Research On Multiple Support Vector Machines Soft Sensor Modeling Based On Particle Bee Colony Optimization Algorithm

Soft sensor technology is a novel type of intelligent measurement technology. It has a very broad application prospects in the industrial process measurement and control system. In the soft sensor technology system, the theoretical study of the soft sensor modeling is one of the hot one. The traditional soft sensor modeling methods have some disadvantages. For example, narrow scope of use, low precision and so on. In the paper, for the soft sensor object with a variety of working conditions, a multiple support vector machines soft sensor modeling based on particle bee colony optimization algorithm is researched. And then, in order to prove the effectiveness of the method, the aluminum strip's grain size soft sensor model is established successfully in the process of aluminum electromagnetic casting.Firstly, the basic principle of soft sensor technology is introduced. Through the literature review, the research trend of soft sensor modeling theory is analyzed. At the same time, some major matters are indicated in the process of soft sensor modeling.Secondly, in order to solve the problem of data clustering, an improved fuzzy C-means clustering based on the subtractive clustering is researched. Two data sets are simulated and analyzed by using the algorithm. The results show that the improved algorithm has many advantages, including greater stability, less computation, and so on. Soon afterwards, the support vector regression modeling basic theory is elaborated. Based on the above research, a multiple support vector machines soft sensor modeling method based on fuzzy clustering is researched. The detailed model building process and the model application process is introduced.Then, in order to solve the support vector machines parameters optimization problem, a novel particle bee colony optimization algorithm is proposed. In the algorithm, the global guidance mechanism of PSO and the basic structure of artificial bee colony algorithm are combined organically. Through a large number of simulation experiments, some advantages are confirmed, including faster convergence speed, higher convergence rate, and so on.Finally, in order to solve the measurement problem of aluminum strip's grain size in the process of aluminum electromagnetic casting, the soft sensor technology is used. The multiple support vector machines soft sensor model of aluminum strip's grain size is established. The simulation results show that, compared with the traditional single model, the model has more robust and better prediction performance.