37
Adithya Arikere Financed by FFI

Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Adithya Arikere

Financed by FFI

Page 2: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•Find opportunities for novel or

enhanced active safety

functionality enabled by electric

drives • Study dynamics of accident scenarios

• Design and implement controllers for

interventions

• Validate controllers in simulations

• Validate controllers in experiments

Project objective

2

Page 3: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•For our purposes: Mathematical model

“mathematical (symbolic) representation of a concept, phenomenon,

relationship, structure, system, or an aspect of the real world”

•Can be used to explain, control or predict system behaviour based on inputs

or past observations

What is a model?

3

Page 4: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

No!

Do we need models?

4

Page 5: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•Versatility

•Online evaluation/computation/adaptation

•Extensibility

•Interactive

•Analysis and understanding

•Build and test hypothesis

•Generate theorems/proofs

•Generate simple solutions

•Knowledge is transferable

•...

Do we want models?

5

Page 6: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•Active safety

•Interventions just a second or two before a crash to

prevent or mitigate the same

•On- or near-limit interventions

•Highly non-linear dynamics

•Cost of failure can be high

Model requirements

6

Page 7: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

“All models are wrong, but some are useful”

Model requirements

7

- George Box (Statistician)

Page 8: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•Different ways to classify:

•Point mass, single-track, two-track

•Analytical/Numerical

•Monolithic/Modular

•Tyre model type

•Degrees of freedom

•...

Classification of models

8

Page 9: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

• Reference frame: Global

• Friction circle representation: Quadratic constraint (Cartesian coordinates)

Point mass (particle) model

9

𝑋

𝑌 𝑌

𝑋

𝑋

𝑌 𝐹𝑌

𝐹𝑋

Page 10: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

• Reference frame: Global

• Friction circle representation: Linear constraint (Polar coordinates)

Point mass (particle) model

10

𝑋

𝑌 𝑌

𝑋

𝑋

𝑌

𝐹

𝜙

Page 11: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Point mass (particle) model

11

• Reference frame: Local

• Friction circle representation: Quadratic constraint (Cartesian coordinates)

𝑋

𝑌 𝑢

𝑋

𝑌

𝐹𝑥

𝐹𝑦

𝜈

Page 12: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Point mass (particle) model

12

• Reference frame: Local

• Friction circle representation: Linear constraint (Polar coordinates)

𝑋

𝑌 𝐹

𝜙

𝑋

𝑌 𝑢

𝜈

Page 13: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Intersection accidents

Opponent Host

Left Turn Across Path – Opposite

Direction (LTAP/OD) 13

13

Page 14: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Intersection accidents

Opponent Host

𝒅

Distance margin

𝒗𝒃

𝒗𝟎

𝒀𝒃

Driver assist interventions only

Steering always performed by driver 14

14

Page 15: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Optimal control solution

15

Global reference frame Local reference frame

States (𝒙)

Hamiltonian

(𝐻)

Control

inputs (𝒖)

Costate

rate (𝜆 )

Costates

(𝜆) ? ? ? ? ?

⋮ ⋮ ⋮

Optimum fo-

rce angle (𝜙) 𝑵𝒐𝒕 𝒔𝒐𝒍𝒗𝒂𝒃𝒍𝒆

Page 16: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Single-track (bicycle) model

16

Equations of motion: Slip angles:

Page 17: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Linear single-track (bicycle) model

17

•Simplification: • Longitudinal velocity constant

• Linear tyres

Page 18: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Linear single-track (bicycle) model

18

•Simplification: • Longitudinal velocity constant

• Linear tyres

• Torque vectoring

Page 19: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Linear single-track (bicycle) model

19

• FWS = Front Wheel Steer

• DYC = Direct Yaw Control (no steering, torque vectoring only)

• YRC = Front Wheel Steering (steady state) + Torque Vectoring (transient)

Page 20: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Optimal control – Single track

20

Page 21: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Optimal control – Single track

21

𝐽′

Page 22: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Optimal control – Single track

22

Page 23: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Single-track (bicycle) model

23

•Common variations: • Constant longitudinal velocity

• Torque vectoring

• Steady state

• Longitudinal slip

• Longitudinal load transfer (dynamic or steady state)

• Lateral load transfer (tyre stiffness and capacity adaptation based on lateral acceleration)

• ....

Page 24: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Single-track (bicycle) model

24

•Use cases: •Understeer/oversteer characteristics

• Steady state or transient yaw or lateral acceleration response

• Stability and controllability analysis

•Reference generation • Commonly used as a yaw rate reference generator for ESC or other such

controllers

• Simplest model that captures yaw degree of freedom

• Validation of point mass results

• Analytical studies

Page 25: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Two-track model

25

Equations of motion:

Page 26: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Two-track model

26

Slip angles:

Page 27: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Two-track model

27

• Lateral transient load transfer: • Adds roll degree of freedom

• Steady state longitudinal load transfer

• Total wheel load

Page 28: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Optimal control – Two track

28

Optimum global force angle from numerical optimal control

Page 29: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Two-track model

30

•Common variations: • Wheel degrees of freedom, longitudinal slip (+4 DoF)

• Pitch and roll degrees of freedom (+2 DoF)

• Suspension modelling

• Steering models

• Rear-wheel steering or individual wheel steering

• Powertrain models

• Front, rear, all-wheel drive, torque vectoring, etc

• Steady state

• ....

Page 30: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Two-track model

31

•Use cases: • Preliminary testing of vehicle subsystems (engine, steering,

suspension, etc)

• Preliminary testing of active safety (or other control) functions

• Yaw stability analysis or verification of stability characteristics

• Simplest model that contains all major features influencing yaw stability

• Validation of point mass results

Page 31: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Tyre models

32

Magic formula tyre model

Page 32: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Tyre models

33

Magic formula tyre model

Page 33: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Tyre models

34

Magic formula tyre model

Page 34: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Tyre models

35

tanh tyre model

Page 35: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Tyre models

36

Linear tyre model

Page 36: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

•Reference generation / preliminary analysis almost always with point mass model

•Single-track / two-track model useful for verification and analysis

•Choice of tyre-model may be important

•High-fidelity models almost always for validation, never for reference generation

•Experiments ultimate validation tool

Summary

37

37

Page 37: Adithya Arikere - sveafordon.com · For our purposes: Mathematical model “mathematical (symbolic) representation of a concept, phenomenon, relationship, structure, system, or an

Thank you!

Questions?