position between two interval data. At last, the data in the current window were reordered by their distance to the first data and it could narrow the query range of data points in k-close distance. Experiment results showed that the algorithm not only possessed better clustering precision with low computing complexity but also was well applied to outlier detection in the motor monitoring system.(2) Since efficiency and accuracy of mining frequent patterns was not high in uncertain data, this paper proposed an algorithm called UFPGA(uncertain frequent pattern genetic algorithm) for mining frequent itemsets of the probability. According to features of uncertain data, the algorithm improved FP-tree(frequent pattern tree) to mine frequent itemsets and employed the genetic algorithm, which reduced the variability of space and increased breeding operator, to search the largest frequent itemsets. UFPGA algorithm shrank the search scope to improve the efficiency of mining frequent itemsets. Results of experiments showed that UFPGA had a good advantage in the time complexity. The algorithm has a positive significance for large-scale uncertain data mining.(3) This paper studied on the production environment of characteristic food of Laoganma and deeply analysed the features of uncertain data in the production and management system. It introduced the uncertain data model to express the uncertain data in the information system of monitoring environment and employed two types of uncertain data processing technologies to deal with the data in the system.