1 Quality checks for source characterization

SOURCE offers various quality indicators for the source characterization process. There are quality checks throughout the whole process, from input data all the way to Blocked Force results. By integrating these checks into your measurement routine, you can recognize problems early, analyze them with your colleagues and find solutions without starting from scratch again. This saves time, costs, and gives you control over the quality of your source characterization.

First-time-right source characterization

Setup checks

Operational consistency (link) shows the consistency of the Virtual Point Transformation with respect to the measured response data. This allows to spot moved or loose sensors quickly. Operational deflection shapes (ODS) (link) help to identify and correct such errors in the sensor setup.

Signal checks

Any signal content not created by the source you want to characterize reduces the quality of the Blocked Force model and is, thus, noise (link). This can be due to other sources within the system or sources outside the system, e.g., test-bench machinery. To identify the signal-noise ratio (SNR), you can perform noise measurements where the source is not running and compare them to operational measurements. ODS animations and operational consistency help to spot and signal problems such as overloads or electromagnetic influence.

Model checks

The Blocked Force model is supposed to only model the source characteristics. This can be evaluated by calculating Blocked Forces also for artificial excitations and noise measurements and comparing the results to the Blocked Forces of the operational measurement.

An artificial excitation simulates the operational excitation of the source in a simplified way under conditions in which additional excitation not produced by the active component (e.g., noise, test-bench machinery) is minimized. For an artificial excitation measurement, the source and any auxiliary system are turned off, and a hammer impact on the source simulates the excitation. You can calculate Blocked Forces for this test condition and perform an on-board validation of the artificial excitations to understand whether all the interfaces are modeled correctly or if a transfer path is missing in the model.

You can investigate the Blocked Force noise (link) to estimate the noise content in the Blocked Force model of an operational excitation. For this, you calculate Blocked Forces for a noise measurement of the source in idle condition. In this state, the Blocked Forces only contain the signal from secondary sources and auxiliary machinery on the test bench. You can compare Blocked Force noise with Blocked Forces of the operational measurements to check whether your source characterization is polluted by noise at some frequencies.

You can verify your model with an on-board validation of the operational measurements. An on-board validation consists of comparing the measurement of the validation sensors with the predicted responses calculated using the Blocked Forces in that same assembly.

The main goal in source characterization is to characterize the source independent of the receiver. Thus, the ultimate validation of a Blocked Force model is a transfer validation: the prediction of the system response to the source excitation in a different assembly. Even without a transfer validation, the Blocked Force model can be validated with the steps above. Whenever you approach modeling of a source for the first time, it is recommended to plan for an additional transfer validation.

Integrating quality checks into the measurement routine

Using quality checks is always a good idea. But the real benefit comes when you include all quality checks in your measurement routine, as you can spot and resolve problems directly when they occur.

The first checks are done on the computer – before approaching the measurement. While preparing your test setup in DIRAC, you can verify the conditioning of each Virtual Point. This checks if your setup allows to identify every degree of freedom. When satisfied with the quality, you can create the physical setup at the test bench or vehicle.

After the component is instrumented, only measure some artificial excitations (e.g., one hammer impact in each direction) in DIRAC. These few measurements allow you to ensure that your measurement equipment functions correctly and that you have physically recreated the same setup as defined in DIRAC. These artificial excitations will also be used later in SOURCE for additional quality checks.

After checking that the component is set up correctly, you must perform the noise floor measurement and the operational measurements. At this stage, you need to check SNR by comparing the noise floor signal and operational measurements. You should also perform other signal checks using ODS and operational consistency.

When you have done all the operationals and are satisfied with their signal quality, it is time to perform the FRF tests in DIRAC. After measuring all excitations of the first Virtual Point group, you must verify the signal and model quality with various indicators. This ensures that the measured FRFs are coherent and without overloads and that the positions of sensors and excitations are correct (consistency). When the dynamics of the VP-transformed FRF look plausible, you can move on to the next Virtual Point.

After finishing all the FRF measurements, you go to SOURCE and post-process the artificial excitation data. You calculate Blocked Forces, perform an on-board validation, and compare the noise floor measurements. Of course, you want your signal to be higher than the measured noise, to avoid modeling noise. If your on-board validation is not polluted by noise but has poor quality, you might not have modeled all interfaces and one or more transfer paths could be missing. If that is the case, consider making some changes to the Design of Experiment.

If the on-board validation of the artificial excitations is good, you can post-process all the operational data and compare them with the noise floor measurement. With this step, you check that none of the measured operational conditions are polluted by noise at any frequency. You can then perform an on-board validation of your operationals to verify the quality of your source characterization results.

With a measurement routine with integrated quality checks, you will build up high-quality test-based models – Virtual Point by Virtual Point and good high-quality source characterization for all the operational conditions.