Inverter AC Current Sampling for Six-Step Harmonic Rejection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In six-step inverter operations for electric vehicles, unwanted (6n−1) and (6n+1) harmonics appear in the measured phase currents, leading to aliasing effects and inaccuracies in alternating current and torque estimation due to insufficient sampling frequency.
Innovation Solution
The method involves sampling multiple values of AC current per subperiod during the six-step operation, transforming these values into a rotating reference frame, averaging them to suppress higher harmonics, and estimating the current value from the averaged samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only one current value is sampled per PWM period (6 values per fundamental period), then the sampling process is simple and fast, but the sampling frequency (6*f0) is insufficient to properly reconstruct frequencies above 3*f0, causing aliasing effects and harmonic distortion
Solution Approach 1:
The fundamental period is divided into six subperiods (PWM periods), and within each subperiod, multiple current values are sampled at different time instants. This segmentation approach increases the total number of samples per fundamental period from 6 to 18 (3 samples per subperiod), thereby increasing the effective sampling frequency to 18*f0 which satisfies the Nyquist criterion for reconstructing frequencies up to 9*f0 and eliminates aliasing of (6n±1) harmonics.
Solution Approach 2:
The sampling is performed periodically at three specific time instants within each PWM period. This periodic sampling pattern ensures that the (6n±1) harmonics are sampled at points where their contribution cancels out when averaged, while the fundamental frequency component is properly captured. The periodic nature of the sampling synchronized with the PWM period enables effective harmonic rejection.
2Measurement precision
If multiple current values are sampled and averaged per subperiod (18 values per fundamental period), then the aliasing effects and harmonic distortion are eliminated, but the computational complexity and processing time increase
Solution Approach 1:
The current values are transformed into the rotating reference frame before the averaging operation. This preliminary transformation aligns the fundamental frequency component with the direct axis of the rotating frame, making it appear as a DC component that can be easily extracted by simple averaging. The (6n±1) harmonics, after transformation, exhibit periodic patterns that cancel out during averaging, eliminating the need for complex filtering algorithms.
3Reliability
If (6n−1) and (6n+1) harmonics are present in the measured phase currents, then the current measurement includes all frequency components, but these harmonics alias to the fundamental frequency in the rotating coordinate system, causing torque estimation inaccuracies
Solution Approach 1:
The invention exploits the specific frequency characteristics of the (6n±1) harmonics in six-step operation by sampling at three specific time instants within each PWM period. At these carefully chosen instants, the (6n±1) harmonics produce values that sum to zero when averaged, effectively converting the harmful harmonic content into a neutral contribution. This allows the use of simple averaging to reject harmonics that would otherwise require complex filtering.
Data Source
AI summary
A method of AC acquisition for an electric machine, includes: a step in which an inverter is operated according to a six-step operation, during which a fundamental period of 2π is divided into six equal time periods, maximum one switching action occurring during each; a step in which a plurality of values of an AC of the inverter from a first reference frame are sampled for each equal time period; a step in which the sampled values are rotated from the first reference frame to a second reference frame; a step in which for each equal time period, the values from the second reference frame are averaged; and a step in which a current value is estimated according to the averaged sampled values.