PID

Proportional-Integral-Derivative Controller.

A PID controller predicts how much an actuator should be given a relevant feedback signal. A common example is in a water boiler where the PID controller predicts how much the heating element should be turned on given a feedback temeperature signal from a temperature sensor.

Dependencies

controlhal

PID

For most use-cases, the PID class takes in up to 3 inputs:

from pid import PID

controller = PID(0.05)  # A "P" controller
controller = PID(0.05, 0.001)  # A "PI" controller
controller = PID(0.05, 0.001, 3.2)  # A "PID" controller

By default, the output is limited to the floating-point range [0, 1], which should be interpreted as 0 to 100%. This output normalization makes it easier to hook up different actuators to a PID controller.

controller = PID(0.5, setpoint=80.0)  # Setpoint can be configured at initialization.
controller.setpoint = 80  # Can directly set the attribute whenever.

The PID controller should be called at a fix interval. If called faster than the specified period (defaults to 0.01 seconds), then the internal state will not be updated, and the previous actuator prediction value will be returned.

while True:  # Main application loop
    temperature = sensor.read()
    actuator_power = controller(temperature)
    actuator.write(actuator_power)

This sensor-read, actuator-write code is wrapped up in controlhal.ControlLoop.

Acknowledgements

This library is modified from m-lundberg's implementation. In turn, his library was inspired by Brett Beauregard's arduino library, PIDLibrary.

I added the following modifications:

Fix some micropython incompatibilities.
Don't accumulate the integral component while the acutator is in saturation. This helps prevent integral windup.
Change a few parameter names to be consistent with some other libraries.
Set default output limits to (0, 1). The error_map feature has been removed in favor of a normalized output.