Source code for pySimBlocks.blocks.controllers.pid

# ******************************************************************************
#                                  pySimBlocks
#                     Copyright (c) 2026 Université de Lille & INRIA
# ******************************************************************************
#  This program is free software: you can redistribute it and/or modify it
#  under the terms of the GNU Lesser General Public License as published by
#  the Free Software Foundation, either version 3 of the License, or (at your
#  option) any later version.
#
#  This program is distributed in the hope that it will be useful, but WITHOUT
#  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
#  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License
#  for more details.
#
#  You should have received a copy of the GNU Lesser General Public License
#  along with this program.  If not, see <https://www.gnu.org/licenses/>.
# ******************************************************************************
#  Authors: see Authors.txt
# ******************************************************************************

from __future__ import annotations

import warnings

import numpy as np
from numpy.typing import ArrayLike

from pySimBlocks.core.block import Block


[docs] class Pid(Block): """Discrete-time PID controller block. Implements a single-input single-output discrete PID controller, similar to the Simulink PID block. The controller computes a control command from an error signal ``e`` using proportional, integral, and/or derivative actions depending on the selected control mode. Output saturation is applied only if ``u_min`` and/or ``u_max`` are set. Anti-windup clamps the integrator state to the saturation bounds. Attributes: controller: Active control mode (``"P"``, ``"I"``, ``"PI"``, ``"PD"``, or ``"PID"``). integration_method: Integration scheme for the I term (``"euler forward"`` or ``"euler backward"``). Kp: Proportional gain as a (1,1) array. Ki: Integral gain as a (1,1) array. Kd: Derivative gain as a (1,1) array. u_min: Lower saturation bound as a (1,1) array, or None. u_max: Upper saturation bound as a (1,1) array, or None. """ def __init__( self, name: str, controller: str = "PID", Kp: ArrayLike = 0.0, Ki: ArrayLike = 0.0, Kd: ArrayLike = 0.0, u_min: ArrayLike | None = None, u_max: ArrayLike | None = None, integration_method: str = "euler forward", sample_time: float | None = None, ): """Initialize a PID controller block. Args: name: Unique identifier for this block instance. controller: Control mode. Must be one of ``{"P", "I", "PI", "PD", "PID"}``. Kp: Proportional gain. Must be scalar-like. Ki: Integral gain. Must be scalar-like. Kd: Derivative gain. Must be scalar-like. u_min: Minimum output saturation bound. None to disable. u_max: Maximum output saturation bound. None to disable. integration_method: Integration scheme for the I term. Must be ``"euler forward"`` or ``"euler backward"``. sample_time: Sampling period in seconds, or None to use the global simulation dt. Raises: ValueError: If ``controller`` or ``integration_method`` is invalid, if any gain is not scalar-like, or if ``u_min > u_max``. """ super().__init__(name, sample_time) controller = controller.upper() allowed_types = {"P", "I", "PI", "PD", "PID"} if controller not in allowed_types: raise ValueError( f"[{self.name}] Invalid controller type '{controller}'. Allowed: {allowed_types}" ) self.controller = controller self.integration_method = integration_method.lower() allowed = ("euler forward", "euler backward") if self.integration_method not in allowed: raise ValueError( f"[{self.name}] Unsupported method '{self.integration_method}'. Allowed: {allowed}" ) self.Kp = self._to_siso("Kp", Kp) self.Ki = self._to_siso("Ki", Ki) self.Kd = self._to_siso("Kd", Kd) self.u_min = None if u_min is None else self._to_siso("u_min", u_min) self.u_max = None if u_max is None else self._to_siso("u_max", u_max) if self.u_min is not None and self.u_max is not None: if float(self.u_min[0, 0]) > float(self.u_max[0, 0]): raise ValueError( f"[{self.name}] u_min ({self.u_min.item()}) must be <= u_max ({self.u_max.item()})." ) self._validate_gains() has_p = "P" in self.controller has_d = "D" in self.controller if has_p or has_d: self.direct_feedthrough = True else: # I-only self.direct_feedthrough = (self.integration_method == "euler backward") self.inputs["e"] = None self.outputs["u"] = None self.state["x_i"] = np.zeros((1, 1), dtype=float) self.state["e_prev"] = np.zeros((1, 1), dtype=float) self.next_state["x_i"] = np.zeros((1, 1), dtype=float) self.next_state["e_prev"] = np.zeros((1, 1), dtype=float) # -------------------------------------------------------------------------- # Public methods # --------------------------------------------------------------------------
[docs] def initialize(self, t0: float) -> None: """Set the output to zero and keep internal states at zero. Args: t0: Initial simulation time in seconds. """ self.outputs["u"] = np.zeros((1, 1), dtype=float)
[docs] def output_update(self, t: float, dt: float) -> None: """Compute the PID control command from the current error input. Args: t: Current simulation time in seconds. dt: Current time step in seconds. Raises: RuntimeError: If input ``e`` is not connected. """ e_in = self.inputs["e"] if e_in is None: raise RuntimeError(f"[{self.name}] Missing input 'e'.") e = self._to_siso("e", e_in) x_i = self.state["x_i"] e_prev = self.state["e_prev"] has_p = "P" in self.controller has_i = "I" in self.controller has_d = "D" in self.controller P = self.Kp * e if has_p else np.zeros((1, 1), dtype=float) if has_i: if self.integration_method == "euler forward": I = x_i else: I = x_i + self.Ki * e * dt else: I = np.zeros((1, 1), dtype=float) D = (self.Kd * (e - e_prev) / dt) if has_d else np.zeros((1, 1), dtype=float) u = P + I + D if self.u_min is not None: u = np.maximum(u, self.u_min) if self.u_max is not None: u = np.minimum(u, self.u_max) self.outputs["u"] = u
[docs] def state_update(self, t: float, dt: float) -> None: """Update the integrator state and store the previous error. Args: t: Current simulation time in seconds. dt: Current time step in seconds. Raises: RuntimeError: If input ``e`` is not connected. """ e_in = self.inputs["e"] if e_in is None: raise RuntimeError(f"[{self.name}] Missing input 'e'.") e = self._to_siso("e", e_in) has_i = "I" in self.controller if has_i: x_i_next = self.state["x_i"] + self.Ki * e * dt else: x_i_next = self.state["x_i"].copy() # Anti-windup: clamp integral state to saturation bounds if self.u_min is not None: x_i_next = np.maximum(x_i_next, self.u_min) if self.u_max is not None: x_i_next = np.minimum(x_i_next, self.u_max) self.next_state["x_i"] = x_i_next self.next_state["e_prev"] = e.copy()
# -------------------------------------------------------------------------- # Private methods # -------------------------------------------------------------------------- def _to_siso(self, name: str, value: ArrayLike) -> np.ndarray: """Normalize a scalar-like value to a (1,1) array; reject anything else.""" if np.isscalar(value): return np.array([[float(value)]], dtype=float) arr = np.asarray(value, dtype=float) if arr.shape == (): return np.array([[float(arr)]], dtype=float) if arr.shape == (1,): return arr.reshape(1, 1) if arr.shape == (1, 1): return arr raise ValueError( f"[{self.name}] '{name}' must be scalar-like ((), (1,), or (1,1)). Got shape {arr.shape}." ) def _validate_gains(self) -> None: """Warn if a gain is zero for a mode that requires it.""" kp = float(self.Kp[0, 0]) ki = float(self.Ki[0, 0]) kd = float(self.Kd[0, 0]) if "P" in self.controller and kp == 0.0: warnings.warn( f"[{self.name}] Kp=0 while controller '{self.controller}' includes a P term.", UserWarning, ) if "I" in self.controller and ki == 0.0: warnings.warn( f"[{self.name}] Ki=0 while controller '{self.controller}' includes an I term.", UserWarning, ) if "D" in self.controller and kd == 0.0: warnings.warn( f"[{self.name}] Kd=0 while controller '{self.controller}' includes a D term.", UserWarning, )