Automated Gate Rejection Algorithms for Real Time Recycled Fiber Blend Variations
Real-time NIR spectroscopy coupled with automated divert gate algorithms stabilizes furnish quality, rejecting off-spec recycled stock before pulper entry.
Kalman Filtering
Recursive Estimation
Mathematical algorithms used for tracking dynamic systems provide a way to filter out noise from industrial sensor data in real time. Standard kalman filtering predicts the next state of a variable, such as paper thickness or roll tension, and then updates that prediction based on the next measurement. This logic allows a control system to ignore sudden, impossible spikes caused by electrical interference or dust on a lens.
Signal Processing
The calculation relies on a series of linear equations that weigh the uncertainty of the current model against the uncertainty of the incoming data. Within a paper mill, kalman filtering helps maintain a steady coating weight even when the base sheet varies slightly. It prevents the machine actuators from overreacting to tiny, random fluctuations that do not represent a real change in the product.
Output from the process is a smoother control response and less wear on mechanical parts.
Process Stability
Speed and accuracy make this method ideal for high-speed converting lines where decisions must happen in milliseconds. Because kalman filtering only needs the previous state and the current observation, it requires very little computer memory. It fails when the underlying process is extremely non-linear or when the noise in the sensor does not follow a normal distribution.