4 Presentations

• 
  10:45 AM - 11:10 AM

  ANNULÉ/CANCELLED Designing a cellular hybrid manufacturing-remanufacturing system considering alternative process routings and contingency process routings

  • Amirreza Hooshyar Telegraphi, presenter, PhD Candidate
  • Akif Bulgak, Université Concordia
  • Michel Gendreau, Polytechnique Montréal
• 
  11:10 AM - 11:35 AM

  Reinforcement learning for forest value chain optimization

  • Ahmed Ragab, presenter, Polytechnique Montreal
  • Ichrak lakhdhar, Natural Resources Canada (NRCan)
  • Mouloud Amazouz, Natural Resources Canada (NRCan)

  Forest value chain optimization (FVCO) represents a dynamic stochastic problem. A big challenge is to select the appropriate technique that finds the optimal actions in a huge number of states. Reinforcement learning can be a promising artificial intelligence (AI) technique used for that purpose.

  Keywords: Value Chain Optimization; Artificial Intelligence; Reinforcement Learning
• 
  11:35 AM - 12:00 PM

  Data-driven optimization of industrial process operations

  • François Aubé, presenter, CanmetÉnergie
  • Mouloud Amazouz, Natural Resources Canada (NRCan)

  Many industrial process control systems are partly adjusted by operators when unexpected events occur such as sudden decline in performance, quality, and change in regime or operating mode. Such duty is difficult due to the complexity of such processes. A machine learning approach is proposed to support operators in their decision making process to better handle such events and ensure real-time optimal operation.

  Keywords: machine learning, operations
• 
  12:00 PM - 12:25 PM

  Open big data and AI platform for efficient and sustainable natural resources operations

  • Mouloud Amazouz, presenter, Natural Resources Canada (NRCan)

  Making environmentally, economically and socially sound and efficient decisions for natural resource management and exploitation has become increasingly difficult. Natural resource sectors stakeholders have access to too much raw data, too many decisions to make, and too little time to do any of it. In addition, the objectives are changing and the diversity of needs is growing. Only computer decision aids can help sort and process large amounts of data (Big data) and expand the ability to make good decisions in the face of these constraints. Artificial intelligence (Al) technology allows inclusion of knowledge processing in the decision support environment. AI and Big Data have started making major changes in the business world, as companies are utilizing the power of data analytics to affect positively their bottom line, resulting in increased productivity and revenues. Unfortunately, such changes are slow to come in the natural resource sectors – the forestry, mining, agriculture sectors, to name a few –, where most of the data collected today are not used, and the data that are used, not fully exploited.
  Despite the advances in technologies such as IoT, sensors, terrestrial Lidar, Aerial Lidar, Satellite Imaging and Drone based imaging that produce large amounts of data, and machine learning algorithms, the forest sector stakeholders are still struggling to make the right decisions. This is due to the unavailability of powerful decision support tools. Such tools should be able to automatically collect, integrate, prepare and process scattered Big data sets from different heterogeneous, structured and unstructured sources. In fact, foresters need to have easy access to data products to take the right decisions not raw data and generic commercial AI platforms.
  The main objective of the project, which will be presented in this talk, is to develop an open Big data and artificial intelligence (AI) platform for efficient and responsible development and use of Canada’s Natural Resources. With a focus on multi-objective value chain optimization (VCO), the vision is to improve decision-making processes at the strategic, tactical and operations planning levels starting with the forest sector, in close collaboration with the Canadian Wood Fiber centre (CWFS).