One of our industrial partners - LMS International, a hi-tech company located in Leuven, Belgium - is offering summer internships to top students from the fields of control and automation. The exact period is negotiable but the position should last at least three months. LMS will offer a small apartment in their own dormitory (just next to their facilities) not far from the center of the historial city of Leuven. On top of that, LMS will pay a monthly living allowance of 250 Euro. LMS hosts a lot of foreign students and they are housed in the same building, therefore, a lot of new contacts with people from all around the world will be certainly made. The work started during the proposed started internship can be turned into a graduate project assignment on a supervision of which researchers both from LMS and from AA4CC will collaborate.
Technical problem description: The flexibility of a car body has an influence on vehicle dynamics. The importance of the flexibility of the body during handling maneuvers has already been shown. Full vehicle multibody models are widely used to improve handling and ride comfort performances of passenger cars. However, when focusing on body, it is difficult to validate the simulation results as the forces at the body/suspension interface cannot be measured.
LMS has a unique test-based technology to identify the individual forces acting in the suspension-to-body connection points and to visualize the body-deformation during handling maneuvers. The body force identification is based on strain data. The frequency range of interest is between 0 and 5 Hz. Strain-gauges are capable of accurately measuring the quasi-static body-deformation while the signals are not saturated by rigid-body DC behavior (as would be the case when DC-accelerometers are used). To identify all suspension-to-body forces a large number of strain-gauges is required to achieve enough over-determination for the inverse method for force-estimation. The strain data are measured with a dynamic measurement system (Scadas Mobile 200 channel system). The time-domain body-forces are identified using an inverse methodology. The method requires transfer functions from force-input in the suspension connections to the strain-responses on the body. These transfer functions are measured in trimmed-body condition and represent the calibration of the forces acting on the body relative to the measured strain. Once these transfer functions are measured. The car is brought in its original conditions and road tests are performed, where dedicated maneuvers either with a test pilot or test robot are performed. With the measured operational strain and the strain-over-force transfer functions the time-domain forces can be estimated using matrix inversion.
The procedure that will be investigated will first compute an initial guess of the suspension forces based on the matrix inversion method that has been used in the past. To improve the accuracy, the obtained force time series are applied to a mathematical model of the car, which calculates accelerations on the body of the car. During the operational measurements with the strain gauges, also accelerations on the same locations as in the mathematical model are recorded. Accelerometer instrumentation is a much easier process then strain gauges. Based on the differences between the measured accelerations and the ones computed by the model, corrections on the initial suspension forces are calculated. The latter will be performed based on methodologies from the control community, called state estimation methodologies. The most well-known state estimator is the Kalman filter. More advanced methods are extended Kalman filter, unscented filter, grid based methods, particle filter, moving horizon estimator, …
In this thesis, a study will be performed on a CAE model to determine good sensor locations (observability of the body forces). The whole procedure will be tested on a CAE model. Also experimental data will be put available