As every new generation of civil aircraft creates more on-wing data and fleets gradually become more connected with the ground, an increased number of opportunities can be identified for more effective Maintenance, Repair and Overhaul (MRO) operations. Data are becoming a valuable asset for aircraft operators. Sensors measure and record thousands of parameters in increased sampling rates. However, data do not serve any purpose per se. It is the analysis that unleashes their value. Data analytics methods can be simple, making use of visualizations, or more complex, with the use of sophisticated statistics and Artificial Intelligence algorithms. Every problem needs to be approached with the most suitable and less complex method. In MRO operations, two major categories of on-wing data analytics problems can be identified. The first one requires the identification of patterns, which enable the classification and optimization of different maintenance and overhaul processes. The second category of problems requires the identification of rare events, such as the unexpected failure of parts. This cluster of problems relies on the detection of meaningful outliers in large data sets. Different Machine Learning methods can be suggested here, such as Isolation Forest and Logistic Regression. In general, the use of data analytics for maintenance or failure prediction is a scientific field with a great potentiality. Due to its complex nature, the opportunities for aviation Data Analytics in MRO operations are numerous. As MRO services focus increasingly in long term contracts, maintenance organizations with the right forecasting methods will have an advantage. Data accessibility and data quality are two key-factors. At the same time, numerous technical developments related to data transfer and data processing can be promising for the future.
Machine learning models have proven to be reliable methods in classification tasks. However, little research has been conducted on the classification of dwelling characteristics based on smart meter and weather data before. Gaining insights into dwelling characteristics, which comprise of the type of heating system used, the number of inhabitants, and the number of solar panels installed, can be helpful in creating or improving the policies to create new dwellings at nearly zero-energy standard. This paper compares different supervised machine learning algorithms, namely Logistic Regression, Support Vector Machine, K-Nearest Neighbor, and Long-short term memory, and methods used to correctly implement these algorithms. These methods include data pre-processing, model validation, and evaluation. Smart meter data, which was used to train several machine learning algorithms, was provided by Groene Mient. The models that were generated by the algorithms were compared on their performance. The results showed that the Long-short term memory performed the best with 96% accuracy. Cross Validation was used to validate the models, where 80% of the data was used for training purposes and 20% was used for testing purposes. Evaluation metrics were used to produce classification reports, which indicates that the Long-short term memory outperforms the compared models on the evaluation metrics for this specific problem.
Nowadays, one of the major current health risks is excessive sitting during work hours. Furthermore, the coronavirus disease 2019 (COVID-19) pandemic and the corresponding government state of emergency forced many people to work from home. These constraints carried out an important change in the lifestyle of people; for instance, the proportion of sitting time in front of a computer during working hours has increased considerably worldwide, particularly through the implementation of teleworking.In order to motivate people to lead a less sedentary life, the Hanze University of Applied Sciences Groningen developed an automated recommender system. We investigated the possibility of automated coaching in order to increase physical activity and help people to reach their daily step goal. By monitoring people’s activity level and progress during the day, we predict personalized recommendations. The effect of these recommendations on the individual’s activity level forms the basis for a personalized coaching approach.Step count data is used to train a machine learning algorithm that estimates the hourly probability of the individual achieving the daily steps goal. The outcome of this prediction is combined with the effect of the type recommendation for the individual to deliver the best recommendation for the individual. To show the practical usefulness, we constructed a platform to manage the data, rules, machine learning algorithms and clustering of participants. Results of initial pilots using the platform and app have given insight in the performance of and challenges associated with algorithm selection and personal model generation for the coaching package caused by the nature of the data. Further research will therefore be done in optimizing machine learning algorithms and tuning for human datasets.
