PSI - Issue 38

A. Débarbouillé et al. / Procedia Structural Integrity 38 (2022) 342–351 A. De´barbouille´ / Fatigue Design (2022) 1– ??

349

8

4. Methodology application

4.1. Model studied In this section, we describe the study case used to evaluate the robustness of the algorithm presented below. To get simulated measures to feed Kalman filter, a numerical simulation of a car rolling at 40 km / h on a track of more than 1000 meters has been done.

Fig. 1. Road profile

Before using the simulated measurement in the ACEKF, some noise is added to them. Two biaxle accelerometers, one biaxle gyrometer, one tachometer and one GPS are implemented onto the numerical car. The sensors noises are uncorrelated. These noises are described in the table 4.1. E ( ) is the noise bias and √ V ( ) defines the standard deviation.

√ V ( ) 25m / s² 0.1rad / s

Name bodies

mass

inertia

Sensors

E ( )

Car body

750 kg

93.75 kg.m²

Accelerometer

0 0 0 0

Front wheel Rear wheel Front junction Rear junction

40 kg 40 kg 45 kg 45 kg

10 kg.m² 10 kg.m² 7.5 kg.m² 7.5 kg.m²

Gyrometer

GPS

2m

tachometer

2m / s

Table 2. Characteristics of bodies

Table 1. parameter for the definition of measurements noises

Fig. 2. Geometry of the 2D model

The track used is a real road profile given in figure 1. The sensors measurements and sensors positions are imported into our algorithm computed in Matlab environment. Covariance matrix of measurement