Application of Model Predictive

Control in a
Resolver-to-Digital Converter
Authors
Mr. Thyago Vasconcelos Estrabis
Dr. Raymundo Cordero García
Dr. Edson Antonio Batista
Dr. Cristiano Quevedo Andrea
Me. Márcio Afonso Soleira Grassi
Application of Model Predictive Control in a Resolver-to-Digital
Converter

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

2
Application of Model Predictive Control in a Resolver-to-Digital 1.
Converter Intro

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Propose System
6.Results
7.Conclusion

3
 Application of Model Predictive Control in a Resolver-to-Digital 1.
Converter Intro

Aircrafts

Electric/Hybrid Vehicles. CNC

Robotic under difficult conditions

4
Application of Model Predictive Control in a Resolver-to-Digital 1.
Converter Intro

• MPC can be applied to get the angular position from a

resolver.

• First time MPC is applied to RDC.

5
Application of Model Predictive Control in a Resolver-to-Digital 2.
Converter Objective

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

6
Application of Model Predictive Control in a Resolver-to-Digital 2.
Converter Objective

• Study the application of MPC in the development of a

resolver to digital converter.

7
Application of Model Predictive Control in a Resolver-to-Digital 3.
Converter Resolver

1.Introduction
2.Objective
3.Resolver and RDC
4.MPC
5.Proposed System
6.Results
7.Conclusion

8
Application of Model Predictive Control in a Resolver-to-Digital 3.
Converter Resolver

• Type of a rotary electrical transformer.

• An angular position sensor widely used in applications where

reliability is required.

• This sensor produces amplitude-modulated output signals.

• Getting the angular position from resolver is more difficult

than from an encoder.

• An ATO is a closed-loop observer whose objective is to

reduce the angle estimation error”.

9
 Application of Model Predictive Control in a Resolver-to-Digital 3.
Converter Resolver

vs(t)
 
ve(t)

Rotary Excitation q Output

transformer winding windings
vc(t)

Structure of the resolver.

10
Application of Model Predictive Control in a Resolver-to-Digital 3.
Converter Resolver

ve(t)

Acq.
system

 
vs(t)

estimated
keaesin(q)
Sample multiplied by -1

Synchronous demodulation.
11
Application of Model Predictive Control in a Resolver-to-Digital 4.
Converter MPC

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

12
Application of Model Predictive Control in a Resolver-to-Digital 4.
Converter MPC

• The MPC is a process control.

• The MPC predict the change in the dependent variables of the

modeled system that will be caused by changes in the
independent variables.

• The term is not used for a specific model.

• The choice of a specific approach depends on the application

13
 Application of Model Predictive Control in a Resolver-to-Digital 4.
Converter MPC

Where y(k) is the system output, xm(k) is a state variable vector

and u(k) is input.

represent the augmented GPC

This method has an integrator embedded

14
Application of Model Predictive Control in a Resolver-to-Digital 4.
Converter MPC

Y    =  Fx (k i ) + FD U Nc

 
−1
⏞
K mpc   =   [1 0…0]   ( FT
F+ Ŕ ) ( F T
F)
The output Y depends on the
current state and the movement of K
 
mpc   =   [ K x K y ]
future control

 
  CA

[ ]
CA ²
F = CA ³
⋮
CA N p

15
Application of Model Predictive Control in a Resolver-to-Digital 5.
Converter P.System

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

16
 Application of Model Predictive Control in a Resolver-to-Digital 5.
Converter P.System

g =q - qe
q + K MPC qe
–
 
Linearized proposed ATO.

g   = sin ( q  −   q e )  ≈  q  −  q e
 

17
 Application of Model Predictive Control in a Resolver-to-Digital 5.
Converter P.System

Proposed RDC algorithm, where the MPC acts as an ATO.

Adapted GPC used in the ATO

18
Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

19
Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Mechanical speed used for the plants 1/s and 1/s².

20
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b) Np = 10 Nc = 1; (c) Np = 10 Nc = 5

Simulation results, with noise, for the plant

1/s and for Np = 2 and Nc = 1: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom.

21
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b) Np = 10 Nc = 1; (c) Np = 10 Nc = 5

Simulation results, with noise, for the plant

1/s and for Np = 10 and Nc = 1: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom.

22
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b) Np = 10 Nc = 1; (c) Np = 10 Nc = 5

Simulation results, with noise, for the plant

1/s and for Np = 10 and Nc = 5: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom.

23
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s² and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b)Np = 10 Nc = 1; (c) Np = 10 Nc = 5.

Simulation results, with noise, for the plant

1/s² and for Np = 2 and Nc = 1: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom.

24
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s² and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b)Np = 10 Nc = 1; (c) Np = 10 Nc = 5.

Simulation results, with noise, for the plant

1/s² and for Np = 10 and Nc = 1: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom.

25
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

Simulation results for the plant 1/s² and for different values of Np and
Nc: (a) Np = 2 Nc = 1; (b)Np = 10 Nc = 1; (c) Np = 10 Nc = 5.

Simulation results, with noise, for the plant

1/s² and for Np = 10 and Nc = 5: (a) Angle
Estimation Error; (b) Angle Estimation Error
Zoom; (c) Estimation error; (d) Estimation
Error Zoom

26
 Application of Model Predictive Control in a Resolver-to-Digital 6.
Converter Results

First order plant simulation with noise Second order plant simulation with noise
Parameters
Np = 2 Np = 10 Np = 10 Np = 2 Np = 10 Np = 10
Nc = 1 Nc = 1 Nc = 5 Parameters
Nc = 1 Nc = 1 Nc = 5
0.0091 0.0034 0.0146 0.0146 0.0031 0.0146
-7.9582 1.5510* -1.4974
-2.5487 2.1468* 2.5487
median 1.6876* -6.0595* -5.2535
median median -4.723* 2.8976* -4.7230*
-0.0256 -0.0077 -0.0328
-0.0256 -0.0077 -0.0328 -0.0325 -0.0093 -0.0325
0.0279 0.0104 0.0381
0.0279 0.0104 0.0381 0.0389 0.0062 0.0389

27
Application of Model Predictive Control in a Resolver-to-Digital 7.
Converter Results

• Nc far from Np – smaller noise amplitude but presented bigger error than
other simulation.

• Nc near Np – or the ratio between the two presented better result for the noisy
plant.

• Np = 2 and Nc = 1 – the best choice with lower peaks values at steady state.

28
Application of Model Predictive Control in a Resolver-to-Digital 7.
Converter Conclusion

1.Introduction
2.Objective
3.Resolver
4.MPC
5.Proposed System
6.Results
7.Conclusion

29
Application of Model Predictive Control in a Resolver-to-Digital 7.
Converter Conclusion

• A modified GPC algorithm for the implementation of a RDC was proposed to

estimate the angular position from the resolver signals.

• This is the first time MPC theory is applied to create an angle tracking
observer (ATO).

• The error was insignificant even during the acceleration

• Find a way to reduce noise interference.

30