Memetic Algorithm for Induction Motor State Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the state and controlling induction motors face challenges in determining and updating system parameters, particularly noise covariance matrices, online during operation, due to changing operational environments and unknown statistical noise characteristics.
Innovation Solution
The implementation of a memetic algorithm that combines evolutionary and local search methods to construct a cost function based on system data, allowing for the online estimation and adjustment of system parameters, including noise covariance matrices, for improved state estimation and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional state estimation methods (Kalman filter, extended Kalman filter) are used for induction motor control, then the control system can operate with a relatively simple structure, but the accuracy of state estimation deteriorates when operational environments change and noise characteristics are unknown
Solution Approach 1:
The patent combines state estimation and parameter estimation into a unified framework. The memetic algorithm integrates evolutionary computation and local search methods to simultaneously estimate system states and unknown parameters (including noise covariance matrices), allowing the system to adapt to changing operational conditions without requiring separate estimation modules.
Solution Approach 2:
The patent dynamically adjusts noise covariance matrices and other system parameters online based on observed data. The memetic algorithm continuously optimizes parameter values to minimize a cost function, enabling the estimator to adapt to varying noise characteristics and operational environments rather than relying on fixed parameter values.
2Adaptability or versatility
If offline parameter determination from experimental data is used, then the system can operate with simple structure, but the adaptability to changing operational environments deteriorates
Solution Approach 1:
The estimation system performs self-adjustment by automatically determining and updating its own parameters online. The memetic algorithm enables the system to self-optimize noise covariance matrices and other parameters based on real-time observed data without requiring external recalibration or manual intervention, allowing continuous adaptation to changing conditions.
Solution Approach 2:
The patent implements a feedback mechanism where the cost function evaluates the difference between predicted and actual measurements, and this information is used to update parameter estimates. The iterative optimization process continuously refines parameter values based on feedback from system performance, enabling adaptation to changing operational environments.
3Loss of time
If experimentation is conducted to determine system parameters, then initial parameter values can be obtained, but the time and cost required deteriorates
Solution Approach 1:
The patent employs preliminary rough estimates of parameters to initialize the estimation algorithm, then rapidly refines these estimates online using the memetic algorithm. This approach avoids the need for extensive offline experimentation while still achieving accurate parameter values through quick online optimization based on initial operational data.
Data Source
AI summary
This document presents methods by which system parameters are tracked online. For adjusting the values of the system parameters, the presented methods take advantage of the data that become available during the system's operation. The values of the parameters can be used for estimating the system state and controlling the system. This document includes methods comprising constructing a cost function based on the data collected during the system operation as a function system parameters and running a memetic algorithm that reduces the cost function value. A memetic algorithm combines an evolutionary or a population-based algorithm with a procedure for local improvement (local search, refinements) of candidate solutions. This document includes methods by which the rotor speed, the rotor flux, and the stator currents of the induction motor can be estimated from electrical measurements while covariance matrices of noise are adjusted online.