Solar Panels String Optimization
Partner: Renewal energy company in Thailand (Undisclosed)
Achieve the best performance available from an array of Photovoltaic (PV) Array, all panels within a string has to perform relatively identical. Any mismatches will result in reducing overall performance to the worst performing panel within the string. As a result, all the panel has to be sorted to arrange equally perform panels together. This operation is not trivial due to the massive number of panels installed in a solarfarm. There are also cost associated with removal/installation of the panels which proportional to the number of tasks needed to perform. In order to reduce this cost, we develop an optimization method that searches for a list of tasks which has to lowest cost to perform.
We define the total cost to move all the panels to be:
total_cost = (cost per panel)*(# of relocation) + (cost per meter)*(distance of travel)
This is not trivial. In order to determine which panels that should be moved, we need to know what are the swaps that give us the shortest path. In the opposite, we cannot determine the shortest path without knowing which panels that should be relocated. Unfortunately, optimizing both of the terms at the same time does not practical due to the massive size of search space.
We developed a mixture of stochastic/heuristic search with integer linear programming to find the relocation steps that has shortest path and with the smallest number of relocated panels.
