skrtks / powerrangers

Heuristieken Smart Grid

powerRangers

For 'programmeertheorie I' we have to find an efficient solution for connecting houses to batteries in a smartgrid for 3 districts. Houses produce solar energy which have to be stored in a battery if a household doesn't use the energy itself. To keep the cotst low the battery has to be filled as suffiecient as possible without overflowing the battery and we try to keep the cables between a house and a battery as short as possible. Houses share cables when they lead to the same battery, this reduces the costs. Cables have to go around houses, because it would be very expensive to lay a cable underneath a house.

A district has a size of 50 x 50 and contains 150 houses with different energy outputs and 5 batteries with a capacity of 1500kW each. We use different algorithms to find a solution to connect houses to batteries and algorithms to lay cables between houses and batteries.

python files

main.py:

• Process command line arguments

smartGrid.py:

• housesList: List containing house objects
• batteries: List containing battery objects
• gridPoints: List containing gridPoint objects
• gridFiller(self): Create grid
• assignGridInfo(self): Assign gridIDs to batteries and houses and change cost to 5000 if there is a house on a gridpoint
• gridDrawer(self): Draw grid with batteries, houses and connections
• fileReader(self, fileHouses, fileBatteries): Read information of houses and batteries from files
• manhattanDistance(self): Calculate manhattan distance for every gridpoint to each battery
• children(self, gridPoint): Returns gridpoint ID's for possible moves from current gridpoint

gridClass.py: Class for grid segments, takes ID, x and y coordinates, cable costs and manhattan distance to batteries

houseClass.py: Define a class for houses, takes ID, x and coordinates, power output, boolean connected, score, batteryID and manhattan distances to batteries

batteryClass.py: Define a class for batteries, takes ID, x and y coordinates, capacity and list of connected houses

connecters.py:

• randomConnecter(smartGrid): Finds connection for houses to batteries with preverence for batteries with the smallest manhattan distance.
• randomWithPreference(smartGrid): Connect all houses by sorting them on power output (biggest to smallest) with nearest batteries
• greedyAlgorithm(smartGrid): Connect all houses by sorting them on power output (biggest to smallest)

dijkstra.py: Dijkstra algorithm to find shortest path between battery and house

pathfinder.py: Algorithm to find a short path between house and battery

CSVWriter.py: Make new CSV file containing results

hillCLimber.py: Swap houses several times to find better solution to append houses to battery

Prerequisites

  Mathlib
  Plotlib
  Python3

Change settings

When running main.py you will be asked which district to run, which algorithm to use and whether to apply the hillclimber.

To change the number of loops for running there are three variables defined in main.py at the top of the file that can be changed:

• numberOfLoops: Choose how many times you want to find new connections and hillclimb (within these loops you can define the number of swaps and connection trials as explaind next).

• numberOfConnections: Choose how many connections you want to try to optimalize the connections. This option is for the randomWithPreference algorithm.

• numberOfSwaps: Choose how many times you want to let the hillclimber swap houses.

For demonstrating the code these variables are set low. For generating our results we set these values to (numberOfLoops: 50, numberOfConnections: 25, numberOfSwaps: 100000).

Built With

  Python3

Authors

  Sam Kortekaas            [skrtks](https://github.com/skrtks)
  Daphne Witmer            [10588094](https://github.com/10588094)
  Ida Segond von Banchet   [igcsvb](https://github.com/igcsvb)

Acknowledgments

  Dijkstra inspiration from Amit Patel [redblobgames] (https://www.redblobgames.com/pathfinding/a-star/implementation.html)
  Pathfinding inspiration from (http://web.mit.edu/eranki/www/tutorials/search/) and (https://gist.github.com/jamiees2/5531924)
  Tech assistent: Maarten van der Sande