Spectrum Reconstruction for Oil Cooler On Condition Monitoring

Machine learning is employed to reconstruct vibration spectra from the condition indicators recorded by a field data acquisition system. Those spectra are used to compute a custom condition indicator, enabling the expansion of an on- condition part life extension program to many more rotorcraft. This alternate, machine-learning driven method for computing an existing condition indicator is submitted for rigorous testing and qualification, and for fielding to the fleet analytics system. DEFINITIONS parameters determined through some algorithm that adjusts them using some set of training data. The data science and model V&V communities have overlapping terminologies which collide in this work. In Validation -- In software or model development, validation is model V&V, validation is the final activity concerned with an activity concerned with the determination that a product ensuring the model is useful for its intended application. But accomplishes the intended objective, such as a software in data science, the final activity concerned with model product performing its intended function or a model of a usefulness is referred to as testing, and employs a test set. In physical system reliably predicting some measured value of data science, validation is an activity that occurs early in the that system. model development process to ensure the model is good but not overfit. This is more similar to model V&V verification, Validation set -- In data science, a validation set is a set of data which is concerned with ensuring a model is correctly coded held back from training in some way, to provide some kind of and configured to the intended design. Also, it is worth insight into the expected out-of-sample performance of a pointing out that model calibration is a form of empirical fit, model. making it very similar to (if not indistinguishable from) data science model training or machine/statistical learning. Verification -- In software or model development, verification is an activity concerned with confirming that a product or Calibration – In model development, calibration is a process model conforms to its intended design, such as a software whereby a limited set of unknown parameters are adjusted in product reflecting a software requirement specification or a order to match the model outputs to some exemplar data, model configured with a set of physically measured typically experimental data (see test data definition). calibration data. Test data -- In modeling and simulation, test data is typically data which is measured from a physical system under some