Neural Network Vector Control for Induction Motor Decoupling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional vector control strategies for three-phase induction motors face challenges in achieving true decoupled torque and flux control due to competing control nature in the current loop, making them sensitive to uncertainties and external disturbances.

Innovation Solution

Implementing a neural network-based vector control method that substitutes proportional-integral (PI) controllers in the current loop with a trained neural network controller using the Levenberg-Marquardt algorithm and Forward Accumulation Through Time (FATT) for dynamic programming-based training, allowing for optimal vector control of induction motors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional PI controllers are used in the current loop of vector control, then the control structure is simple and easy to implement, but true decoupled torque and flux control cannot be achieved due to competing control nature

Engineering Contradiction:

Improveease of implementationVSAvoiddecoupling control performance

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent replaces the conventional PI controller (a simple proportional-integral control mechanism) with a neural network-based controller that incorporates dynamic programming. This substitution transforms the control approach from a fixed-gain linear controller to an adaptive intelligent controller capable of achieving true decoupled control of torque and flux in the induction motor current loop.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent employs dynamic programming to optimize the control parameters of the neural network controller. By using value iteration and policy iteration methods, the controller adapts its parameters dynamically to achieve optimal decoupling performance, allowing the system to handle parameter variations and uncertainties in the induction motor.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If neural network-based control is implemented to achieve true decoupled control, then control performance and robustness are improved, but computational burden increases

Engineering Contradiction:

Improvecontrol robustnessVSAvoidcomputational complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent pre-trains the neural network controller offline using dynamic programming algorithms (value iteration and policy iteration) to obtain optimal control policies. This preliminary action allows the controller to be prepared in advance with optimized parameters, reducing the computational burden during real-time operation while maintaining high control performance and robustness.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The neural network controller is designed to adapt and learn from operating conditions automatically. Through online adjustment mechanisms, the controller can self-optimize its parameters based on real-time feedback, reducing the need for complex external computation and adjustment systems.

Inventive Principle:

Principle #25Self-service

3Stability of the object's composition

If conventional vector control is used, then the control system is stable, but it is sensitive to uncertainties and external load disturbances

Engineering Contradiction:

Improvecontrol stabilityVSAvoidrobustness to disturbances

Core Design Contradiction:

Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The patent implements a closed-loop feedback mechanism where the neural network controller continuously monitors the actual torque and flux and adjusts its control actions accordingly. The dynamic programming framework incorporates feedback from system states to optimize control decisions, enabling the system to maintain stability while adapting to uncertainties and external disturbances.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent transforms the static PI controller into a dynamic neural network-based controller that can adapt its control strategy in real-time. The controller's parameters and structure can dynamically adjust based on operating conditions, allowing it to maintain stability under varying loads and disturbances while achieving true decoupled control.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10990066B2Systems, methods and devices for vector control of induction machines using artificial neural networks

Publication Date: 2021.04.27 UNIVERSITY OF ALABAMA

AI summary

Described herein is a neural network-based vector control method for the induction motor. The disclosure includes an approach to implement optimal vector control for an induction motor by using an NN; a NN controller to substitute two decoupled proportional-integral (PI) controllers in current loop; and, a mechanism to train the NN controller by using a Levenberg-Marquardt (LM)+forward accumulation through time (FATT) algorithm.