Statistical algorithms designed to provide accurate estimates of process variables even in the presence of extreme outliers or faulty sensor data. They ensure that plant optimization software remains stable and reliable despite occasional measurement noise or instrument failures.