Research And Application Of Intelligent Pig-keeping Based On Association Rules

After nearly30years development, with the promoted and application in large-scalebreeding technology, the pig industry in China is also get some improvement.But, incurrent situation, our raising technology is still lower than the international level, largescale pig industry is faced some problems like lack of talents、low efficiency and high cost,to be resolved.Therefore, in recent years, many large pig farms began to adopt theautomatic intelligent equipment to improve the efficiency of pig-keeping and thecompetitiveness in the industry. In the whole process of intelligent pig-keeping, whetherfor the pigs’ data and information collection, or to analyzed these information to findindividual physiology characteristic, production performance and health to get precisefeeding programs, we all need to have a comprehensive analyse and research in a largenumber of breeding datas, as to help farm staff for a right judgment.And for these records,with multiple attribuete,it is very important for us to find the relationship quickly andmining the precious experiment between them.Data Mining is one of the most active fields, and usees to find out the implict valuableknowledge from massive data.Association Rules is used to be foundd interesting dependentrelationships between items of database, and it is a main branch in the field.But the mainproblems of this method are how to improve the efficiency of algorithms and how to applyit in a certain application area.This paper focus on the research of association rules algorithm in the field ofintelligence pig-keeping.Aim at the characteristic of the application field,design theinproved algorithm based on category and increamental updating,and the application in thisfield.The main problems and related results of the research are as follows:For the algorithm research, this paper first analyzes two typical existing algorithms,Apriori and FP-Growth.On the background of the project of Chongqing about large-scalepig-keeping, according to the request of farmers and the characteristic of datas, in order tosolve these problems, the association rules model is researched in the following twoaspects:(1)For breeding needs of the users, proposed the conception of classificationproperties and rule ordred tree, the algorithm named CACC is designed;(2)According tothe characteristic of the pig farm in data updated, proposed the conception of classified andbasic solution, the algorithm named IUAC is designed.Finally, these two methods are eachcampared with FP-Growth and Apriori on the standard data sets to prove that these arefaster and feasible. For application, use the large amounts of data from automatic feeding systrm, after thepretreatment of the data, we use CACC and IUAC to have experiments to miningAssociation Rules and to realize the model of the best breeding schenme, for the referenceto breeding users.Through the analysis of the results, it is showed that the rules matcheswith Pigs Technology Theory, and also have a good feeding reference value.