Soft Sensing Of The Lubricant Yield Based On ANFIS In Acetone Benzol Dewaxing Crystal Process

The acetone-benzol dewaxing process is an important part in the process of producing lubricant, and it costs about more than 60% of the cost, and this process also plays an important role in determining the quality and receiving rate of the lubricant. This paper researches on one acetone-benzol dewaxing device and creates the soft sensing model of it in an improved ant clustering algorithm and adaptive neural fuzzy inference system. The main contents are as follows.Soft sensing technique works very well in theory and application area since it is created in 1970s, this technique's base idea is to get the value of primary variables with secondary variables. This paper chooses the getting rate of lubricant as the primary variable and chooses those factors which affect the primary variable as the secondary variable, such as the cooling speed, the component of solvents and the ratio of solvents. Then the paper analyses the sample data with an improved ant clustering algorithm, and determine the number of cluster, the centers of clusters and the rules of cluster algorithm. Based on this, the paper proposes a fuzzy neural network based on this improved ant clustering algorithm in order to solve the problems that determine the fuzzy rules and the number of the rules. Finally, this paper put this soft sensing model based on ANFIS into the practice of acetone-benzol dewaxing crystal process, and the results express very well.