Hybrid NSGA-II Algorithm on Multi-objective Inventory Management Problem - Intelligent Computing for Sustainable Energy and Environment - page 166

Tabl e 1. Diections for local search

r Q

Shortage cost high Should increase
Install cost high Should decrease Should decrease

After the NSGA-II, we examine every solution for its shortage cost and install
cost, if shortage cost is high (higher than α

mean ), then r increases to see if
new solutions are better under Pareto rule. Also, we check for solutions to see if
the install cost is too high and respond accordingly.

×

Serial Algorithm Combing NSGA-II and Local Search. GA is a solution
for global search so that it has worse local search ability. To combine it with
local search is good for problem solving. The combination has often been used
in single-objective problem. And some results have been made in multi-objective
problems. We here use a serial method to combine NSGA-II and local search. In
this algorithm, the property of this problem we analyzed before is also adopted
to simplify all this process. The algorithm process contains .

4 Simulation and Remarks

4.1 Different Demand Situations

It can be seen that when the market, or the demand is more unstable (distri-
bution variance is larger), the value of cost become larger. And we can see the
degree of risk from the value of cvar more obviously. This verifies the feasibility
of our model.

(a) Demand distribution is (5,4) (b) Demand distribution is (5,16)

Fig. 1. Result of cost and cvar under different demand distribution