Motor Controller Inverse Model Compensation for Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motor control systems for electric power steering face issues with motor operation noise due to noise sensitivity in current detection and parameter variations, particularly when using high-pass filters in feedforward control, which increase torque command value noise and sensitivity.
Innovation Solution
A motor control system that employs feedforward control with compensation for self-inductance using an inverse model, correcting phase characteristics and gain reductions with angular velocity-based adjustments, and includes dead band compensation to reduce noise sensitivity and operation noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedback control is used for motor current control, then control accuracy is improved, but motor operation noise is generated due to noise in current detector
Solution Approach 1:
The control system is divided into feedforward control and feedback control paths. The feedforward control path uses inverse model compensation to handle the main control task without using noisy current detection values, while the feedback control path handles only the error correction. This segmentation allows the system to achieve control accuracy without being affected by current detector noise in the main control path.
Solution Approach 2:
An inverse model of the motor and drive circuit is introduced as an intermediary to compensate for parameter variations and characteristics. This inverse model allows the feedforward control to predict and compensate for system behavior without directly using noisy current detection values, thereby reducing motor operation noise while maintaining control accuracy.
2Stability of the object's composition
If high-pass filter is used in feedforward control to process actual currents, then parameter variation compensation is improved, but torque command value noise sensitivity is increased
Solution Approach 1:
The system changes the parameter representation by using an inverse model that accounts for parameter variations in the motor and drive circuit. Instead of directly filtering actual currents with high-pass filters, the inverse model compensates for parameter variations in a way that does not amplify noise, thereby reducing torque command value noise sensitivity while maintaining parameter variation compensation.
3Object-generated harmful factors
If current detection value is not directly used to suppress motor operation noise, then motor operation noise is reduced, but motor output torque varies due to parameter variations
Solution Approach 1:
The inverse model performs preliminary compensation for parameter variations and drive circuit characteristics before the control action is applied. By pre-compensating for these variations in the feedforward path, the system can avoid using noisy current detection values while still maintaining stable motor output torque through the compensated feedforward commands.
Solution Approach 2:
A feedback mechanism is introduced to compensate for the limitations of feedforward control. The feedback path uses current detection values only for error correction, not for the main control command, thereby suppressing motor operation noise while maintaining torque stability through the combination of inverse model feedforward compensation and feedback error correction.
Data Source
AI summary
A technique for reducing an operation noise even when a high-pass filter is used for controlling a motor is provided. A controller is used in a motor control system for driving a motor by using a drive circuit and an inverter. The controller includes a current control block. The controller performs feedforward control by using a current value, compensates for a term of a self-inductance of the motor included in a current control block by an inverse model, compensates for a phase characteristic of a transfer function of the inverse model by an advance component, and corrects a gain characteristic of the transfer function of the inverse model by a function of a physical quantity obtained based on an angular velocity of the motor, thereby compensating for a phase delay and a gain reduction of a torque output generated by the self-inductance.