Motor Parameter Estimation With Iterative MRAC and ANN Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for online estimation of motor parameters, such as phase resistance and permanent magnet flux linkage, fail to accurately account for non-modeled voltage losses during load and speed changes, leading to deviations in estimated values and errors in coil temperature estimation.
Innovation Solution
An equipment and method that utilize a combination of artificial neuronal network and Model-Reference-Adaptive-Control models to iteratively estimate motor parameters, considering non-modeled operating parameters, with initially determined constants as start values, to improve estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a MRAC model sets permanent magnet flux linkage to a constant value to estimate phase resistance, then the estimation process can be simplified, but non-modeled voltage losses during load and speed changes are not accounted for, causing deviations in estimated parameters
Solution Approach 1:
The patent transforms the static estimation approach into a dynamic iterative process. The first and second estimation algorithms initially assume constant parameters, but the third and fourth algorithms dynamically revise these estimates by incorporating non-modeled voltage losses. This dynamic revision process continues until convergence, allowing the system to adapt to changing operating conditions while maintaining model simplicity.
Solution Approach 2:
The patent implements a feedback mechanism where the output of the first and second estimation algorithms serves as input to the third and fourth revision algorithms. The revised estimates are then fed back into the system, creating an iterative loop that progressively improves accuracy by accounting for previously neglected voltage losses, thereby resolving the contradiction between model simplicity and estimation accuracy.
2Device complexity
If an ANN model sets phase resistance to a constant value to estimate permanent magnet flux linkage, then the estimation process can be simplified, but non-modeled voltage losses during load and speed changes are not accounted for, causing deviations in estimated parameters
Solution Approach 1:
The patent applies the same dynamic principle to the ANN-based estimation. While the initial ANN model assumes constant phase resistance, the iterative revision process dynamically adjusts this assumption by incorporating non-modeled voltage losses. This allows the system to maintain the simplicity of the ANN model while improving accuracy through dynamic parameter revision.
Solution Approach 2:
The feedback mechanism operates similarly for the ANN model, where the initial estimation and subsequent revision algorithms form an iterative loop. The revision algorithm uses the initial estimation results and operating parameters to compute corrected values, feeding them back into the system for further refinement, thereby resolving the accuracy-complexity contradiction.
3Productivity
If simultaneous estimation of permanent magnet flux linkage and phase resistance is attempted under reference control with Id=0 A, then both parameters can be estimated together, but rank deficiency makes the simultaneous estimation impossible
Solution Approach 1:
The patent segments the simultaneous estimation problem into separate sequential steps. Instead of attempting to estimate both parameters simultaneously (which causes rank deficiency), the system first estimates one parameter while treating the other as constant, then revises both estimates in subsequent iterations. This segmentation eliminates the rank deficiency issue while still achieving simultaneous parameter estimation through the iterative process.
Solution Approach 2:
The patent applies preliminary action by performing initial constant-parameter estimation before conducting the revision step. The first and second algorithms provide preliminary estimates that serve as a foundation for the subsequent revision algorithms. This preliminary action allows the system to bypass the rank deficiency problem in the initial phase while setting up the conditions for accurate final estimation.
Data Source
AI summary
A method and equipment (apparatus) for estimating motor parameters includes: receiving an operating parameter of an electric motor, estimating an estimated first motor parameter based on the operating parameter and on an initially determined second motor parameter and estimating an estimated second motor parameter based on the operating parameter and on an initially determined first motor parameter. The equipment (apparatus) and the method further include estimating a revised estimated second motor parameter based on the estimated first motor parameter and on the operating parameter, and estimating a revised estimated first motor parameter based on the estimated second motor parameter and on the operating parameter.