Amos Ng

Evolutionary optimisation of noisy multi-objective problems using confidence-based dynamic resampling

Many real-world optimisation problems approached by evolutionary algorithms are subject to noise. When noise is present, the evolutionary selection process may become unstable and the convergence of the optimisation adversely affected. In this paper, we present a new technique that efficiently deals with noise in multi-objective optimisation. This technique aims at preventing the propagation of inferior solutions in the evolutionary selection due to noisy objective values. This is done by using an iterative resampling procedure that reduces the noise until the likelihood of selecting the correct solution reaches a given confidence level. To achieve an efficient utilisation of resources, the number of samples used per solution varies based on the amount of noise in the present area of the search space. The proposed algorithm is evaluated on the ZDT benchmark problems and two complex real-world problems of manufacturing optimisation. The first real-world problem concerns the optimisation of engine component manufacturing in aviation industry, while the second real-world problem concerns the optimisation of a camshaft machining line in automotive industry. The results from the optimisations indicate that the proposed technique is successful in reducing noise, and it competes successfully with other noise handling techniques.

SIMULATION-BASED DECISION SUPPORT FOR MANUFACTURING SYSTEM LIFE CYCLE MANAGEMENT

Previous research has highlighted the role of virtual engineering tools in the development of manufacturing machinery systems. Simulation models created for this purpose can potentially be used to provide support for other tasks, such as operational planning and service and maintenance. This requires that the simulation models can be fed with historic data as well as with snapshot data. Furthermore, the models must be able to communicate with other business software. The paper describes how simulation models can be used for operational production planning and for service and maintenance support. Benefits include a better possibility to verify production plans and the possibility to monitor and service manufacturing machinery from remote locations. Furthermore, the expanded and continuously updated models provide a good tool to study the effect of, for instance, planned new product introduction in existing manufacturing systems. The paper also presents directions for future research. One ambition is to add AI tools to the system so as to develop a semi-autonomous system for decision support.

A parallel surrogate-assisted multi-objective evolutionary algorithm for computationally expensive optimization problems

This paper presents a new efficient multi-objective evolutionary algorithm for solving computationally-intensive optimization problems. To support a high degree of parallelism, the algorithm is based on a steady-state design. For improved efficiency the algorithm utilizes a surrogate to identify promising candidate solutions and filter out poor ones. To handle the uncertainties associated with the approximative surrogate evaluations, a new method for multi-objective optimization is described which is generally applicable to all surrogate techniques. In this method, basically, surrogate objective values assigned to offspring are adjusted to consider the error of the surrogate. The algorithm is evaluated on the ZDT benchmark functions and on a real-world problem of manufacturing optimization. In assessing the performance of the algorithm, a new performance metric is suggested that combines convergence and diversity into one single measure. Results from both the benchmark experiments and the real-world test case indicate the potential of the proposed algorithm.

Industrial cost modelling and multi-objective optimisation for decision support in production systems development

Recent developments in cost modelling, simulation-based multi-objective optimisation, and post-optimality analysis have enabled the integration of costing data and cost estimation into a new methodology for supporting economically sound decision-making in manufacturing enterprises. Within this methodology, the combination of production engineering and financial data with multi-objective optimisation and post-optimality analysis has been proven to provide the essential information to facilitate knowledge-driven decision-making in real-world production systems development. The focus of this paper is to present the incremental cost modelling technique specifically designed for the integration with discrete-event simulation models and multi-objective optimisation within this methodology. A complete example, using the simulation model and data modified from a previous real-world case study, is provided in this paper to illustrate how the methodology and cost modelling are applied for the optimal investment decision support.

A Metamodel-Assisted Steady-State Evolution Strategy for Simulation-Based Optimization

Evolutionary algorithms (EAs) have proven to be highly useful for optimization of real-world problems due to their powerful ability to find near-optimal solutions of complex problems [8]. A variety of successful applications of EAs has been reported for problems such as engineering design, operational planning, and scheduling. However, in spite of the great success achieved in many applications, EAs have also encountered some challenges. The main weakness of using EAs in real-world optimization is that a large number of simulation evaluations are needed before an acceptable solution can be found. Typically, an EA requires thousands of simulation evaluations and one single evaluation may take a couple of minutes to hours of computing time. This poses a serious hindrance to the practical application of EAs in real-world scenarios, and to address this problem the incorporation of computationally efficient metamodels has been suggested, so-called metamodel-assisted EAs [11]. The purpose of metamodels is to approximate the relationship between the input and output variables of a simulation by computationally efficient mathematical models. If the original simulation is represented as

DISTRIBUTED VIRTUAL MANUFACTURING FOR DEVELOPMENT OF MODULAR MACHINE SYSTEMS

To support all phases of an agile modular manufacturing machine life cycle with CAE and Virtual Manufacturing tools, a number of different engineering applications (e.g. specialist software based tools) are typically used for design, simulation, analysis, programming, control and monitoring of a machine. These applications mainly exist today as small applications islands where each of them manages their own data. When a manufacturing machine is designed, simulated, programmed, analyzed, tested, or operated, the information, connected to that specific machine, used and generated by each application island is stored separately by each application. These application islands often use different storage technologies. Each one of the applications has an information structure to separate the information connected to each machine; however, they do not necessarily use the same information structure. Another issue concerning these applications is the functionality that is implemented in them to manage information; ...

OPTIMISE: An Internet-Based Platform for Metamodel-Assisted Simulation Optimization

Computer simulation has been described as the most effective tool for de-signing and analyzing systems in general and discrete-event systems (e.g., production or logistic systems) in particular (De ...

Advanced machine service support using Internet-enabled three-dimensional-based virtual engineering

In the era of globalization, one of the key factors for manufacturing machine builders/suppliers to remain competitive is their capability to provide cost-effective and comprehensive machine service and maintenance for their clients at anytime, anywhere. Previous research has highlighted the role of virtual engineering tools in the design and development life cycle of manufacturing machinery systems. Virtual engineering models created during the development phase can potentially be used to provide valuable functions for many other tasks during the operational phase, including service and maintenance support. This paper introduces an innovative Internet-enabled three-dimensional-based virtual engineering framework that can be used for such purposes. Specifically, it addresses a system architecture that is designed to facilitate the tight integration between virtual engineering tools and a set of Internet-based reconfigurable modular maintenance supporting tools. This system architecture has been verified by implementations using different toolsets atop of various Internet technologies (e.g. XML Web services and LabViews Datasocket). Implementation details and successful industrial-based test cases are also provided in this paper.