Automated Robot-Based 3D Vibration Measurement System

The expectations of automobile customers for lower noise, reduced vibration levels and improved ride comfort have become ever more demanding on the engineering community. These demands together with the needs of automobile manufacturers to improve their product development efficiencies and costs are the main driving force behind the design of a revolutionary, fully automated, robot-based system for acquiring three-dimensional structure-borne vibration data with far higher measurement point density than is typically obtained using the conventional accelerometer approach. This article describes the system and presents results obtained from a production car body and chassis of an eco-vehicle. It also hints at potential future applications for structural health monitoring in the aerospace industry. Experimental modal analysis is traditionally a labor-intensive process that requires attaching accelerometers and cables to the surface of the structure under test. In the automotive industry, the surfaces are so large and complex that finite-element models used to simulate vibrational behavior generate a mesh comprising hundreds of thousands of elements. Limited availability of accelerometer channels and high-count sensor cost typically limit the number of simultaneous measurement points to around 100. To characterize structural behavior, many more locations must be measured and the accelerometers are therefore moved to a new set of locations and the test repeated. In the case of a body in white, dummy masses are added to the structure so that when the accelerometers are moved, overall structural behavior is unchanged. The effect of the mass of the large number of accelerometers, dummy masses and cables is of course significant and needs to be taken into account in the model when attempting to validate the simulation with measured data. For some structures, it may be impossible to completely model the effects of added mass in the simulation. For acoustics analysis at higher frequencies, measuring the required higher measurement point density is often impossible using a contact sensor approach. In addition to the resulting time and cost constraints, the entire process is complex and subject to human error. Each triaxial accelerometer needs to be positioned and axially aligned very precisely. Their locations on the structure also need to be known very accurately relative to the nodal points in the FE model if the data are going to be used to update the CAE model. Every one of the large number of sensors must be calibrated on a regular basis, and transducers cannot be allowed to fail, come unglued or suffer from a broken cable during the test. To reduce costs and improve time to market, noise, vibration and harshness (NVH) studies are today relying more and more on the simulation process. Fewer prototypes are being built, resulting in the need for significantly more thorough testing than in the past. This requires improved test productivity, data reliability, density and accuracy. All of this leads to the need for a fully integrated prototype testing approach to providing feedback for updating and refining the CAE models. A common approach for noncontact experimental modal analysis is scanning-laser Doppler vibrometry. Scanning laser Doppler vibrometers (SLDVs), introduced in the 1980s, 1 are systems for noncontact measurement of vibration utilizing the optical Doppler effect, which causes a shift in the frequency of light backscattered from a moving surface. The frequency shift is given by: surface. Sophisticated frequency decoding electronics produce a realtime output signal proportional to instantaneous velocity of the surface relative to the SLDV in the direction of the laser. Computercontrolled, galvanometer-driven mirrors scan the laser over the test surface to obtain vibration data at hundreds or even thousands of locations. Because the vibrometer output is directional and can be acquired, stored and processed in the same way as signals from an accelerometer, it’s possible using FFT signal processing together with multichannel data acquisition to acquire all of the information necessary for a modal analysis. The additional channels are used to simultaneously gather data from other sources such as load cells, accelerometers, and microphones. A complete experimental modal analysis for model updating requires the vibration amplitude and phase spectra in all three orthogonal axes of the test object’s coordinate system. It also needs to know the geometric shape of the object under test and the exact coordinate location of each measurement.