Neural Network Rotor Angle Offset Compensation for Motor Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motor control systems face inaccuracies and instability due to deviations in sensed or estimated rotor angle values from true rotor angle values, leading to suboptimal torque output and potential loss of motor control, particularly at higher speeds, which current methods like look-up tables or non-linear functions fail to accurately compensate for.
Innovation Solution
Implementing neural network circuitry trained to generate a rotor angle offset based on instant rotor speed, minimizing d-axis instant voltage values across various rotor speeds, thereby providing a more accurate and adaptive compensation for angle errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods like look-up tables or non-linear functions are used to estimate rotor angle offset, then the system is simpler to implement, but the accuracy of rotor angle compensation deteriorates leading to suboptimal torque output and potential loss of motor control
Solution Approach 1:
The patent replaces traditional mechanical/mathematical estimation methods (look-up tables, non-linear functions) with neural network circuitry that performs adaptive learning and compensation. The neural network is trained to minimize d-axis instant voltage values, enabling accurate rotor angle offset compensation without relying on pre-defined mathematical models, thus resolving the contradiction between accuracy and complexity.
Solution Approach 2:
The patent changes the approach from using fixed mathematical parameters (look-up tables, non-linear functions) to using trainable neural network parameters (weights and biases) that adapt to minimize d-axis voltage. This parameter transformation enables the system to achieve high accuracy while maintaining practical implementability through learned optimization rather than complex analytical models.
2Reliability
If accurate rotor angle compensation is implemented using neural network circuitry, then motor control stability is improved, but the device complexity increases due to additional neural network components
Solution Approach 1:
The patent replaces complex control algorithms and mathematical computations with neural network circuitry that performs compensation through learned patterns. This substitution maintains motor control stability and reliability while the hardware implementation of the neural network keeps the overall device complexity manageable compared to software-based complex control systems.
3Productivity
If traditional estimation methods are used, then the device complexity is lower, but the torque output and motor performance deteriorate at higher speeds due to inaccurate angle compensation
Solution Approach 1:
The patent replaces traditional estimation methods with neural network circuitry that dynamically compensates for rotor angle deviations. This enables accurate torque control at higher speeds where traditional methods fail, as the neural network learns to minimize d-axis voltage across the full speed range, improving productivity without requiring excessively complex control systems.
Solution Approach 2:
The patent introduces dynamic adaptation through neural network learning, allowing the compensation strategy to change based on operating conditions including speed. The neural network is trained to handle varying speeds and operational conditions, enabling the system to maintain optimal torque output dynamically rather than relying on static estimation methods that degrade at higher speeds.
Data Source
AI summary
An apparatus for driving a motor includes motor circuitry and neural network circuitry. The motor circuitry is configured to generate, based on an error compensated rotor angle and current at a plurality of phases of the motor, a d-axis instant current value and generate a d-axis instant voltage value based on the d-axis instant current value. The motor circuitry is further configured to generate voltage at the plurality of phases based on the d-axis instant voltage value. The neural network circuitry is configured to generate a rotor angle offset based on an instant rotor speed at the motor. The neural network circuitry has been trained to generate the rotor angle offset to minimize the d-axis instant voltage value for each of a plurality of rotor speeds at the motor. The error compensated rotor angle is based on the rotor angle offset.