【摘要】 针对焙烧球团料层透气性软测量二次变量过多的状况,提出了对二次变量“主次分离”的处理方法,该法可降低当二次变量过多时软测量的建模难度。同时提出了一种“松散型模糊神经网络”的建模方法,该法充分利用了模糊技术和神经网络技术的各自优势,得到了较高的测量精度。更多还原

【Abstract】 Since the status of overmuch secondary variables in process of soft-sensing on diathermancy of pallet, a method of separating primary and secondary variables is put forward. In this way the model-establishment difficulty can be reduced when secondary variable is overmuch. Meanwhile a new model-establishment method of relax fuzzy neural network is used, each virtue of fuzzy technology and neural technology is imposed in this method, and high measure precise can be reached.更多还原