ML Wiki
Machine Learning Wiki - A collection of ML concepts, algorithms, and resources.

Waste Utilization Problem

Waste Utilization Problem

This is a Multi-Objective Optimization Problem

  • 2 cities $X$ and $Y$ produce garbage
  • there are two incinerators $I_1$ and $I_2$ with come capacity
  • each incinerator has some cost of utilization per unit of waste
  • there are some transportation costs per unit from a city to an incinerator

How much garbage to send from $X$ and $Y$ to $I_1$ and $I_2$?

Example:

  • $X$ produces 100 tons of garbage, $Y$ produces 150 tons
  • transportation costs: $X \to I_1: 2, X \to I_2: 3; Y \to I_1: 3, Y \to I_2: 4$
  • cost of utilization: $I_1: 2, I_2: 1$
  • capacities of $I_1$ and $I_2$ are 150

Image

Define the following variables:

  • $XI_1, XI_2$ - garbage sent from $X$ to $I_1$ and $I_2$ respectively
  • $YI_1, YI_2$ - garbage sent from $Y$ to $I_1$ and $I_2$ respectively

Constraints:

  • amount of produced garbage = amount of incinerated garbage
    • $XI_1 + XI_2 = 100$
    • $YI_1 + YI_2 = 150$
  • $I_1$ and $I_2$ have capacity:
    • $XI_1 + YI_1 \leqslant 150$
    • $XI_2 + YI_2 \leqslant 150$
  • all must be positive
    • $XI_1, XI_2, YI_1, YI_2 \geqslant 0$

So we have the following objectives:

  • we want to minimize the total cost of incineration
  • : $z_1 = 2 \cdot (XI_1 + YI_1) + 1 \cdot (XI_2 + YI_2)$
  • and we want to minimize the transportation cost
  • : $z_2 = 2 \cdot XI_1 + 3 \cdot YI_1 + 3 \cdot XI_2 + 4 \cdot YI_2$

We evaluate all feasible solutions against $z_1$ and $z_2$

  • and get something similar to this
  • Image
  • the Pareto-optimal solutions dominate all other feasible solutions

How to select the best one?

  • Weighed Sum
    • will select only 4 solutions, the rest is ignored
    • Image
  • Ideal Point
    • we find the closest point to the ideal
    • Image
  • Routing Optimization for Waste Management http://www.ma.iup.edu/~jchrispe/ORArticles/WasteManagement.pdf
  • Ant Colony optimization for Waste Utilization problem: http://thescipub.com/pdf/10.3844/jmssp.2009.199.205

Sources