![]() | ![]() | ![]() |
| | |||||||
| Electronic Projects Design/Ideas/Reviews Are you building an electronic project or want to? Maybe you need some assistance? Come and submit your electronic questions here and let our experienced members find a solution. |
| | LinkBack | Thread Tools | Display Modes |
| | (permalink) |
| Hi, I'd like to know how do you control a system with a long delay in it's response. E.g.: I have to control the flow in a duct. I can regulate analogicaly the opening of a floodgate, but the measurement point (from where I get the feedback signal) is 500 mts downstream. So, initially I will be measuring 0 m3/s. If I use a PID control, my output will grow and grow opening the gate more and more. By the time the water reaches the feedback point the flow will be much bigger than I need it to be. Other example could be controlling a temperature where the heater element is away from the termometer. And I guess there must be a lot other systems like this. Can anyone tell me how do you control a system like this ? Thanks a lot. | |
| |
| | (permalink) |
| I found it. It's called the "Smith Predictor" Thanks anyway | |
| |
| | (permalink) |
| This isn't a simple problem but there are some common approaches. You might decide to adjust the gate (or valve) to a predetermined position that will result in the desired flowrate at such a great distance downstream. A "positioner" does that on a valve - the controller output indicates an absolute position. Another way is similar but not the same - measure the flow rate at the floodgate and have a controller set the position of the floodgate to maintain the setpoint (the flow rate at the controller). When the flowrate downstream needs to increase you reset the flow rate controller setpoint to a new value that corresponds to what you want. There are other, more elaborate schemes but these are two improvements over what you've described. In either case described above the response won't be quick but you'll avoid the severe overshoot that is likely to result - the gate would be wide open long before the water got to the downstream point - unless the system were incredibly slow. Slowing the response down is a valid approach but may not get you what you want.
__________________ stevez | |
| |