Subventions et des contributions :

Subvention ou bourse octroyée s'appliquant à plus d'un exercice financier (2017-2018 à 2018-2019).

An electrical power grid comprises of three basic elements: power generating stations, load centres, andx000D
transmission lines. A transmission line connects a generating station with a load centre, and the topology of ax000D
power grid evolves over time. Drawing a new transmission line from a generating station to a load centre is ax000D
complex decision process involving: (D1) satisfaction of feasibility constraints, policies, and environmentalx000D
constraints, among others; (D2) cost-benefit analysis; and (D3) optimization of power flow. Before carrying outx000D
steps D2 and D3, a grid operator may want to evaluate and rank all possible instances of a new transmissionx000D
line by considering D1. In this project, we will develop a framework to evaluate and rank the instances of ax000D
new transmission line in a given power grid. The inputs to the framework will be: (i) the topology of the powerx000D
grid; (ii) a GPS (Global Positioning System) annotated geographic map of the physical space covered by thex000D
grid; (iii) important features of the geographic space, namely, cities, water bodies, mountains, and off-limitx000D
areas; and (iv) regulatory and environmental constraints. The output from the framework will be a ranked listx000D
of potential routes for a new transmission line. Each instance of a new transmission line will be described byx000D
means of a tuple: <source, destination, rank, sequence of GPS locations>. The <source, destination> pair willx000D
identify the two end points of a transmission line, the sequence of GPS locations will define the physical route,x000D
and the rank of the transmission line will indicate the degree of its satisfaction of all its input constraints. Thex000D
ranking problem on hand is a challenging one because of two reasons: (i) there is a need to identify a physicalx000D
route in the geographic context; and (ii) all possible routes need to be evaluated against the given constraints tox000D
produce a total ordering. We will achieve two milestones in this project: model the geographic space coveredx000D
by the power grid as a multi-feature labeled graph; and design a machine-learning based algorithm to evaluatex000D
and rank the top routes. Results from this research will enable the industry partner, SNC-Lavalin, to considerx000D
developing a software system for the automation of selection of routes for new transmission lines.