The Research On Artificial Bee Colony Algorithm For Fuzzy Clustering Of Data Mining

With the progress and development of the socioeconomic level, the automation and the intelligence of industry production turns the degree higher. The improvement of automatization degree made enterprises to amass and store more and more historical data of industrial processes. On the one hand,with the enterprises to improve their own development. These data is containing the rich extremely valuable information and the experience. On the other hand, production practice and scientific research of industrial processes are based on a large number of historical data. These use some methods to analyses and process, in order to perform various tasks. For example: online monitoring, process identification, fault diagnosis and design of control strategy. Therefore, data mining as one of the main method of techniques to locate the useful information in the vast amount of data causes the attention of people more and more.Artificial Bee Colony algorithm simulates the intelligent foraging behavior of honey bee swarms. It is a very simple, robust and population based stochastic optimization algorithm. So, it has been paid more and more attention by scholars. Although FCM and KFCM have been applied to pattern recognition, image processing and computer vision and many other fields, but there are still some flaws. Aiming at the shortcoming of the FCM algorithm, that is easily plunging into local minimum, and sensitivity to initialization and noise data. Therefore, a new artificial bee colony (ABC)-based fuzzy algorithm (ABFM) is put forward in this paper. KFCM not only to certain extent overcomes limitation of data intrinsic shape dependence and can correctly clustering, but also overcome sensitivity to initialization and noise data and improve the algorithm robustness. However, like FCM algorithm, KFCM algorithm still exists some drawbacks, such as the sensitivity to initialization and the tendency to get trapped in local minima. Therefore, an improved kernel fuzzy C-Means based on artificial bee colony (ABC-KFCM) is proposed.ABFM and ABC-KFCM Improve the search efficiency and reduce the search process of the phenomenon of local optimum. However, when in higher dimensions and larger numbers of clustering the effect is not obvious. Therefore, we apply Boltzmann selection mechanism instead of roulette and uses min intervals to make the initial group more symmetrical and better capacity of global search. According to the test, It not only can effectively solve faults of FCM and KFCM but also has more accurate in clustering and higher efficiency for the data set of higher dimensions and larger numbers of clusteringTE is a coupling, parameter varying, nonlinear, multi-variables process.The traditional dimension reduction methods are not obtain the desired effect. Therefore, we apply manifold learning to reduce the dimension of data, and then use ABFM algorithm to Cluster analysis. The simulation results show that it is feasible for the improved algorithm is the feasibility,and achieve the desired results.