A multicriteria fuzzy linear programming method for water supply system development planning (Q581252)

Starting with considerations on a certain decision problem concerning development planning of a water supply system, the author formulates a model of multicriterial linear programming which in general notation assumes the following form: (1) $\sum^{n}_{j=1}c_{kj}x_ j\to \min,\quad k=1,\dots,5$ (2) $\sum^{n}_{j=1}a_{ij}x_ j\leq (\geq)b_ i,\quad i=1,\dots, m_ 1$ (3) $\sum^{n}_{j=1}a_{ij}x_ j=0,\quad i=m_ 1+1,\dots, m_ 2$ (4) $x_ j\geq 0,\quad j=1,\dots, n$ Next, replacing in (1)--(2) all the coefficients by fuzzy numbers (of the L-R type) and in addition introducing the fuzzy goals \(\tilde g_ k\), \(k=1,\dots,5\), corresponding to particular criterial functions in (1), the author receives the following problem of fuzzy programming: (5) $\sum^{n}_{j=1}\tilde c_{kj}x_ j\lesssim \tilde g_ k,\quad k=1,\dots, 5$ (6) $\sum^{n}_{j=1}\tilde a_{ij}x_ j\lesssim (\gtrsim)\tilde b_ i,\quad i=1,\dots, m_ 1$ (7) $\sum^{n}_{j=1}a_{ij}x_ j=0,\quad i=m_ 1+1,\dots, m_ 2$ (8) $x_ j\geq 0,\quad j=1,\dots, n$ in which the first two conditions are fuzzy and the others are crisp. Next, based on the comparison index for fuzzy numbers of two types (optimistic and pessimistic indices), the author replaces the problem (5)--(8) by a certain crisp problem of multicriterial programming (which consists in maximization of the comparison indices). The size of the problem is almost twice as large as the size of the initial problem. This problem, in the case when the membership functions of the fuzzy coefficients in (5)--(6) are linear, becomes a multicriterial linear fractional programming problem. To solve it, the author proposes to use the interactive algorithm of Choo and Atkins.