I want to control a process via a sliding mode controller. I do not have exact equations of the process, thus estimated equations are used. The actual output is in range of [0 °C , 500 °C], and estimation error is in [-3°C , 3°C]. The estimation is almost accurate but if I use SMC, there will be a high-frequency chattering due to estimation error even if I use Sat or Tanh instead of Sgn function. Is it logical to divide the output into a constant e.g. in order to put the output in the range of [0 , 10] and then control the process based on new output (mapped output)? Is there any other suggestion available for this problem?
Hi Mehdi Mousavi,
I don't understand what you meant by dividing the output signal of a plant into a constant. At first, you mentioned the range of plant output is [0, 500]°C. Then, you also stated that output in the range of [0 , 10]. Moreover, the system order and the behavior of the system are unknown. So, it can be mind-crackling to answer your problem.
Despite the info insufficiency, I think it is important to ensure that the designed SMC is robust against the modeling error. Yes, you heard me right, it's the modeling error, not the output error ±3°C. You don't need to switch to sat(s) or tanh(s) during the early stage of the control design.
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping