Titulo Estágio
Deep Reinforcement Learning for Condition Based Monitoring in Aircraft Planning
Áreas de especialidade
Sistemas Inteligentes
Local do Estágio
Laboratório de Redes Neuronais (LARN-CISUC)
Enquadramento
While your airplane is on its way to reach your destination, all airplane systems and components are also on their way, very slowly, from a healthy state to one of malfunction. The terabytes of data a modern airplane generates each day can be used to determine the health condition of all airplane parts, from wheels and brakes to air conditioning to structural integrity.
When a light indicates something to be wrong with an airplane system, it often may not be clear which part of a (sub) system is causing the error. “Using historical and actual airplane data, we can help pinpoint the root cause, saving time and money". Most maintenance in aviation is preventive (also known as pre-scheduled or interval-maintenance). Therefore, many systems and components are inspected while they are still in good health.
This proposal deals with Condition-Based Monitoring (CBM) approach where unscheduled airplane maintenance events are reduced, and airplane availability is increased.
Objetivo
This internship involves a decision support systems for a reliable and consistent application of CBM where uncertainties have to be taken into account The idea is to use machine learning techniques to build a decision support tool for optimizing maintenance.
The main objective of this proposal to deploy a Deep Reinforcement Learning for Condition Based Monitoring in Airplane. The first main goal is described in the following steps:
(i) Step 1: Data Collecting: to preprocessing and data analysis from sensors;
(ii) Step 2: Feature Engineering: Design hand-crafted features for representing the problem, etc.
(iii) Step 3: Feature Analytics: Perform data analytics to choose the best features for model building
The second main goal is to develop a model able to handle condition-based monitoring in aircraft planning, using machine learning and pattern recognition techniques, namely, deep reinforcement learning. The next steps should be followed:
(iv) Step 4: Model Building: Train and evaluate using machine learning techniques, especially neural networks.
(v) Step 5: Model Validation: Validate the overall Model with real data.
Plano de Trabalhos - Semestre 1
• Literature Review;
• Analysis of the RL system on several environments, possibly, using Gym and Baselines from OPENAI (https://gym.openai.com/);
• Select, prepare, and preprocess a collection of sensor datasets from aircrafts (e.g., health condition, Remaining Useful Time (RUL) of aircraft components, severity measures of degradation/failure extracted from the health diagnostics, previous scheduled intervals, time of the plane on the ground, etc.);
• Preliminary use of RL algorithms using previous datasets
• Writing of intermediate report.
Plano de Trabalhos - Semestre 2
•Develop a Deep Reinforcement Learning (DRL) system (e.g. Deep Q-Learning and/or Actor Critic Algorithm to for estimating the network parameters) as a component of a decision support system model for optimal planning of aircraft predictive maintenance;
•Analyze experimental results: e.g., study parameter values; compare performance of the reduced datasets vs. previous results, etc.;
•Writing of scientific article;
•Writing of the thesis.
Condições
This work will be carried out in the Laboratory of Neural Networks (LARN) of CISUC, where there will be a regular supervision and feedback on the behalf of the supervisor and co-supervisor.
Familiarity with machine learning and data mining algorithms and software tools are essential. Participating students will acquire valuable knowledge and experience with model building and data science by mining massive datasets, which skills are currently in high demand for various technology employers due to the relevance to various applications.
Observações
Grant under REMAP Project will be available during the second semester internship depending on the 1st semester candidate evaluation. A 3-month scholarship of 745 euros per month is foreseen for this work, renewable for another 3 months.
Logistics @Laboratory of Neural Networks (LARN)
DEI-FCTUC
Orientador
Orientadora: Bernardete Ribeiro (bribeiro@dei.uc.pt); Coorientadores: Catarina Silva, Penousal Machado
bribeiro@dei.uc.pt 📩