Random uncertain motor parameters identification combining fourth-order moment and trust region

Area of Science:

• Engineering
• Computational Science

Background:

• Motor parameter identification faces challenges with low efficiency and ill-conditioned data due to uncertainty propagation.
• Existing methods struggle with the accuracy and computational cost of surrogate models in iterative identification processes.

Purpose of the Study:

• To develop a novel method for efficient and accurate random uncertain motor parameter identification.
• To reduce the dependence on surrogate model accuracy and improve computational efficiency in identifying motor parameters.

Main Methods:

• Combines the fourth-order moment method for probability calculation with trust region model management for optimization.
• Inner layer calculates cumulative probability and probability density function; outer layer transforms identification into a deterministic optimization problem.
• Utilizes genetic intelligent technology to further reduce computational cost and refines parameter search intervals iteratively.

Main Results:

• Successfully transforms random uncertain motor parameter identification into a deterministic optimization problem.
• Demonstrates effective identification of motor parameters by minimizing probability distribution discrepancies.
• Obtains the probability distribution of random uncertain motor parameters, including mean, standard deviation, skewness, and kurtosis coefficients.

Conclusions:

• The proposed method effectively achieves random uncertain motor parameter identification.
• The integration of fourth-order moment and trust region technology enhances identification accuracy and computational efficiency.
• Numerical results validate the method's capability in complex uncertain environments.