James Curry

Parallel ant colony for nonlinear function optimization with graphics hardware acceleration

This paper presents a massively parallel Ant Colony Optimization — Pattern Search (ACO-PS) algorithm with graphics hardware acceleration on nonlinear function optimization problems. The objective of this study is to determine the effectiveness of using Graphics Processing Units (GPU) as a hardware platform for ACO-PS. GPU, the common graphics hardware found in modern personal computers, can be used for data-parallel computing in a desktop setting. In this research, the classical ACO is adapted in the data-parallel GPU computing platform featuring ‘Single Instruction — Multiple Thread’ (SIMT). The global optimal search of the ACO is enhanced by the classical local Pattern Search (PS) method. The hybrid ACO-PS method is implemented in a GPU+CPU hardware platform and compared to a similar implementation in a Central Processing Unit (CPU) platform. Computational results indicate that GPU-accelerated SIMT-ACO-PS method is orders of magnitude faster than the corresponding CPU implementation. The main contribution of this paper is the parallelization analysis and performance analysis of the hybrid ACO-PS with GPU acceleration.

SIMD tabu search for the quadratic assignment problem with graphics hardware acceleration

This paper presents a single instruction multiple data tabu search (SIMD-TS) algorithm for the quadratic assignment problem (QAP) with graphics hardware acceleration. The QAP is a classical combinatorial optimisation problem that is difficult to solve optimally for even small problems with over 30 items. By using graphic hardware acceleration, the developed SIMD-TS algorithm executes 20 to 45 times faster than traditional CPU code. The computational improvement is made possible by the utilisation of the parallel computing capability of a graphics processing unit (GPU). The speed and effectiveness of this algorithm are demonstrated on QAP library problems. The main contribution of this paper is a fast and effective SIMD-TS algorithm capable of producing results for large QAPs on a desktop personal computer equivalent to the results achieved with a CPU cluster.

Advanced tutorial—simulation-based scheduling and control

This paper presents a simulation control system developed by the Texas A&M Computer Aided Manufacturing Laboratory (TAMCAM). This control system is a research tool that TAMCAM uses to explore the advantages and disadvantages of on-line simulation for process control. The simulation control system is developed directly from information about the shop floor stored in a relational database. Processing times and fallout rates are also estimated directly from external data sources. The system is implemented in Arena, Microsoft Access, and Microsoft Visual C++. The advanced tutorial will demonstrate the software tools described in this paper.

Particle Swarm with graphics hardware acceleration and local pattern search on bound constrained problems

This paper presents a Particle Swarm - Pattern Search Optimization (PS2) algorithm with graphics hardware acceleration for bound constrained nonlinear optimization problems. The objective of this study is to determine the effectiveness of using Graphics Processing Units (GPU) as a hardware platform for Particle Swarm Optimization (PSO). GPU, the common graphics hardware which can be found in many personal computers, can be used for desktop data-parallel computing. The classical PSO is adapted in the data-parallel GPU computing platform featuring ‘Single Instruction - Multiple Thread’ (SIMT). PSO is also enhanced by adding a local Pattern Search (PS) improvement. The hybrid PS2 optimization method is implemented in the GPU environment and with a Central Processing Unit (CPU) in a PC. Computational results indicate that GPU-accelerated SIMT-PS2 method is orders of magnitude faster than the corresponding CPU implementation. The main contribution of this paper is the parallelization analysis and performance analysis of the hybrid PS2 with GPU acceleration.

Rescheduling parallel machines with stepwise increasing tardiness and machine assignment stability objectives

Nervousness in machine assignments during rescheduling can cause problems for the implementation of a scheduling system. This paper examines rescheduling due to the arrival of new jobs to the system. Parallel machine scheduling problems with stepwise increasing tardiness cost objectives, non-zero machine ready times, constraints that limit machine reassignments, and machine reassignment costs are considered. Simulation experiments and individual scheduling problems indicate that nervousness can be controlled at a low cost in some parallel machine scheduling environments. The rescheduling problems in the simulation are solved with a branch-and-price algorithm. Significant gains in schedule stability can be achieved by selecting the alternative optimal solution with the fewest machine reassignments.

Multi-walk Parallel Pattern Search Approach on a GPU Computing Platform

This paper studies the efficiency of using Pattern Search (PS) on bound constrained optimization functions on a Graphics Processing Unit (GPU) computing platform. Pattern Search is a direct search optimization technique that does not require derivative information on non-linear programming problems. Pattern Search is ideally suited to a GPU computing environment due to its low memory requirement and no communication between threads in a multi-walk setting. To adapt to a GPU environment, traditional Pattern Search is modified by terminating based on iterations instead of tolerance. This research designed and implemented a multi-walk Pattern Search algorithm on a GPU computing platform. Computational results are promising with a computing speedup of 100+ compared to a corresponding implementation on a single CPU.

Experimental Investigation of the Tool Wear and Tool Life in Micro Hard Milling

The objective of this project is to experimentally investigate the influence of Minimum Quantity Lubrication (MQL) on tool wear and tool life in micro hardmilling. The experiments were performed on stainless steel using uncoated WC micro-mill with the nominal diameter of 508 microns. The tool wear is characterized by the volume of the material loss at the tool tip. In order to reveal the progression of the tool wear, the worn tool was examined periodically under SEM after a fixed amount of workpiece material removal (1.25 mm3 or 5 slots in this study). The tool life was characterized as the amount of material removed, instead of the conventional cutting times. The feedrate and the spindle speed were fixed, and two levels of axial depth of cut (50 and 75 microns) were compared. The higher depth of cut leads to longer tool life. The machining performance under MQL is superior to the dry machining for both process conditions in terms of the tool life. The cutting forces in feed direction and the surface roughness at the bottom of the slots were also examined during the experiments. The magnitude of the machining forces showed cyclic pattern for both MQL and dry machining. The SEM images and the cutting force signals suggested that the dominant mode of the tool wear in micro-milling is edge chipping and abrasive wear at the tool tip. The loss of the micro-grain of WC at the cutting edge leads to edge chipping, which reduces the effective cutting diameter; the abrasive wear enlarge the edge radius, causing the cutting force increase. As the cutting edge radius reaches a certain dimension, the whole edge was stripped off, a new edge formed with a smaller edge radius, and the cycle restarts. Under MQL cutting conditions, three cycles were observed before tool failure, while under dry machining conditions, the tool only experienced two cycles before tool breakage. The surface roughness at the bottom of the slots improved significantly with the application of MQL for all levels of the tool wear. The surface roughness did not increase drastically as the tool wear increased. It reached a plateau after the tool wear went into gradual wear state. Further experiments and theoretical analysis will be pursued in the future to gain a deeper understanding of tool wear mechanism in micro-milling.Copyright

Prototype software model for designing intruder detection systems with simulation

This article explores using discrete-event simulation for the design and control of defence oriented fixed-sensor- based detection system in a facility housing items of significant interest to enemy forces. The key issues discussed include software development, simulation-based optimization within a modeling framework, and the expansion of the framework to create real-time control tools and training simulations. The software discussed in this article is a flexible simulation environment where the data for the simulation are stored in an external database and the simulation logic is being implemented using a commercial simulation package. The simulation assesses the overall security level of a building against various intruder scenarios. A series of simulation runs with different inputs can determine the change in security level with changes in the sensor configuration, building layout, and intruder/guard strategies. In addition, the simulation model developed for the design stage of the project can be modified to produce a control tool for the testing, training, and real-time control of systems with humans and sensor hardware in the loop.

Sequence Dependent Parallel Machine Scheduling Using Parallel Ant Colony Optimization With Graphics Hardware Acceleration

This paper studies the effectiveness of using parallel Ant Colony Optimization for sequence dependent parallel machine scheduling on a Graphics Processing Unit (GPU) hardware platform. Parallel machine scheduling is a traditional NP-hard combinatorial optimization problem. In this research, a hybrid ant colony optimization method that combines the ‘Apparent Tardiness Cost with Setups’ (ATCS) dispatching rule with massive ants is proposed to solve the parallel machine scheduling problem quickly and efficiently. The computational results demonstrate that the proposed method is effective and solve the problems order of magnitude faster with a GPU accelerated implementation.Copyright

Using a natural language generation approach to document simulation results

Simulation experiments generate large data sets that must be converted into recommendations for decision makers. This article explores using a Natural Language Generation (NLG) approach for writing summaries of simulation experiments. The article discusses the steps required to convert simulation experiment data to text and highlights the unique aspects of data to text for simulation experiments. Automation of report generation can potentially reduce the time and cost of simulation studies and improve the reporting of results. A prototype software system was developed and applied to a simulation to illustrate the benefits of a NLG approach.