Rodrigo Lankaites Pinheiro

A Variable Neighbourhood Search for the Workforce Scheduling and Routing Problem

The workforce scheduling and routing problem (WSRP) is a combinatorial optimisation problem where a set of workers must perform visits to geographically scattered locations. We present a Variable Neighbourhood Search (VNS) metaheuristic algorithm to tackle this problem, incorporating two novel heuristics tailored to the problem-domain. The first heuristic restricts the search space using a priority list of candidate workers and the second heuristic seeks to reduce the violation of specific soft constraints. We also present two greedy constructive heuristics to give the VNS a good starting point. We show that the use of domain-knowledge in the design of the algorithm can provide substantial improvements in the quality of solutions. The proposed VNS provides the first benchmark results for the set of real-world WSRP scenarios considered.

A genetic algorithm for a workforce scheduling and routing problem

The Workforce Scheduling and Routing Problem refers to the assignment of personnel to visits across various geographical locations. Solving this problem demands tackling scheduling and routing constraints while aiming to minimise the total operational cost. This paper presents a Genetic Algorithm (GA) tailored to tackle a set of real-world instances of this problem. The proposed GA uses a customised chromosome representation to maintain the feasibility of solutions. The performance of several genetic operators is investigated in relation to the tailored chromosome representation. This paper also presents a study of parameter settings for the proposed GA in relation to the various problem instances considered. Results show that the proposed GA, which incorporates tailored components, performs very well and is an effective baseline evolutionary algorithm for this difficult problem.

Analysis of Objectives Relationships in Multiobjective Problems Using Trade-Off Region Maps

Understanding the relationships between objectives in many-objective optimisation problems is desirable in order to develop more effective algorithms. We propose a technique for the analysis and visualisation of complex relationships between many (three or more) objectives. This technique looks at conflicting, harmonious and independent objectives relationships from different perspectives. To do that, it uses correlation, trade-off regions maps and scatter-plots in a four step approach. We apply the proposed technique to a set of instances of the well-known multiobjective multidimensional knapsack problem. The experimental results show that with the proposed technique we can identify local and complex relationships between objectives, trade-offs not derived from pairwise relationships, gaps in the fitness landscape, and regions of interest. Such information can be used to tailor the development of algorithms.

Extended decomposition for mixed integer programming to solve a workforce scheduling and routing problem

We propose an approach based on mixed integer programming (MIP) with decomposition to solve a workforce scheduling and routing problem, in which a set of workers should be assigned to tasks that are distributed across different geographical locations. We present a mixed integer programming model that incorporates important real-world features of the problem such as defined geographical regions and flexibility in the workers’ availability. We decompose the problem based on geographical areas. The quality of the overall solution is affected by the ordering in which the sub-problems are tackled. Hence, we investigate different ordering strategies to solve the sub-problems. We also use a procedure to have additional workforce from neighbouring regions and this helps to improve results in some instances. We also developed a genetic algorithm to compare the results produced by the decomposition methods. Our experimental results show that although the decomposition method does not always outperform the genetic algorithm, it finds high quality solutions in practical computational times using an exact optimization method.

Towards an Efficient API for Optimisation Problems Data

The literature presents many application programming interfaces (APIs) and frameworks that provide state of the art algorithms and techniques for solving optimisation problems. The same cannot be said about APIs and frameworks focused on the problem data itself because with the peculiarities and details of each variant of a problem, it is virtually impossible to provide general tools that are broad enough to be useful on a large scale. However, there are benefits of employing problem-centred APIs in a R&D environment: improving the understanding of the problem, providing fairness on the results comparison, providing efficient data structures for different solving techniques, etc. Therefore, in this work we propose a novel design methodology for an API focused on an optimisation problem. Our methodology relies on a data parser to handle the problem specification files and on a set of efficient data structures to handle the information on memory, in an intuitive fashion for researchers and efficient for the solving algorithms. Also, we present the concepts of a solution dispenser that can manage solutions objects in memory better than built-in garbage collectors. Finally, we describe the positive results of employing a tailored API to a project involving the development of optimisation solutions for workforce scheduling and routing problems.

A Variable Neighbourhood Search for Nurse Scheduling with Balanced Preference Satisfaction

The nurse scheduling problem (NSP) is a combinatorial optimisation problem widely tackled in the literature. Recently, a new variant of this problem was proposed, called nurse scheduling problem with balanced preference satisfaction (NSPBPS). This paper further investigates this variant of the NSP as we propose a new algorithm to solve the problem and obtain a better balance of overall preference satisfaction. Initiall, the algorithm converts the problem to a bottleneck assignment problem and solves it to generate an initial feasible solution for the NSPBPS. Posteriorly, the algorithm applies the Variable Neighbourhood Search (VNS) metaheuristic using two sets of search neighbourhoods in order to improve the initial solution. We empirically assess the performance of the algorithm using the NSPLib benchmark instances and we compare our results to other results found in the literature. The proposed VNS algorithm exhibits good performance by achieving solutions that are fairer (in terms of preference satisfaction) for the majority of the scenarios.

A Development and Integration Framework for Optimisation-based Enterprise Solutions

The operational research literature often presents papers result of collaborative work between universities and third parties. However, despite the popularity of these scenarios, to the best of our knowledge no software development methodology or framework can be found in the literature to aid the development of decision support systems in such environment. Therefore we propose a methodological framework intended to aid on the communication between academics and information system developers, integration of the information management system and the decision support module, and to provide a better development environment for the academics. Moreover, the framework aims to provide liberty for academics and developers to work as they consider better, hence not restraining anyone. The framework is divided into three main components: the building of a data model, a data extractor and validator and a solution visualization and auxiliary platform. These elements directly or indirectly improve and collaborate to the research and the development of the decision support system. We also present the application of the framework to a real scenario and the positive outcome obtained.

Using Goal Programming on Estimated Pareto Fronts to Solve Multiobjective Problems.

Modern multiobjective algorithms can be computationally inefficient in producing good approximation sets for highly constrained many-objective problems. Such problems are common in real-world applications where decision-makers need to assess multiple conflicting objectives. Also, different instances of real-world problems often share similar fitness landscapes because key parts of the data are the same across these instances. We we propose a novel methodology that consists of solving one instance of a given problem scenario using computationally expensive multiobjective algorithms to obtain a good approximation set and then using Goal Programming with efficient single-objective algorithms to solve other instances of the same problem scenario. We propose three goal-based objective functions and show that on a real-world home healthcare planning problem the methodology can produce improved results in a shorter computation time.

An application programming interface with increased performance for optimisation problems data

An optimisation problem can have many forms and variants. It may consider different objectives, constraints, and variables. For that reason, providing a general application programming interface (A...

An Evolutionary Algorithm for Graph Planarisation by Vertex Deletion

A non-planar graph can only be planarized if it is structurally modified. This work presents a new heuristic algorithm that uses vertices deletion to modify a non-planar graph in order to obtain a planar subgraph. The proposed algorithm aims to delete a minimum number of vertices to achieve its goal. The vertex deletion number of a graph G = (V, E) is the smallest integer k � 0 such that there is an induced planar subgraph of G obtained by the removal of k vertices of G. Considering that the corresponding decision problem is NP-complete and an approximation algorithm for graph planarization by vertices deletion does not exist, this work proposes an evolutionary algorithm that uses a constructive heuristic algorithm to planarize a graph. This constructive heuristic has time complexity of O(n + m), where m = |V| and n = |E|, and is based on the PQ-trees data structure and on the vertex deletion operation. The algorithm performance is verified by means of case studies.