论文部分内容阅读
针对一类串联型工业大系统,提出了多阶段逆模型建模方法:将串联大系统分为若干个阶段,以产品质量指标作为过程设计的起点,用逆向推理的方法,建立各个阶段的逆模型;根据产品质量指标的要求,直接求出各个阶段的控制变量设定值。将该方法应用于胶液生成过程的软测量建模,采用多阶段建模方法和整体建模方法分别建立了基于BP神经网络的胶液生成过程逆模型,并从误差平方和MSE和命中率等方面对两种建模方法的建模精度进行了比较。结果表明,多阶段建模方法可以获得更高的建模精度;同时,具有更大的灵活性;而且逆模型方法可以根据质量指标求出控制变量设定值,更便于实际应用。
Aiming at a series of industrial systems of tandem type, a multi-stage inverse model modeling method is proposed: the tandem system is divided into several stages, the product quality index is taken as the starting point of the process design, and the reverse inference method is used to establish the inverse of each stage Model; according to the requirements of product quality indicators, direct determination of the control variables in all stages of the set value. The method is applied to the soft sensor modeling in the process of glue formation. The inverse model of the glue production process based on the BP neural network is established by the multi-stage modeling method and the overall modeling method respectively. From the square error MSE and the hit ratio And other aspects of the modeling accuracy of the modeling methods were compared. The results show that the multi-stage modeling method can achieve higher modeling accuracy and more flexibility. Moreover, the inverse model method can obtain the control variable set value according to the quality index, which is more convenient for practical application.