The optimisation deep-draw clearance design for deep-draw dies

In this paper, the modelling of deep-drawing processing using neural networks is established. The relationships between process parameter (material thickness, punch diameter, die-cavity diameter and materials-clearance ratio) and deep-drawing performance (the dimensional error of diameter and cylinder) are created, based on a neural network. A simulated annealing (SA) optimisation algorithm with a performance index is then applied to the neural network to search for the optimal design parameters of the drawing-die. Experimental results have shown that deep-drawing performance can be enhanced by using this approach.     

An empirical approach to modelling fluid dispensing for electronic packaging

Fluid dispensing is a popular process in semiconductor manufacturing industry which is commonly used in die-bonding as well as microchip encapsulation for electronic packaging. Modelling the fluid dispensing process is important to understand the process behaviour as well as determine optimum operating conditions of the process for a high-yield, low cost and robust operation. Previous studies of fluid dispensing mainly focus on the development of analytical models. However, an analytical model for fluid dispensing, which can provide accurate results, is very difficult to develop because of the complex behaviour of fluid dispensing and high degree of uncertainties of the process in a real world environment. In this project, an empirical approach to modelling fluid dispensing was attempted. Two common empirical modelling techniques, statistical regression and neural networks, were introduced to model fluid dispensing process for electronic packaging. Development of neural network based process models using genetic algorithm (GA) and Levenberg−Marquardt algorithm are presented. Validation tests were performed to evaluate the effectiveness of the developed process models from which a multiple regression model and a GA trained neural network with the architecture of 3-15-1 were identified to be the process models of the fluid dispensing respectively for the encapsulation weight and encapsulation thickness.     

Process optimisation of transfer moulding for electronic packages using artificial neural networks and multiobjective optimisation techniques

Transfer moulding is the most common process for the encapsulation of electronic packages in semiconductor manufacturing. Quality of the moulding is affected by a large number of mould design parameters and process parameters. Currently, the parameters setting is performed by experienced engineers in a trial and error manner and often the optimal setting can not be obtained. In the face of global competition, the current practice is inadequate. In this research, a process optimisation system for transfer moulding of electronic packages is described which involves design of experiments (DOE) techniques, artificial neural networks (ANNs), multiple regression analysis and the minimax method. The system is aimed to determine the optimal mould design parameters and process parameter settings of transfer moulding of electronic packages for multiobjective problem. Implementation of the optimisation system has demonstrated that the time for the determination of optimal mould design parameters and process parameters setting can be greatly reduced and the parameters setting recommended by the system can contribute to the good quality of moulded packages without relying on experienced engineers.     

The process modelling of epoxy dispensing for microchip encapsulation using fuzzy linear regression with fuzzy intervals

Epoxy dispensing is a popular way to perform microchip encapsulation for chip-on-board (COB) packages. However, the determination of the proper process parameters setting for a satisfactory encapsulation quality is difficult due to the complex behaviour of the encapsulant during the dispensing process and the inherent fuzziness of epoxy dispensing systems. Sometimes, the observed values from the process may be irregular. In conventional regression models, deviations between the observed values and the estimated values are supposed to have a probability distribution. However, when data is scattered, the obtained regression model has too wide of a possibility range. These deviations in processes such as epoxy dispensing can be regarded as system fuzziness that can be dealt with satisfactorily using a fuzzy regression method. In this paper, the fuzzy linear regression concept with fuzzy intervals and its application to the process modelling of epoxy dispensing for microchip encapsulation are described. Two fuzzy regression models, expressing the correlation between various process parameters and the two quality characteristics, respectively, were developed. Validation experiments were performed to demonstrate the effectiveness of the method for process modelling.     

Structure optimisation and statistical modelling of technological machine systems

A method aimed at the optimisation of the technological process structure and the layout diagram for machining systems based on the optimum variants guided search using designer-computer dialogue has been considered. Mathematical models and results of modelling various machining systems built in accordance with building-block approach are also discussed. The method by which the calculation of productivity and reliability of lines of complex structure can be achieved, based on the generalisation of the modelling, data, and which has been applied to the solution of practical optimisation problems is also presented.     

A cost-tolerance model for process sequence optimisation

This paper concerns the modelling of cost-tolerance data for various manufacturing processes and the optimisation of process sequences based on minimum production cost. A natural spline model representing the cost-tolerance relationship is introduced. A methodology to optimise the process sequences is developed using an expert system approach. An example to illustrate the methodology and the optimisation model is presented     

Modelling and optimisation of laser shock peening using an integrated simulated annealing-based method

Laser shock peening is an innovative surface treatment technique, which has been successfully applied to improve fatigue performance of metallic components. Laser shock peening improves the surface morphology and microstructure of the material. In this paper, three Nd3+:YAG laser process parameters (voltage, focus position and pulse duration) are varied in an experiment, in order to determine the optimal process parameters that could simultaneously meet the specifications for seven correlated responses of processed Nimonic 263 sheets. The modelling and optimisation of a process were performed using the advanced, problem-independent method. First, responses are expressed using Taguchi’s quality loss function, followed by the application of multivariate statistical methods to uncorrelate and synthesise them into a single performance measure. Then, artificial neural networks are used to build the process model, and simulated annealing was utilised to find the optimal process parameters setting in a global continual space of solutions. Effectiveness of the proposed method in the development of a robust laser shock peening was proved in comparison to several commonly used approaches from the literature, resulting in the highest process performance measure, the most favourable response values and the corresponding process parameters optimum. Besides the improved surface characteristics, the optimised laser shock peening (LSP) showed an improvement in terms of microhardness and formation of favourable microstructural transformations.     

A metamodel using neural networks and genetic algorithms for an integrated optimal design of mechanisms

This work examines the possibility of using genetic algorithms and some neural networks to optimise mechanisms. A detailed example shows that using this stochastic method along with neural networks is very efficient. We can thus speak of a metamodel for optimisation in the context of integrated design .     

Towards the integration of process and production planning: an optimisation model for cutting parameters

This paper aims to contribute to process and production planning integration through the development of a new cutting parameters optimisation model. The developed model considers simultaneously the technology-related constraints and a shop floor constraint determined by the available time at each workstation. The latter, being a constraint related to the part machining time, is associated with the set of all elementary machining operations and implies the development of a new multi-operation optimisation model. In this approach, part machining time is a new variable for shop floor scheduling. Since the limiting factor of workstation available time at every scheduling date depends on the shop floor status, optimum part machining time can range from the time for minimum cost to the time for maximum production rate. The introduction of the available time in the optimisation process allows for the generation of improved schedules according to several performance measures. The proposed optimisation model is non-linear, uni-criterion and multi-variable. The search of the optimal solution is carried out using sequential quadratic programming.     

Predicting product quality with backpropagation: A thermoplastic injection moulding case study

This application uses live data from a thermoplastic injection moulding manufacturer to examine the feasibility and effectiveness of using backpropagation artificial neural networks for predictive quality control. Preprocessing and post processing of live data, formulating neural predictive strategies, selecting architecture and parameters, and handling of temporal aspects are topics. Performance of the neural networks are compared to other quality control methods, including control charts and statistical techniques. This case study demonstrates that even manufacturers who have modest expertise in computing and limited hardware and software availability can successfully use neural networks for data analysis and modelling.