Combination Of Heuristics Research Articles

The effect of combining simple priority heuristics in flow-dominant shops

SUMMARY Previous scheduling research indicates that no single priority heuristic is effective on all the relevant performance criteria. However, few heuristics have shown to be effective on particular performance criteria. This study examines this issue further by applying various scheduling heuristics at different states of three flow-dominant shops (i.e. entrance, intermediate, and exit workstations), under a variety of experimental conditions. The experimental factors used are shop load, due date tightness, and shop structure. Twenty-eight different combinations of heuristics were investigated. The results indicate that the application of particular priority heuristics at various stages of the production process has a positive impact on the shop performance. By selectively utilizing different scheduling heuristics, it is possible to gain the advantages of each heuristic and at the same time minimize their associated shortcomings. Furthermore, the relative rankings of the heuristic combinations are fair...

Hierarchical control strategy for flotation columns

An integral solution to the control problem has to deal with problems associated with instrumentation, process modelling, control strategies and control algorithms. The term “hierarchical” control is used here to mean the intelligent integration of available knowledge, control algorithms and logic rules of decision. The selection of information, algorithms and rules is based on the experimental and theoretical knowledge of the column flotation process. The software support was designed and implemented in object oriented computer programmes. This technique allowed the implementation of a combination of control techniques, algorithms and heuristics, in the form of a control strategy. This strategy has shown to fit well with processes exhibiting complex dynamic characteristics. The hierarchical control was developed and tested in a pilot and an industrial flotation column including the following general aspects: coordination of distributed conventional control loops, verification and diagnosis of sensors, on-line predictive estimation of operating variables and diagnosis of common operating problems.

Joint performance of greedy heuristics for the integer knapsack problem

This paper analyzes the worst-case performance of combinations of greedy heuristics for the integer knapsack problem. If the knapsack is large enough to accomodate at least m units of any item, then the joint performance of the total-value and density-ordered greedy heuristics is no smaller than (m + 1) (m + 2) . For combinations of greedy heuristics that do not involve both the density-ordered and total-value greedy heuristics, the worst-case performance of the combination is no better than the worst-case performance of the single best heuristic in the combination.

Sequencing mixed-model assembly lines to level parts usage and minimize line length

Abstract The problem of sequencing units on a mixed-model assembly line can be viewed with several objectives in mind. Past research has focused mainly on two separate performance measures: (1) minimizing the length of the line (which is equivalent to minimizing the risk of stopping the conveyor when system variability is present and the station lengths are fixed); or (2) maintaining a rate of assembly equal to the demand rate for each model type in the production schedule. The latter is the more appropriate in a just-in-time environment. We present a bicriteria formulation of the problem that can be used to examine the tradeoffs between line length and parts usage. The resultant model takes the form of a mixed integer nonlinear program and is solved with a combination of heuristics and branch and bound. Results are reported for a wide range of problem sizes, as defined by the number of stations on the line, the number of different model types, and the total number of units to be assembled. In almost all ...

Recognition-based segmentation of on-line run-on handprinted words: Input vs. output segmentation

The performance of two methods for recognition-based segmentation of strings of on-line handprinted capital Latin characters is reported. The input strings consist of a time-ordered sequence of X, Y coordinates, punctuated by pen-lifts. The methods are designed to work in “run-on mode” where there is no constraint on the spacing between characters. While both methods use a neural network recognition engine and a graph-algorithmic post-processor, their approaches to segmentation are quite different. The first method, which we call INSEG (for input segmentation), uses a combination of heuristics to identify particular pen-lifts as tentative segmentation points. The second method, which we call OUTSEG (for output segmentation), relies on the empirically trained recognition engine for both recognizing characters and identifying relevant segmentation points. The best results are obtained with the INSEG method: 11% error on handprinted words from an 80,000 word dictionary.

An aggregation of PRO and CON evidence for medical decision support systems

One promising way to increase the classification accuracy of medical decision support systems is to implement heuristic combinations of pattern recognition and artificial intelligence tools. A parallel between “cognition” model and differential diagnostic task is sketched accentuating the aggregation of activating and restraining inputs and corresponding PRO and CON evidence in medicine. On the basis of this paradigm a trainable model of a fuzzy neuron is proposed which resembles some elements from the physician's decision process. An example from aviation medicine is presented which demonstrates the enhanced performance.

A generalized scheme for mapping parallel algorithms

A generalized mapping strategy that uses a combination of graph theory, mathematical programming, and heuristics is proposed. The authors use the knowledge from the given algorithm and the architecture to guide the mapping. The approach begins with a graphical representation of the parallel algorithm (problem graph) and the parallel computer (host graph). Using these representations, the authors generate a new graphical representation (extended host graph) on which the problem graph is mapped. An accurate characterization of the communication overhead is used in the objective functions to evaluate the optimality of the mapping. An efficient mapping scheme is developed which uses two levels of optimization procedures. The objective functions include minimizing the communication overhead and minimizing the total execution time which includes both computation and communication times. The mapping scheme is tested by simulation and further confirmed by mapping a real world application onto actual distributed environments. >

Computer aided molecular design: a novel method for optimal solvent selection

Abstract The task of selecting a solvent or solvent mixture with desirable combination of physical properties to meet the needs of specific applications, has largely been tackled using a combination of heuristics and costly experimental studies. This material selection problem is here formulated as the combinatorial molecular design problem of choosing a set of structural groups making up a target molecule with the desired properties as predicted by available group contribution techniques. A novel mixed-integer nonlinear programming (MINLP) technique is used to solve the problem yielding compounds with optimum value of an appropriate performance index, subject to material balances, process and design limitations and feasibility of molecular structures. The strategy is applied with excellent results to solvent design examples for liquid-liquid extraction and multicomponent gas absorption using varying combinations of objective functions and constraints to reflect directly a multiplicity of operational objectives.

Scheduling of manufacturing systems using the lagrangian relaxation technique

Scheduling is one of the most basic but the most difficult problems to be solved in the manufacturing industry. Generally, some degree of time-consuming and impractical enumeration is required to obtain optimal scheduling solutions. Industry has thus relied on a combination of heuristics and simulation to solve the problem, resulting in unreliable and often infeasible solutions. Yet, there is a real need for an improvement in scheduling methods in the complex and turbulent manufacturing environment. The logical strategy is to find scheduling methods which consistently generate good schedules efficiently. However, it is often difficult to measure the quality of a schedule without knowing the optimum.In this paper, the practical solution of three manufacturing scheduling problems are examined. As each problem is formulated, contraints are added or modified to reflect increasing real world complexity. The first problem considers scheduling singleoperation jobs on identical, parallel machines; the second problem is concerned with scheduling multiple-operation jobs with simple fork/join precedence constraints on identical, parallel machines; and, lastly, the job shop problem is considered, where multiple-operation jobs with general precedence constraints are scheduled on multiple machine types.In this paper, Lagrangian relaxation is used to decompose each problem into job- or operation-level scheduling subproblems, which are easier to solve than the original problem and have intuitive appeal. This technique then results in algorithms which generate near-optimal, feasible schedules efficiently, while giving a lower bound on the optimal cost. In addition, in day-to-day scheduling operations, the Lagrange multipliers from the last schedule can be used to initialize the algorithm, further reducing the computation time in generating a new schedule. All algorithms are demonstrated with examples using actual factory data. The consistent application of the Lagrangian relaxation technique to solve these problems demonstrates its advantages over scheduling methods currently used in industry. It is therefore believed that the methodology forms the basis of a new generation of scheduling systems that provide good schedules efficiently while effectively supporting management goals.

Polymer process design—I. Continuous production of chain growth homopolymers

A procedure for the systematic design of polymer production processes is described which uses a combination of heuristics and design estimates to choose among alternatives and to identify the most promising candidate designs. The procedure is described in detail here for free radical solution homopolymers. This approach gives rapid estimates of sensitivities to parameter values; this can be used to identify the critical data in the design. As examples, the design of continuous plants for the manufacture of polystyrene, poly(methyl methacrylate), poly(vinyl acetate) and polyethylene are considered; the best candidate designs are similar to several existing processes. The number and types of reactors are found to depend upon the operating conditions and the form of the final product. For purification, a flash is used at low viscosities. For further purification, the wiped film device is preferred when relatively large residence times are acceptable, but the extruder has a significant economic advantage when the time of exposure to high temperatures must be limited. Sensitivities are used to rank a variety of parameters as to their relative influence on the economics and the product quality.

A heuristic algorithm for loading new work on circuit pack assembly lines

SUMMARY This paper describes a scheduling algorithm used to suggest the sequence in which new products should be started on circuit pack assembly lines. For a real application many aspects, such as inaccuracy of data, incomplete information, variability and unpredictability of many events, etc., make it very difficult to have a realistic model which is analytically tractable. The scheduling approach developed here is an effort to capture the right details and to leave an allowance for the aspects not represented in the model. It shifts the emphasis from optimization with’ idealistic models to a conservative approach with more realistic models. A combination of heuristics are used to achieve the objectives of meeting weekly due dates, of keeping W1P, flow time and number of set-ups low while maintaining capacity. Due dates are given the highest priority. Within the limitations imposed by due dates, a weighted, cumulative load imbalance measure is used to balance the load on all operations. Set-up considerations determine the choice if this measure is close to (say, within 15% of) the minimum. This provides a simple, robust and reasonably sound solution to an enormously difficult problem.

A Comparison of Due Date Setting, Resource Assignment, and Job Preemption Heuristics for the Multiproject Scheduling Problem*

This research examines a heuristic, rule‐based approach for setting due dates in a multiproject, multijob, or assembly shop. Due date estimation is a challenging problem because the operating environment is capacitated, involves the allocation of multiple resources, and allows for the preemption of resources from one project or job to another. The dynamic, continuous arrival of new jobs or orders frequently results in the preemption of resources through the application of managerially determined priority policies. These preemption policies have a significant impact on the ultimate completion time of a job or a project.A three‐factor, full‐factorial computer simulation experiment is used to assess the relative effectiveness of combinations of four due date setting heuristics, five resource assignment heuristics, and three resource preemption heuristics. Recommendations are made for the selection of due date and resource assignment heuristic combinations under the three preemption policies examined.

The HASTUS Vehicle and Manpower Scheduling System at the Société de transport de la Communauté urbaine de Montréal

Varying service levels during the day and complex work rules make transit scheduling a difficult task. Over the last 10 years, a joint university-S.T.C.U.M.-private enterprise effort has produced HASTUS, a comprehensive set of computer tools that assists in building cost-effective schedules. The HASTUS-Macro decision support system uses linear programming methods to estimate the costs of changes in union contracts and service levels. HASTUS-Bus employs network-flow methods to generate optimal vehicle schedules. HASTUS-Micro generates operator assignments through a combination of specially-formulated heuristics and optimal matching algorithms. Annual savings in excess of $3,000,000 (three percent of operating costs) have been reported in manpower scheduling and $1,000,000 in vehicle scheduling. This success at the S.T.C.U.M. has led to installations in nearly 40 major cities worldwide.

Rule‐based mineral sequences in geochemical flow processes

AbstractThe infiltration of an aqueous solution into a permeable medium generally results in the dissolution of some of the minerals initially present and the possible precipitation of others. When the infiltration velocities are small, as is the case for many natural processes, conditions of local equilibrium apply and the dissolution and precipitation processes exhibit a wave‐type behavior reminiscent of chromatogrphic fronts. Zones of constant composition (state) will be separated by narrow regions within which the aqueous and solid phase concentrations exhibit sharp changes. Because of this wave‐like structure, an algebraic solution of the coupled material balance equations exists, but in a form that involves a trial and error solution procedure which has heretofore discouraged its application. This paper describes the essence of a scheme which uses a combination of graph theory and heuristics to minimize trials and thereby render the algebraic solution practical. The scheme offers an alternative to time‐sequencing solutions (e.g., finite difference) of the set of partial differential equations.

A study of combining heuristics for scheduling projects with limited multiple resources

The main objective of this study was to find a simple and quick procedure on the microcomputer for scheduling activities of a constrained multiple resource single project network that would minimize project duration. From this research there are two different types of results presented. First, a combination of simple heuristics which find the average of the minimum project durations for the constrained resource problem is presented. This combination not only supports the previous research on successful simple heuristic methods which set the priorities for constrained resource problems, but also produces results which are significantly better than those obtained by single heuristics. Second, a procedure for determining this combination of heuristics is introduced. A computer algorithm, COMAL, was developed for this study with constrained resource problems, but in the future its use may be expanded into other fields.

Simple and integrated heuristic algorithms for scheduling tasks with time and resource constraints

We consider the problem of scheduling a set of n tasks in a system having r resources. Each task has an arbitrary, but known, processing time and a deadline, and may request use of a number of resources. A resource can be used either in shared mode or exclusive mode. In this article, we study algorithms used for determining whether or not a set of tasks is schedulable in such a system, and if so, determining a schedule for it. This scheduling problem is known to be NP-complete and hence we methodically study a set of heuristics that can be used by such an algorithm. Due to the complexity of the problem, simple heuristics do not perform satisfactorily. However, an algorithm that uses combinations of these simple heuristics works very well compared to an optimal algorithm that takes exponential time complexity. For the combination that performs the best, we also determine the scheduling costs as a function of the size of the task set scheduled.

Preemptive Scheduling Under Time and Resource Constraints

We consider the problem of scheduling a set of n preemptable tasks in a system having r resources. Each task has an arbitrary, but known, worst case processing time and a deadline, and may request simultaneous use of a number of resources. A resource can be used either in shared mode or exclusive mode. In this paper, we develop and evaluate algorithms for determining whether or not a set of preemptive tasks is schedulable in such a real-time system, and if so, determining a schedule for it. This scheduling problem is known to be computationally intensive. In many real-time application environments, tasks are scheduled dynamically, and hence the scheduling algorithms used must have low run-time costs. To keep run-time costs low, we propose the use of suboptimal but practical algorithms that employ computationally simple heuristics. The computational complexity of our algorithms for scheduling n tasks in a system having r resources is O(rn <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), which is very much lower than that of known optimal algorithms. We report on the results of simulation studies performed on such heuristic preemptive scheduling algorithms and the sensitivity of the performance of the algorithms with respect to various scheduling parameters. These studies show that due to the complexity of the problem, straightforward heuristics do not perform satisfactorily. However, an algorithm that uses combinations of such heuristics in conjunction with limited backtracks works very well.

Large-Scale Production Planning

This paper discusses some principles of handling large-scale production planning problems. In practice a combination of optimization theory and heuristics is needed to develop a satisfactory approach. The main ideas are illustrated by two case studies: a power generator scheduling problem and a multi-item production planning problem.

A Combined Approach to the Pallet Loading Problem

Abstract In this paper the two-dimensional pallet loading problem is considered: that is, the problem of loading a rectangular pallet of size L by W, drawing from a set of n rectangular boxes. The objective is to maximize the area covered on the pallet by the boxes loaded. The problem is approached using a combination of dynamic programming and heuristics. The structured solutions resulting from the application of the dynamic program have two serendipitous characteristics: any item may be placed on the periphery of the pallet for easy access, and some control may be retained over the center of gravity of the pallet. Computational results are given.

The m-Center Problem: Minimax Facility Location

The m-Center Problem is to locate a given number of (emergency) facilities anywhere along a road network so as to minimize the maximum distance between these facilities and fixed demand locations assigned to them. Fundamental properties of the m-Center Problem are examined. The problem is modeled using integer programming, and is successfully attacked using a binary search technique and a combination of exact tests and heuristics. Computational results are given.