Servo Motor Controller Coefficient Learning for Vibration Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimizing the coefficients of filters and velocity feedforward units in motor controllers to suppress vibrations and trajectory errors is challenging, especially when determining multiple parameters such as attenuation coefficients, central frequencies, and bandwidths, and adjusting these coefficients during changes in rotation velocity.
Innovation Solution
A machine learning device that performs machine learning to optimize the coefficients of filters and velocity feedforward units based on measurement information from an external instrument, position commands, and position errors, using an evaluation function to minimize vibrations and trajectory errors, allowing the external instrument to be removed post-learning to reduce costs and improve reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple coefficients (attenuation coefficients, central frequency, bandwidth) of filter and velocity feedforward unit are manually optimized, then vibration suppression and trajectory accuracy improve, but the complexity and time required for parameter tuning increase significantly
Solution Approach 1:
The system performs self-learning by automatically optimizing filter and velocity feedforward unit coefficients through machine learning algorithms. The control device learns optimal parameters by processing measurement information and position errors without requiring manual intervention, enabling the system to self-adjust and eliminate vibrations while maintaining trajectory accuracy.
Solution Approach 2:
The system performs preliminary learning operations by collecting measurement information during acceleration and deceleration phases, then uses this data to pre-optimize coefficients before actual operation. This preliminary action allows the system to prepare optimal parameters in advance, reducing the complexity of real-time parameter tuning during motor operation.
2Measurement precision
If external measuring instruments are used for coefficient optimization, then learning accuracy improves, but system cost increases and reliability decreases due to additional failure points
Solution Approach 1:
The invention extracts and utilizes measurement information that is already available within the motor control system itself, such as position errors from encoders and current measurements from motors. By taking out and leveraging these internal measurement sources, the system eliminates the need for external measuring instruments, thereby maintaining learning accuracy while improving reliability and reducing costs.
Solution Approach 2:
The system makes existing components serve multiple functions: position sensors originally designed for feedback control are also utilized for machine learning operations. This multi-functionality allows the system to perform both control and learning tasks using the same measurement infrastructure, eliminating the need for dedicated external measuring instruments.
3Ease of manufacture
If traditional methods are used to determine filter and feedforward coefficients, then implementation simplicity is maintained, but vibration suppression effectiveness and trajectory accuracy deteriorate
Solution Approach 1:
The invention replaces traditional manual tuning methods with automated machine learning algorithms. Instead of relying on operator expertise and iterative manual adjustment, the system uses computational algorithms to automatically determine optimal coefficients, thereby maintaining implementation simplicity while dramatically improving vibration suppression effectiveness and trajectory accuracy.
Data Source
AI summary
Vibration of a machine end and an error of a moving trajectory are suppressed. A machine learning device performs machine learning of optimizing first coefficients of a filter provided in a motor controller that controls a motor and second coefficients of a velocity feedforward unit of a servo control unit provided in the motor controller on the basis of an evaluation function which is a function of measurement information after acceleration and deceleration by an external measuring instrument provided outside the motor controller, a position command input to the motor controller, and a position error which is a difference between the position command value and feedback position detection value from a detector of the servo control unit.