Inferential Sensor

Inferential sensors, or "soft sensors," estimate difficult-to-measure process variables using readily available data, aiming to improve accuracy and efficiency in industrial settings. Current research focuses on enhancing the predictive power of these sensors by employing multi-model architectures, often incorporating linear models combined with sophisticated switching logic, such as Support Vector Machines (SVMs), to seamlessly transition between models and optimize data partitioning. This approach addresses limitations of simpler single-model sensors, particularly in handling nonlinear processes, leading to improved performance in applications like chemical engineering and petrochemical refining.

Papers