Decreasing the moisture content variation of flakes in the drying process of Oriented Strand Board (OSB) manufacturing using multivariate time-series pattern discovery

减少干燥过程中薄片含水率的变化对于降低定向刨花板（OSB）的制造成本至关重要。对于一个典型规模的工厂来说，1%的湿度标准偏差降低值超过100万美元/年的制造成本。OSB的制造过程包括将原木切割成薄片，干燥薄片，加入胶水和蜡，将薄片形成均匀的垫子，然后将垫子热压成大约4英尺乘8英尺的面板。干燥过程必须克服各种挑战，如原料原木的水分含量变化很大，从不同操作特性的多个干燥机中混合薄片，干燥机内10分钟的时间延迟以及干燥机与放置湿度传感器的垫成型站之间的几个小时的时间延迟等。工厂维护着一个大型数据库，其中记录了数百个高频率（每分钟或更少）的操作变量。因此，多元时间序列模式发现有可能导航这个庞大的数据库，并确定最佳的操作策略和条件，以最大限度地减少薄片湿度的变化。

Decreasing the variation in moisture content of flakes from the drying process is important for cutting the**manufacturing cost of Oriented Strand Board (OSB). A 1% reduction in the standard deviation of the moisture**worths more than $1 million manufacturing cost/year for a typical sized mill. The manufacturing process of**OSB involves cutting logs into flakes, drying the flakes, adding glue and wax, forming the flakes into a**uniform mat and then hot pressing the mat into panels approximately 4 foot by 8 feet. The drying process must**overcome challenges such as the highly variable moisture content of the raw logs, the mixing of flakes from**multiple dryers with different operating characteristics, the time delays of 10 minutes within the dryers and the**time delays of several hours between the dryers and the mat forming station where the moisture sensors are**located, etc. The mill maintains a large database with hundreds of operating variables recorded at high**frequency (every minute or less). Multivariate time-series pattern discovery thus has the potential to navigate**this large database and identify the optimum operating strategy and conditions that minimize the flake moisture**variability.