RedOnion

PID.1

Derived: PID

PID(R) regulator (with extra parameters). Used to overcome certain problems of direct control. P-only regulator is pass-through (doing nothing), but other features like ouput-change-limiting can help smooth the control signal. But that alone could stop regulating near the target, when there is some minimal threshold for action to be taken. Rotation controll (roll, killRot) with P-only regulator could settle with small but non-zero offset, keeping the vessel/ship rotating by small amount indefinitely. PI-regulator overcomes this by accumulating the error and increasing the control signal until effect can be observed (rotation speed / angular velocity changes). But that has its own problem - induced oscillation (and wind-up). Two other factors - R and D - can be used to overcome that oscillation. D directly lowers output signal when effects are observed (decrease raw-roll control signal when observing change in angular velocity) and R lowers the accumulator used by I to dampen the oscillation (reducing both oscillation/overshooting and wind-up caused by big changes in input signal). Both also react to outside disturbances (like drag).

Instance Properties:

• param: Params - All the parameters.
• P: double - Proportional factor (strength of direct control).
• I: double - Integral factor (dynamic error-correction, causes oscillation as side-effect).
• D: double - Derivative factor (dampening - applied to output, reduces the oscillation).
• R: double - Reduction factor for accumulator (dampening - applied to accumulator used by integral factor, works well against both oscillation and windup).
• scale: double - Difference scaling factor.
• targetChangeLimit: double - Maximal abs(target - previous target) per second. NaN or +Inf means no limit (which is default). This can make the output smoother (more human-like control) and help prevent oscillation after target change (windup).
• outputChangeLimit: double - Maximal abs(output-input) and also abs(target-input) for integral and reduction factors. Helps preventing overshooting especially after change of Target (windup). NaN or +Inf means no limit (which is default).
• errorLimit: double - Limit of abs(target-input) used by P and I factors. Prevents over-reactions and also reducing windup.
• accumulatorLimit: double - Limit of abs(accumulator) used by I and R factors. Another anti-windup measure to prevent overshooting.
• input: double - Feedback (true state - e.g. current pitch; error/difference if Target is NaN).
• maxInput: double - Highest input allowed.
• minInput: double - Lowest input allowed.
• target: double - Desired state (set point - e.g. desired/wanted pitch; NaN for pure error/difference mode, which is the default). The computed control signal is added to Input if Target is valid, use error/difference mode if you want to add it to Target.
• maxTarget: double - Highest target allowed.
• minTarget: double - Lowest target allowed.
• output: double - Last computed output value (control signal, call Update() after changing Input/Target).
• maxOutput: double - Highest output allowed.
• minOutput: double - Lowest output allowed.
• dt: double - Time elapsed since last update (in seconds), Time.tick after reset.
• lastDt: double - Time elapsed between last and previous update.

Instance Methods:

• reset(): void - Reset internal state of the regulator (won’t change PIDR and limits).
• resetAccu(): void - Reset accumulator to zero.
• update(): double - Update output according to time elapsed (and Input and Target).
• update(): double, dt double, input double
  • Set input and update output according to time elapsed (provided as dt).
• Update(): double, dt double, input double, target double
  • Set input and target and update output according to time elapsed (provided as dt).
• update(): double, dt double
  • Update output according to time elapsed (provided as dt, using current Input and Target).