Multi-Winding Motor Predictive Flux Control Without Low-Pass Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current control methods for electrical converters with multiple three-phase windings face challenges in distinguishing fundamental current components from ripple, leading to suboptimal torque control and increased harmonic distortion, particularly in medium-voltage drives where low-pass filtering limits controller gain and allows for suboptimal optimized pulse patterns.
Innovation Solution
The implementation of model predictive pulse pattern control (MP3C) that controls instantaneous flux along a reference trajectory defined by the converter's pulse pattern, allowing for discontinuous switching angles and eliminating the need for low-pass filtering, while extending flux-based control to electrical machines with multiple three-phase windings by calculating average and difference flux references and translating them into winding flux references for each system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If low-pass filtering is applied to extract fundamental current component, then current ripple is reduced, but controller gain is limited and drive dynamic performance deteriorates
Solution Approach 1:
The invention extracts only the necessary information (fundamental current component) through appropriate sampling at specific time instants, rather than applying low-pass filtering to the entire signal. This allows obtaining the fundamental component without the performance degradation caused by filtering.
Solution Approach 2:
The sampling time instants are pre-determined based on the switching pattern to coincide with moments when only the fundamental current component is present at the converter terminals. This preliminary positioning of sampling points eliminates the need for filtering and preserves dynamic performance.
2Stability of the object's composition
If switching angles of OPPs evolve smoothly without discontinuities, then control stability is maintained, but OPP optimality deteriorates resulting in higher current ripple
Solution Approach 1:
The invention transitions from static, continuous switching angle evolution to dynamic switching angle selection that can change discontinuously between different modulation intervals. This allows the switching angles to be optimized independently in each interval based on current operating conditions, achieving better ripple performance.
Solution Approach 2:
The switching angles are optimized periodically at each modulation index interval, allowing discontinuous jumps between intervals. Each periodic optimization cycle independently determines the optimal switching angles for that specific operating range, improving overall performance.
3Productivity
If MP3C is applied to electrical machines with multiple sets of windings, then control performance is improved, but control complexity increases
Solution Approach 1:
The control of multiple winding systems is segmented into independent model predictive controllers for each winding, with each controller handling its own flux reference trajectory. This modular segmentation manages complexity by treating each winding independently while achieving coordinated control through shared flux references.
Solution Approach 2:
A universal flux-based control framework is applied that works for both single-winding and multi-winding machines. The same MP3C principles and algorithms are used across all winding systems, providing a unified multi-functional control approach that reduces overall system complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling an electrical converter (14) comprises: determining an average flux reference ψS,avrefγavref from a reference torque (Tref) and an estimated torque (T̂), which is determined from measured currents (iSabc,I, iSabc,II) and measured voltages (vSabc,I, vSabc,II) and/or reconstructed voltages (v̂Sabc,I, v̂Sabc,II) in the winding systems (20a, 20b); determining at least one difference flux reference ψS,diffrefγdiffref from at least one difference current reference iSdq,diffref; determining a winding flux reference ψS,IrefγIrefψS,IIrefγIIref for each winding system (20a, 20b) from the average flux reference ψS,avrefγavref and the at least one difference flux reference ψS,diffrefγdiffref by adding and/or subtracting the at least one difference flux reference ψS,diffref,γdiffref to the average flux reference (ψS,avref, γavref); determining switching commands (uabc,I, uabc,II) for each converter branch (14a, 14b) from the winding flux reference ψS,Iref,∡ψS,Iref,ψS,IIref,∡ψS,IIref and an estimated winding flux (ψ̂S,I, ψ̂S,I, ψ̂S,II, ψ̂S,II) for the winding system (20a, 20b), which is connected to the respective converter branch (14a, 14b), wherein the estimated winding flux (ψ̂S,I, ψ̂S,I, ψ̂S,II, ψ̂S,II) is determined from the measured current (iSabc,I, iSabc,II) and the measured voltages (vSabc,I, vSabc,II) and/or the reconstructed voltages (v̂Sabc,I, v̂Sabc,II) in the winding system (20a, 20b), and wherein the winding flux reference (ψS,Iref,∡ψS,Iref,ψS,IIref, ∡ψS,IIref) and the estimated winding flux (ψ̂S,I, ψ̂S,I, ψ̂S,II, ψ̂S,II) are provided to a dedicated model predictive controller (52) for the respective converter branch (14a, 14b), which determines the switching commands (uabc,I, uabc,II); and applying the switching commands (uabc,I, uabc,II) to each converter branch (14a, 14b).