Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Mixed phenomenological and neural approach to induction motor speed estimation B. Beliczynski, L. Grzesiak and B. Ufnalski Institute of Control and Industrial Electronics Warsaw University of Technology, POLAND Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Outline Problem statement 1 Feedforward neural network without preprocessing 2 Feedforward neural network with preprocessing 3 Some results and conclusion(s) 4 Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Hint #1: To reduce cost. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Hint #1: To reduce cost. Hint #2: To increase reliability – it is not uncommon for electric drives to operate in harsh environments. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Hint #1: To reduce cost. Hint #2: To increase reliability – it is not uncommon for electric drives to operate in harsh environments. Why neuroestimators as soft sensors? Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Hint #1: To reduce cost. Hint #2: To increase reliability – it is not uncommon for electric drives to operate in harsh environments. Why neuroestimators as soft sensors? Hint #1: They are innately able to cope with plant nonlinearities. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Motivation There are applications of induction motor drives that require only a moderate control precision. In such cases we tend to go into speed-sensorless operation, i.e. encoders are replaced with soft speed sensors. Why soft sensors? Hint #1: To reduce cost. Hint #2: To increase reliability – it is not uncommon for electric drives to operate in harsh environments. Why neuroestimators as soft sensors? Hint #1: They are innately able to cope with plant nonlinearities. Hint #2: They can be trained using real plant data – no mathematical model of the plant is needed. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Problem statement ω m u s u s ˆ ω m m o ? Induction ? motor i s Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Model of the induction machine – just to make you familiar with the plant we are talking here (not needed to synthesize the soft speed neurosensor) u s = R s i s + d d t ψ s + j ω ψ s ψ s (1) 0 = R r i r + d � � d t ψ r + j ω ψ s − p b ω m (2) ψ r ψ s = L s i s + L m i r (3) ψ r = L r i r + L m i s (4) J d d t ω m = m e − c t ω m − m o (5) m e = 3 � � 2 p b ψ s × i s (6) Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Straightforward approach ω m ( k ) = f ( ω m ( k − 1 ) , m o ( k ) , ψ s ( k − 1 ) , i s ( k − 1 ) , u s ( k )) (7) Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Recurrent neural network u s α u s α z -1 z -1 u s β u s β ω m ˆ ω m ˆ RNN RNN i s α (dynamic NN) (dynamic NN) m o ˆ i s β z -1 z -1 Bad idea! Practically impossible to get any robustness. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Feedforward neural network u s a u s b u s c u s a u s α FFNN ω m ˆ i s a (static u s b u s β FFNN FFNN ω m ˆ ω m ˆ NN) i s b i s a (static i s α (static NN) NN) i s c i s b i s β Still bad idea! Practically impossible to get any robustness. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Rotating reference frame u s d u s α u s q u s β ˆ ω m i s d FFNN i s α αβ /dq i s q i s β γ Flux estimator Much better but still sensitive to flux estimation errors. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Stationary reference frame u 1 u 2 u s α Nonlinear transformations, e.g. u s β u 3 ω m ˆ the Akagi power FFNN u 4 i s α components and/or i s β u 5 the instantaneous impedance components u 6 Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Problem statement Feedforward neural network without preprocessing Feedforward neural network with preprocessing Some results and conclusion(s) Nonlinear preprocessing � s α + u 2 u 2 u 1 = | u s | = (8) s β � s α + i 2 i 2 u 2 = | i s | = (9) s β u 3 = ℜ ( u ∗ s i s ) = u s α i s α + u s β i s β (10) u 4 = ℑ ( u ∗ s i s ) = u s α i s β − u s β i s α (11) � u s � = u s α i s α + u s β i s β = u 3 u 5 = ℜ (12) i 2 s α + i 2 u 2 i s s β 2 � u s � = u s β i s α − u s α i s β = − u 4 u 6 = ℑ (13) . s α + i 2 u 2 i s i 2 2 s β Almost the best we can do but still heuristic i.e. rather subjective choice. Bartlomiej Beliczynski, Lech Grzesiak and Bartlomiej Ufnalski Speed neuroestimation in induction motor drives
Recommend
More recommend
