Traditional heuristic versus Hopfield neural network approaches to a car sequencing problem

This paper considers the problem of optimally sequencing different car models along an assembly line according to some contiguity constraints, while ensuring that the demands for each of the models are satisfied. This car sequencing problem (CSP) is a practical NP-hard combinatorial optimisation problem. The CSP is formulated as a nonlinear integer programming problem and it is shown that exact solutions to the problem are difficult to obtain due to the indefinite quadratic form of the CSP objective function. Two traditional heuristics (steepest descent and simulated annealing) are employed to solve the CSP approximately. Several Hopfield neural network (HNN) approaches are also presented. The process of mapping an optimisation problem onto a HNN is demonstrated explicitly, and modifications to the existing neural approaches are presented which guarantee feasibility of solutions. Further modifications are proposed to improve the solution quality by permitting escape from local minima in an attempt to locate the global optimum. Results from all solutions techniques are compared on a set of instances of the CSP, and conclusions drawn.     

Two local search approaches for solving real-life car sequencing problems

The NP-hard problem of car sequencing appears as the heart of the logistic process of many car manufacturers. The subject of the ROADEF’2005 challenge addressed a car sequencing problem proposed by the car manufacturer RENAULT, more complex than the academic problem generally addressed in the literature. This paper describes two local search approaches for this problem. In the first part, a new approach by very large-scale neighborhood search is presented. This approach, designed during the qualification stage preceding the final, is based on an original integer linear programming formulation. The second part is dedicated to the approach which enabled us to win the ROADEF’2005 challenge. Inspired by the latest works on the subject, this one is based on very fast explorations of small neighborhoods. Our contribution here is mainly algorithmic, in particular by showing how much exploiting invariants speeds up the neighborhood evaluation and contributes to the diversification of the search. Finally, the two approaches are compared and discussed through an extensive computational study on RENAULT’s benchmarks. The main conclusion drawn at this point is that sophisticated metaheuristics are useless to solve car sequencing problems. More generally, our victory on ROADEF’2005 challenge demonstrates that algorithmic aspects, sometimes neglected, remain the key ingredients for designing and engineering high-performance local search heuristics.     

A Multi-agent Based Self-adaptive Genetic Algorithm for the Long-term Car Pooling Problem

Rising vehicles number and increased use of private cars have caused significant traffic congestion, noise and energy waste. Public transport cannot always be set up in the non-urban areas. Car pooling, which is based on the idea that sets of car owners having the same travel destination share their vehicles has emerged to be a viable possibility to reduce private car usage around the world. In this paper, we present a multi-agent based self-adaptive genetic algorithm to solve long-term car pooling problem. The system is a combination of multi-agent system and genetic paradigm, and guided by a hyper-heuristic dynamically adapted by a collective learning process. The aim of our research is to solve the long-term car pooling problem efficiently with limited exploration of the search space. The proposed algorithm is tested using large scale instance data sets. The computational results show that the proposed method is competitive with other known approaches for solving long-term car pooling problem.     

An approximate moving boundary method for American option pricing

We present a method to solve the free-boundary problem that arises in the pricing of classical American options. Such free-boundary problems arise when one attempts to solve optimal-stopping problems set in continuous time. American option pricing is one of the most popular optimal-stopping problems considered in literature. The method presented in this paper primarily shows how one can leverage on a one factor approximation and the moving boundary approach to construct a solution mechanism. The result is an algorithm that has superior runtimes-accuracy balance to other computational methods that are available to solve the free-boundary problems. Exhaustive comparisons to other pricing methods are provided. We also discuss a variant of the proposed algorithm that allows for the computation of only one option price rather than the entire price function, when the requirement is such.     

Analysis and design of sequencing rules for car sequencing

This paper presents novel approaches for generating sequencing rules for the car sequencing (CS) problem in cases of two and multiple processing times per station. The CS problem decides on the succession of different car models launched down a mixed-model assembly line. It aims to avoid work overloads at the stations of the line by applying so-called sequencing rules, which restrict the maximum occurrence of labor-intensive options in a subsequence of a certain length. Thus to successfully avoid work overloads, suitable sequencing rules are essential. The paper shows that the only existing rule generation approach leads to sequencing rules which misclassify feasible sequences. We present a novel procedure which overcomes this drawback by generating multiple sequencing rules. Then, it is shown how to apply both procedures in case of multiple processing times per station. For both cases analytical and empirical results are derived to compare classification quality.     

Proportionate flow-shop scheduling with rejection

In many heavily loaded manufacturing systems, managers routinely make use of outsourcing options in order to maintain reasonable Quality of Service for customers. Thus, there is a strong need to provide tools for managers to economically coordinate sourcing and scheduling decisions. Our main aim is to provide such tools for an important set of flow-shop scheduling problems where rejection (outsourcing) is allowed and processing times are machine-independent. Our scheduling problems are essentially bicriteria problems, which combine a scheduling objective and the total outsourcing cost. We study several problems which differ according to the scheduling criterion considered. Moreover, each problem is divided into four different variations depending on the way the two criteria are dealt with. For example, in one variation the two criteria are aggregated into a single objective function; in two other variations the aim consists of minimizing one criterion subject to ensuring that the value of the other criterion will not exceed a predefined threshold. From a theoretical point of view, a computational complexity classification is provided for all variations of the problems under consideration. Moreover, optimization algorithms have been constructed to solve all problem variations, and approximation schemes have been developed for solving hard variations. Those schemes enable managers to solve large instances of hard variations while controlling the maximal gap between the obtained solution and the (unknown) optimal solution.     

Analysis of a Rollout Approach to Sequencing Problems with Stochastic Routing Applications

The paper considers sequencing problems, the traveling salesman problem being their natural representative. It studies a rollout approach that employs a cyclic heuristic as its main base algorithm. The theoretical analysis establishes that it is guaranteed to improve (at least in a weak sense) the quality of any feasible solution to a given sequencing problem. Besides other applications, the paper shows that it is well suited for applications that are embedded in dynamic and stochastic environments. The computational performance of the approach is investigated with applications to two stochastic routing problems. The dynamic version of the heuristic appears to be the first algorithm available in the literature to approximately solve a variant of one of these problems.     

Algebraic relations between the total least squares and least squares problems with more than one solution

Summary This paper completes our previous discussion on the total least squares (TLS) and the least squares (LS) problems for the linear systemAX=B which may contain more than one solution [12, 13], generalizes the work of Golub and Van Loan [1,2], Van Huffel [8], Van Huffel and Vandewalle [11]. The TLS problem is extended to the more general case. The sets of the solutions and the squared residuals for the TLS and LS problems are compared. The concept of the weighted squares residuals is extended and the difference between the TLS and the LS approaches is derived. The connection between the approximate subspaces and the perturbation theories are studied.It is proved that under moderate conditions, all the corresponding quantities for the solution sets of the TLS and the modified LS problems are close to each other, while the quantities for the solution set of the LS problem are close to the corresponding ones of a subset of that of the TLS problem.This work was financially supported by the Education Committee, People's Republic of China     

Integrated machine scheduling and vehicle routing with time windows

This paper integrates production and outbound distribution scheduling in order to minimize total tardiness. The overall problem consists of two subproblems. The first addresses scheduling a set of jobs on parallel machines with machine-dependent ready times. The second focusses on the delivery of completed jobs with a fleet of vehicles which may differ in their loading capacities and ready times. Job-dependent processing times, delivery time windows, service times, and destinations are taken into account. A genetic algorithm approach is introduced to solve the integrated problem as a whole. Two main questions are examined. Are the results of integrating machine scheduling and vehicle routing significantly better than those of classic decomposition approaches which break down the overall problem, solve the two subproblems successively, and merge the subsolutions to form a solution to the overall problem? And if so, is it possible to capitalize on these potentials despite the complexity of the integrated problem? Both questions are tackled by means of a numerical study. The genetic algorithm outperforms the classic decomposition approaches in case of small-size instances and is able to generate relatively good solutions for instances with up to 50 jobs, 5 machines, and 10 vehicles.     

Bounds in Multistage Linear Stochastic Programming

Multistage stochastic programs, which involve sequences of decisions over time, are usually hard to solve in realistically sized problems. Providing bounds for optimal solution may help in evaluating whether it is worth the additional computations for the stochastic program vs. simplified approaches. In this paper we generalize measures from the two-stage case, based on different levels of available information, to the multistage stochastic programming problems. A set of theorems providing chains of inequalities among the new quantities are proved. Numerical results on a case study related to a simple transportation problem illustrate the described relationships.     

A sweep-based algorithm for the fleet size and mix vehicle routing problem

This paper presents a new sweep-based heuristic for the fleet size and mix vehicle routing problem. This problem involves two kinds of decisions: the selection of a mix of vehicles among the available vehicle types and the routing of the selected fleet. The proposed algorithm first generates a large number of routes that are serviced by one or two vehicles. The selection of routes and vehicles to be used is then made by solving to optimality, in polynomial time, a set-partitioning problem having a special structure. Results on a set of benchmark test problems show that the proposed heuristic produces excellent solutions in short computing times. Having a fast but good solution method is needed for transportation companies that rent a significant part of their fleet and consequently can take advantage of frequent changes in fleet composition. Finally, the proposed heuristic produced new best-known solutions for three of the test problems; these solutions are reported.     

A methodology to solve large-scale cooperative transportation planning problems

In this paper, we suggest a methodology to solve a cooperative transportation planning problem and to assess its performance. The problem is motivated by a real-world scenario found in the German food industry. Several manufacturers with same customers but complementary food products share their vehicle fleets to deliver their customers. After an appropriate decomposition of the entire problem into sub problems, we obtain a set of rich vehicle routing problems (VRPs) with time windows for the delivery of the orders, capacity constraints, maximum operating times for the vehicles, and outsourcing options. Each of the resulting sub problems is solved by a greedy heuristic that takes the distance of the locations of customers and the time window constraints into account. The greedy heuristic is improved by an appropriate Ant Colony System (ACS). The suggested heuristics to solve the problem are assessed within a dynamic and stochastic environment in a rolling horizon setting using discrete event simulation. We describe the used simulation infrastructure. The results of extensive simulation experiments based on randomly generated problem instances and scenarios are provided and discussed. We show that the cooperative setting outperforms the non-cooperative one.     

Disrupted capacitated vehicle routing problem with order release delay

With the popularity of the just-in-time system, more and more companies are operating with little or no inventories, which make them highly vulnerable to delays on supply. This paper discusses a situation when the supply of the commodity does not arrive at the depot on time, so that not enough of the commodity is available to be loaded on all vehicles at the start of the delivery period. New routing plans need to be developed in such a case to reduce the impact the delay of supply may have on the distribution company. The resulting vehicle routing problem is different from other types of vehicle routing problems as it involves waiting and multiple trips. Two approaches have been developed to solve the order release delay problem, both of which involve a Tabu Search algorithm. Computational results show the proposed approaches can largely reduce the disruption costs that are caused by the delayed supply and they are especially effective when the length of delay is long.     

A fixed-point iteration approach for multibody dynamics with contact and small friction

Acceleration–force setups for multi-rigid-body dynamics are known to be inconsistent for some configurations and sufficiently large friction coefficients (a Painleve paradox). This difficulty is circumvented by time-stepping methods using impulse-velocity approaches, which solve complementarity problems with possibly nonconvex solution sets. We show that very simple configurations involving two bodies may have a nonconvex solution set for any nonzero value of the friction coefficient. We construct two fixed-point iteration algorithms that solve convex subproblems and that are guaranteed, for sufficiently small friction coefficients, to retrieve, at a linear convergence rate, the unique velocity solution of the nonconvex linear complementarity problem whenever the frictionless configuration can be disassembled. In addition, we show that one step of one of the iterative algorithms provides an excellent approximation to the velocity solution of the original, possibly nonconvex, problem if for all contacts we have that either the friction coefficient is small or the slip velocity is small.Subject Index. 70E55, 75M10, 75M15, 90C33A partial version of this work has appeared in the proceedings of the NATO Advanced Studies Institute on Virtual Nonlinear Multibody Systems, Prague, 2002.     

Pareto optimality and a class of set covering heuristics

The set covering problem has many diverse applications to problems arising in crew scheduling, facility location and other business areas. Since the problem is known to be hard to solve optimally, a number of approximate (heuristic) approaches have been designed for it. These approaches (with one exception) divide into two main groups, greedy heuristics and dual saturation heuristics. We use the concept of a Pareto optimal dual solution to show that an arbitrary dual saturation heuristic has the same worst-case performance guarantee as the two best known heuristics of that type. Moreover, this poor performance level is always attainable by those two heuristics.     

Car sequencing versus mixed-model sequencing: A computational study

The paper deals with the two most important mathematical models for sequencing products on a mixed-model assembly line in order to minimize work overload the mixed-model sequencing (MMS) model and the car sequencing (CS) model. Although both models follow the same underlying objective, only MMS directly addresses the work overload in its objective function. CS instead applies a surrogate objective using so-called sequencing rules which restrict labor-intensive options accompanied with the products in the sequence. The CS model minimizes the number of violations of the respective sequencing rules, which is widely assumed to lead to minimum work overload. This paper experimentally compares CS with MMS in order to quantify the gap in the solution quality between both models. The paper studies several variants of CS with different sequencing rule generation approaches and different objective functions from the literature as well as a newly introduced weighting factor. The performance of the different models is evaluated on a variety of random test instances. Although the objectives of CS and MMS are positively linearly correlated, results show that a sequence found by CS leads to at least 15% more work overload than a solution found by MMS. For none of the considered test instances and for none of the three different objective functions, CS is able to produce competitive results in terms of solution quality (work overload) compared to MMS. The results suggest that decision makers using CS should investigate whether MMS would lead to better sequencing orders for their specific instances.     

Unified encoding for hyper-heuristics with application to bioinformatics

This paper introduces a new approach to applying hyper-heuristic algorithms to solve combinatorial problems with less effort, taking into account the modelling and algorithm construction process. We propose a unified encoding of a solution and a set of low level heuristics which are domain-independent and which change the solution itself. This approach enables us to address NP-hard problems and generate good approximate solutions in a reasonable time without a large amount of additional work required to tailor search methodologies for the problem in hand. In particular, we focused on solving DNA sequencing by hybrydization with errors, which is known to be strongly NP-hard. The approach was extensively tested by solving multiple instances of well-known combinatorial problems and compared with results generated by meta heuristics that have been tailored for specific problem domains.     

A two-stage model for a day-ahead paratransit planning problem

We consider a dynamic planning problem for paratransit transportation. The focus is on a decision to take one day ahead: which requests to serve with own vehicles, and which requests to subcontract to taxis? We call this problem the day-ahead paratransit planning problem. The developed model is a non-standard two-stage integer recourse model. Both stages consist of two consecutive optimization problems: the clustering of requests into routes, and the assignment of these routes to vehicles. To solve this model, a genetic algorithm approach is used. Computational results are presented for randomly generated data sets.     

An efficient implementation of a VNS/ILS heuristic for a real-life car sequencing problem

The car sequencing problem consists in sequencing a given set of cars to be produced in a single day. We address one of the variants of this problem, in which the objective of minimizing the number of violations of assembly constraints has a stronger weight than the minimization of the number of paint color changes. We present and describe in details a VNS/ILS approach for approximately solving this problem. Computational results on real-life test instances are reported. The work presented in this paper obtained the second prize in the challenge ROADEF’2005 sponsored by Renault.     

Solving the open vehicle routeing problem via a single parameter metaheuristic algorithm

In this paper, we consider the open vehicle routeing problem (OVRP), in which routes are not sequences of locations starting and ending at the depot but open paths. The problem is of particular importance for planning fleets of hired vehicles, a common practice in the distribution and service industry. In such cases, the travelling cost is a function of the vehicle open paths. To solve the problem, we employ a single-parameter metaheuristic method that exploits a list of threshold values to guide intelligently an advanced local search. Computational results on a set of benchmark problems show that the proposed method consistently outperforms previous approaches for the OVRP. A real-world example demonstrates the applicability of the method in practice, demonstrating that the approach can be used to solve actual problems of routing large vehicle fleets.