In order for biomass drying processes to be efficient, it is crucial to achieve the target residual water content within a close margin, since more conservative drying would result in a waste of energy. A method for a reliable estimation of the water content is therefore of obvious importance. Ideally, such a method does not require any expensive sensors. We show reduced order models and extended Kalman filters can be combined to reliably determine the water content and temperature of wood particles based on only surface temperature measurements. The proposed observer works reliably if measurements are only available for parts of a particle face. It can therefore still be applied if particle surfaces are partially obstructed, which is a prerequisite for use in industrial processes and units, such as rotary dryers. The extended Kalman filter uses a reduced order model that is obtained by applying proper orthogonal decomposition and Galerkin projection to coupled PDEs that model heat conduction and water diffusion in anisotropic particles. In contrast to the original PDE simulation model, the reduced model and the filter based on it are suitable for real time computations and monitoring.

为了使生物质干燥过程高效，至关重要的是要在很小的范围内达到目标残留水含量，因为更保守的干燥会浪费能源。因此，一种用于可靠地估计水含量的方法非常重要。理想地，这种方法不需要任何昂贵的传感器。我们展示了降阶模型，可以结合使用扩展的卡尔曼滤波器来仅基于表面温度测量值可靠地确定木材颗粒的水含量和温度。如果仅对粒子面的一部分进行测量，建议的观察者将可靠地工作。因此，如果颗粒表面被部分阻塞，仍然可以使用，这是在工业过程和单元（例如旋转干燥机）中使用的先决条件。扩展的卡尔曼滤波器使用降阶模型，该模型是通过对耦合的PDE应用适当的正交分解和Galerkin投影获得的，该模型对各向异性颗粒中的导热和水扩散进行建模。与原始的PDE仿真模型相比，精简模型和基于它的滤波器适用于实时计算和监视。