Back to results
US 11,796,977 B2Granted
PID controller autotuner using machine learning approaches
AssigneeHoneywell International Inc.
InventorsVladimir Havlena, Daniel Pachner
Life of this patent
Estimating the term…
Abstract
An approach for auto tuning a PID controller that may incorporate determining set-points for controlled variables, obtaining controlled variables from a controlled process module, taking a difference between the set-points and the controlled variable as inputs to a proportional, integral and derivative (PID) controller, calculating a loss or reward from the difference between the controlled variables and the set-points, and achieving set-point tracking. The loss or reward needs to be an increasing or decreasing function of a control error value. Also incorporated may be adding loss or reward components based on controlled variables time difference or control action time difference, which may effect a self-tuned performance of the PID controller.
Claims17
A proportional, integral, and derivative (PID) controller system comprising:
- PID controller configured to output a control action to a controlled process module, the control action is based on an output provided to the PID controller and is configured to affect a value of one or more controlled variables;
- a machine learning module implemented by one or more processors having a reinforcement module that provides the output to the PID controller via an output connection to the PID controller;
- and
- a controlled process module configured to receive, as an input, the control action outputted from the PID controller, the controlled process module provides values of controlled variables to the PID controller via a differentiator that also takes in set-points for the controlled variables and the differentiator outputs a control error to the PID controller, wherein the control error is a difference between the set-points and the values of the controlled variables;
- and
- wherein:
- the values of controlled variables and set-points for the controlled variables are input to the machine learning module;
- and
- the machine learning module calculates the instantaneous loss or reward which is an increasing or decreasing, respectively, function of the control error;
- the machine learning module determines the time-differentiated and augmented variable process states based on process states outputted from the controlled process module, the set-points for the controlled variables, and the values of the controlled variables;
- the reinforcement module is configured to receive the instantaneous loss or reward, the time differences of control actions, and the time-differentiated and augmented variable process states and determine the output provided to the PID controllers based on the instantaneous loss or reward, the time differences of control actions, and the time-differentiated and augmented variable process states.
8 more elements
A method for auto tuning a proportional, integral, and derivative (PID) controller comprising:
- determining set-points for controlled variables;
- obtaining controlled variables from a controlled process module;
- taking a difference between the set-points and the controlled variable as inputs to the PID controller, the differences between the set-points and the controlled values are controlled error values;
- calculating a loss or reward from the difference between the controlled variables and the set-points;
- achieving set-point tracking, wherein the loss or reward is an increasing or decreasing function of the control error value;
- determining control action time differences;
- providing an output to the PID controller, wherein the output is based on the control action time differences, the time-differentiated and augmented variable process states, and the loss or reward calculated;
- and
- passing an output of a control action to the controlled process module from the PID controller, wherein the control action is based on the output provided to the PID and is configured to affect a value of one or more of the controlled variables.
6 more elements
An auto tuner system comprising:
- a machine learning module implemented by one or more processors;
- a proportional, integral and derivative (PID) controller having an input connected to the machine learning module and configured to determine a control action based on an output received from the machine learning module, the control action is configured to affect a value of one or more controlled variables;
- a controlled process module having an input connected to the PID controller and configured to receive the control action determined;
- and
- a differentiator implemented by one or more processors having an input connected to the controlled process module and an output connected to the PID controller;
- and
- wherein the machine learning module comprises:
- a process states augmentation and differentiation submodule having a first input of values of controlled variables from the controlled process module, a second input of set-points for the controlled variables, and a third input of process states from the controlled process module, and an output to a reinforcement learning submodule of time-differentiated and augmented variable process states determined based on the first input, the second input, and the third input;
- and
- the reinforcement learning submodule providing the output to the PID controller, the reinforcement learning submodule determines the output to the PID controller based on a calculated loss or reward, control action time differences, and the time-differentiated and augmented variable process states determined.
7 more elements
Dependencies are inferred from the claim text.