A Decision Support System for Data-Driven Driver-Experience Augmented Vehicle Routing Problem

Logistics delivery companies typically deal with delivery problems that are strictly constrained by time while ensuring optimality of the solution to remain competitive. Often, the companies depend on intuition and experience of the planners and couriers in their daily operations. Therefore, despite the variability-characterizing daily deliveries, the number of vehicles used every day are relatively constant. This motivates us towards reducing the operational variable costs by proposing an efficient heuristic that improves on the clustering and routing phases. In this paper, a decision support system (DSS) and the corresponding clustering and routing methodology are presented, incorporating the driver’s experience, the company’s historical data and Google map’s data. The proposed heuristic performs as well as [Formula: see text]-means algorithm while having other notable advantages. The superiority of the proposed approach has been illustrated through numerical examples.

OPERATIONAL VARIABLE JOB SCHEDULING WITH ELIGIBILITY CONSTRAINTS: A RANDOMIZED CONSTRAINT‐GRAPH‐BASED APPROACH

In this study, we consider the problem of Operational Variable Job Scheduling, also referred to as parallel machine scheduling with time windows. The problem is a more general version of the Fixed Job Scheduling problem, involving a time window for each job larger than its processing time. The objective is to find the optimal subset of the jobs that can be processed. An interesting application area lies in Optimal Berth Allocation, which involves the assignment of vessels arriving at the port to appropriate berths within their time windows, while maximizing the total profit from the served vessels. Eligibility constraints are also taken into consideration. We develop an integer programming model for the problem. We show that the problem is NP‐hard, and develop a constraint‐graph‐based construction algorithm for generating near‐optimal solutions. We use genetic algorithm and other improvement algorithms to enhance the solution. Computational experimentation reveals that our algorithm generates very high quality solutions in very small computation times. Santrauka Nagrinėjama nepastovių darbų planavimo problema, susijusi su mašinų darbo planavimu laiko tarpais. Ši problema yra bendresnė fiksuotų darbų planavimo problemos versija, kai laiko tarpai ilgesni už darbų trukmę. Siekiama rasti optimalų atliekamų darbų poaibį. Geras pavyzdys yra optimali laivų priežiūra prieplaukoje, kai laivai prisišvartuoja laiko tarpais, taip maksimizuodami iš laivų gaunamą pelną. Taip pat įvertinami tinkamumo ribojimai. Problemai spręsti sukurtas sveikųjų skaičių programavimo modelis. Sukurtas apribojimų ir kreivių pavidalo algoritmas, gebantis generuoti apytikslius sprendinius. Jiems patikslinti naudojamas genetinis algoritmas ir kiti korekciniai algoritmai. Kompiuteriniai eksperimentai atskleidė, kad sukurtieji algoritmai generuoja labai tikslius sprendinius per labai trumpą laiko tarpą.