GOALI/IUCP: Prediction of Wood Pulp K-Number with Neural Networks

小行星9522981 木质素将木材纤维结合在一起，必须去除木质素以生产用于牛皮纸的高强度纸浆。 卡伯值或K值表示制浆过程中木质素的去除程度，并且可能是该过程中测量质量的关键变量。 困难在于它是离线测量。 更重要的是，在原料进入制浆蒸煮器和测量K值之间通常有四个小时的过程延迟。 这使得建模和控制变得困难。 这个学术联络与工业项目的赠款机会使用神经网络模型来预测K数作为一个更容易获得的过程参数的函数。 这是改善该过程对木片进料变化的控制和响应的第一步。 来自缅因州大学和南达科他州的研究小组。Warren公司将使用来自运营工厂的数据开发表征和预测模型，并在集成到蒸煮器操作中时比较其长期预测能力。 在整个过程中，研讨会和讲习班是技术转让和模型改进的一部分。 这项研究的影响将是更均匀的纸浆质量，甚至与可变的原料，并在随后的漂白和造纸过程中更均匀的质量。

9522981 Musavi Lignin holds wood fibers together, and must be removed to produce high strength pulp for kraft paper. The Kappa- or K-number indicates the degree of lignin removal by a pulping process, and is probably the key variable for measuring quality in this process. A difficulty is that it is an off-line measurement. More importantly, there is usually a four hour process delay between when raw materials enter a pulping digester and when the K-number is measured. This makes modeling and control difficult. This Grant Opportunity for Academic Liaison with Industry project uses neural network models to predict K-number as a function of a number of more readily available process parameters. This is a first step in improving the control and responsiveness of this process to changes in chip feed stock. The research team from the University of Maine and S.D. Warren Company will develop characterization and prediction models using data from an operating plant, and compare their long term predicative capability when integrated into digester operations. Throughout, seminar and workshops are part of the technology transfer and model improvement. The impact of this research will be more uniform quality of pulp, even with variable feed stock, and more uniform quality in subsequent bleaching and papermaking processes.