When a measurement used in a control application is characterized by a significant amount of process or measurement noise then to avoid taking control action on noise filtering may be applied in the transmitter or in the control system. However, when traditional filtering techniques are used it is necessary to increase the PID reset time to compensate for the added lag introduced into the control loop. As a result a slower control response will be observed for changes introduced for by unmeasured disturbances. At Emerson Exchange last week I co-hosted a workshop with Willy Wojsznis which addressed the benefits of using Kalman filtering vs traditional filtering techniques when a control measurement contains significant measurement or process noise.
You can see and listen to this presentation using the embedded link shown below. The presentation is approximately 32 minutes in length.
- 0-5:20 background information on the Kalman filter.
- 5:20-13:15 scalar Kalman filter implementation
- 13:15-16:45 modifying the Kalman filter for non-zero mean, implementation in DeltaV
- 16:45-19:42 test results of Kalman filter vs traditional filter in control
- 19:42-24:55 accessing DeltaV composite for Kalman filter, reference information
- 24:55-31:55 questions and answers
The composite we created for Kalman filtering can be found through application exchange at the DeltaV Interactive Portal. This composite may be downloaded and imported on any DeltaV system. Also, a test module and documentation on the Kalman filter are available at this site.
I will be posting a series of blogs over the next month on four other workshops I co-hosted at Emerson Exchange. If you did not attend Emerson Exchange or missed these workshops at Emerson Exchange then I hope you find these postings of interest.