Tanja Dorst

In recent years, data science and engineering have faced many challenges concerning the increasing amount of data. In order to ensure findability, accessibility, interoperability, and reusability (FAIRness) of digital resources, digital objects as a synthesis of data and metadata with persistent and unique identifiers should be used. In this context, the FAIR data principles formulate requirements that research data and, ideally, also industrial data should fulfill to make full use of them, particularly when Machine Learning or other data-driven methods are under consideration. In this contribution, the process of providing scientific data of an industrial testbed in a traceable and FAIR manner is documented as an example.

Humans spend most of their lives indoors, so indoor air quality (IAQ) plays a key role in human health. Thus, human health is seriously threatened by indoor air pollution, which leads to 3.8 ×10⁶ deaths annually, according to the World Health Organization (WHO). With the ongoing improvement in life quality, IAQ monitoring has become an important concern for researchers. However, in machine learning (ML), measurement uncertainty, which is critical in hazardous gas detection, is usually only estimated using cross-validation and is not directly addressed, and this will be the main focus of this paper. Gas concentration can be determined by using gas sensors in temperature-cycled operation (TCO) and ML on the measured logarithmic resistance of the sensor. This contribution focuses on formaldehyde as one of the most relevant carcinogenic gases indoors and on the sum of volatile organic compounds (VOCs), i.e., acetone, ethanol, formaldehyde, and toluene, measured in the data set as an indicator for IAQ. As gas concentrations are continuous quantities, regression must be used. Thus, a previously published uncertainty-aware automated ML toolbox (UA-AMLT) for classification is extended for regression by introducing an uncertainty-aware partial least squares regression (PLSR) algorithm. The uncertainty propagation of the UA-AMLT is based on the principles described in the Guide to the Expression of Uncertainty in Measurement (GUM) and its supplements. Two different use cases are considered for investigating the influence on ML results in this contribution, namely model training with raw data and with data that are manipulated by adding artificially generated white Gaussian or uniform noise to simulate increased data uncertainty, respectively. One of the benefits of this approach is to obtain a better understanding of where the overall system should be improved. This can be achieved by either improving the trained ML model or using a sensor with higher precision. Finally, an increase in robustness against random noise by training a model with noisy data is demonstrated.

Measurement data can be considered complete only with an associated measurement uncertainty to express knowledge about the spread of values reasonably attributed to the measurand. Measurement uncertainty also allows to assess the comparability and the reliability of measurement results as well as to evaluate decisions based on the measurement result. Artificial Intelligence (AI) methods and especially Machine Learning (ML) are often based on measurements, but so far, uncertainty is widely neglected in this field. We propose to apply uncertainty propagation in ML to allow estimating the uncertainty of ML results and, furthermore, an optimization of ML methods to minimize this uncertainty. Here, we present an extension of a previously published automated ML toolbox (AMLT), which performs feature extraction, feature selection and classification in an automated way without any expert knowledge. To this end, we propose to apply the principles described in the “Guide to the Expression of Uncertainty in Measurement” (GUM) and its supplements to carry out uncertainty propagation for every step in the AMLT. In previous publications we have presented the uncertainty propagation for some of the feature extraction methods in the AMLT. In this contribution, we add some more elements to this concept by also including statistical moments as a feature extraction method, add uncertainty propagation to the feature selection methods and extend it to also include the classification method, linear discriminant analysis combined with Mahalanobis distance. For these methods, analytical approaches for uncertainty propagation are derived in detail, and the uncertainty propagation for the other feature extraction and selection methods are briefly revisited. Finally, the use the uncertainty-aware AMLT is demonstrated for a data set consisting of uncorrelated measurement data and associated uncertainties.

Process sensor data allow for not only the control of industrial processes but also an assessment of plant conditions to detect fault conditions and wear by using sensor fusion and machine learning (ML). A fundamental problem is the data quality, which is limited, inter alia, by time synchronization problems. To examine the influence of time synchronization within a distributed sensor system on the prediction performance, a test bed for end-of-line tests, lifetime prediction, and condition monitoring of electromechanical cylinders is considered. The test bed drives the cylinder in a periodic cycle at maximum load, a 1 s period at constant drive speed is used to predict the remaining useful lifetime (RUL). The various sensors for vibration, force, etc. integrated into the test bed are sampled at rates between 10 kHz and 1 MHz. The sensor data are used to train a classification ML model to predict the RUL with a resolution of 1 % based on feature extraction, feature selection, and linear discriminant analysis (LDA) projection. In this contribution, artificial time shifts of up to 50 ms between individual sensors' cycles are introduced, and their influence on the performance of the RUL prediction is investigated. While the ML model achieves good results if no time shifts are introduced, we observed that applying the model trained with unmodified data only to data sets with time shifts results in very poor performance of the RUL prediction even for small time shifts of 0.1 ms. To achieve an acceptable performance also for time-shifted data and thus achieve a more robust model for application, different approaches were investigated. One approach is based on a modified feature extraction approach excluding the phase values after Fourier transformation; a second is based on extending the training data set by including artificially time-shifted data. This latter approach is thus similar to data augmentation used to improve training of neural networks.