Research On Fault Diagnosis Of Batch Process Based On Fuzzy Clustering

Intermittent process with non-linear,time-varying,strong coupling characteristics,it is difficult to accurately establish the process monitoring model.However,the data-driven monitoring model has better accuracy and adaptability by processing production process data and analyzing implicit information to detect,analyze,evaluate and handle the running status of complex processes.In this paper,the typical batch fermentation process of beer fermentation is taken as the object.Based on the theory of cluster analysis,the fault diagnosis algorithm of batch process is studied.The performance of the proposed algorithm is verified by simulation experiments.The main research contents and conclusions are as follows:Aiming at the problem that the clustering algorithm based on fuzzy C-means clustering is ineffective in clustering non-spherical or ellipsoidal datasets,a kernel function is proposed in FCM clustering algorithm to map the input data to high-dimensional feature space,amplifying differences between the samples and improving the algorithm for the purpose of classification accuracy of nonspherical or ellipsoid data set.Aiming at the problem that the FCM algorithm is sensitive to the initial value and easy to fall into the local optimal value,this paper proposes an intelligent optimization algorithm particle swarm optimization(PSO)algorithm to find the global optimal solution to the clustering center,to overcome the FCM algorithm for the initial value of random selection and easy to fall into the local optimal problem.In the form of simulation experiments,the fault diagnosis algorithm of beer fermentation process based on PSO-KFCM was verified.By comparing the experimental results of KFCM and FCM,it is concluded that the diagnostic accuracy of PSO-KFCM based intermittent process fault diagnosis algorithm is significantly improved.