基于核极限学习机的煮糖结晶自适应控制研究

Research of Adaptive Control for Sucrose Crystallization Based on Kernel Extreme Learning Machine

关键词:  煮糖结晶  自适应控制  工艺曲线  核极限学习机  粒子群算法

Key words:  sucrose crystallization  adaptive control  process curve  kernel extreme learning machine  particle swarm optimization

摘要:

针对煮糖结晶过程难以进行自动控制的问题，提出一种基于预测模型的自适应控制方法。以逐步浓缩上升煮糖工艺为基础，基于核极限学习机构建糖膏液位和糖膏锤度的预测模型；以预测工艺偏差作为适应度函数，利用粒子群算法在线优化蒸汽阀和入料阀开度，并自动调节阀门用于跟踪理想工艺曲线。结果表明：与人工煮糖相比，自适应控制的煮糖过程更稳定且更接近理想工艺曲线，达到卸糖液位和锤度所需的时间相对减少7.06%。基于核极限学习机的煮糖结晶自适应控制方法具有可行性，可为进一步实现工业煮糖自动控制提供理论参考。

Abstract:

In order to solve a difficulty of automatic control of sucrose crystallization process, an adaptive control method based on a predictive model was proposed. Basis on the gradual concentration and rise method of sucrose crystallization process, the prediction model of massecuite level and brix was established by kernel extreme learning machine. Taking the predictive process deviation as a fitness function, the particle swarm optimization algorithm is used to optimize the opening of steam valve and feed valve online and automatically adjust the valve to track the ideal process curve. The results showed that compared with a manual sucrose crystallization process, the process of adaptive control was more stable and closer to the ideal process curve, and the time to reach the unloading massecuite level and brix was reduced by 7.06%. The adaptive control method of sucrose crystallization based on kernel extreme learning machine is feasible, which can provide a theoretical reference for further realization of the automatic control of industrial sucrose crystallization.