Neuro-genetic approach to optimize parameter design of dynamic multiresponse experiments

Engineers have widely applied the Taguchi method, a traditional approach for robust experimental design, to a variety of quality engineering problems for enhancing system robustness. However, the Taguchi method is unable to deal with dynamic multiresponse owing to increasing complexity of the product or design process. Although several alternative approaches have been presented to resolve this problem, they cannot effectively treat situations in which the control factors have continuous values. This study incorporates desirability functions into a hybrid neural network/genetic algorithm approach to optimize the parameter design of dynamic multiresponse with continuous values of parameters. The objective is to find the optimal combination of control factors to simultaneously maximize robustness of each response. The proposed approach is based on three stages which (1) use neural networks for constructing a response function model of a dynamic multiresponse system, (2) use exponential desirability functions for evaluating overall performance of a specific factor combination, and (3) use a genetic algorithm to optimize parameter design. Effectiveness of the proposed approach is illustrated with a simulated example. Analysis results reveal that the approach has higher performance than the traditional experimental design.     

Application of SVR,Taguchi loss function,and the artificial bee colony algorithm to resolve multiresponse parameter design problems: a case study on optimizing the design of a TIR lens

Multiresponse parameter design problems have become increasingly important and have received considerable attention from both researchers and practitioners since there are usually several quality characteristics that must be optimized simultaneously in most modern products/processes. This study applies support vector regression (SVR), Taguchi loss function, and the artificial bee colony (ABC) algorithm to develop a six-staged procedure that resolves these common and complicated parameter design problems. SVR is used to model the mathematical relationship between input control factors and output responses, and the ABC algorithm is used to find the optimal control factor settings by searching the well-constructed SVR models in which the Taguchi loss function is applied to evaluate the overall performance of a product/process. The feasibility and effectiveness of the proposed approach are demonstrated via a case study in which the design of a total internal reflection (TIR) lens is optimized while fabricating an MR16 light-emitting diode lamp. Experimental results indicate that the proposed solution procedure can provide highly robust design parameter settings for TIR lenses that can be directly applied in real manufacturing processes. Comparisons with the Taguchi method reveal that the Taguchi method is an undesirable and inappropriate method for resolving multiple-response parameter design problems, while the ABC algorithm can search the solution spaces in continuous domains modeled via SVR instead of in the limited discrete experiment levels, thus finding a more robust design than that obtained by the traditional analysis of variance. Consequently, the proposed integrated approach in this study can be considered feasible and effective and can be popularized as a useful tool for resolving general multiresponse parameter design problems in the real world.     

Improving the lighting performance of a 3535 packaged hi-power LED using genetic programming,quality loss functions and particle swarm optimization

The lighting performance of a 3535 packaged hi-power LED (light-emitting diode) is mainly influenced by its geometric design and the refractive properties of its materials. In the past, engineers often determined the settings of the geometric parameters and selected the refractive properties of the materials through a trial-and-error process based on the principles of optics and their own experience. This procedure was costly and time-consuming, and its use did not ensure that the settings of the design parameters were optimal. Therefore, this study proposed a hybrid approach based on genetic programming (GP), Taguchi quality loss functions, and particle swarm optimization (PSO) to solve the multi-response parameter design problems. The feasibility and effectiveness of the proposed approach was demonstrated by a case study on improving the lighting performance of an LED. The confirmation results showed that all of the key quality characteristics of an LED fulfill the required specifications, and the comparison found that the proposed hybrid approach outperforms the traditional Taguchi method in solving this multi-response parameter design problem. The proposed hybrid approach can be extended to solve parameter design problems with multiple responses in various application fields.     

基于Taguchi-BPNN-SEDEA的精密产品多元质量非参数稳健优化

针对设计参数不确定性和模型结构未知情形下精密产品多元质量波动问题,同时兼顾主体结构对轻量化设计要求,提出一种基于Taguchi-BPNN-SEDEA的多元质量非参数稳健优化方法.首先,通过正交试验设计和有限元分析获取多元质量数值,运用Taguchi方法将多元质量数值转化为信噪比来衡量精密产品稳健性;其次,运用BPNN非参数模型构建多元质量信噪比预测模型,以避免由参数模型设定导致的误差;在此基础上,提出改进的DEA基本模型,采用SEDEA非参数稳健优化方法,将设计参数不确定性下BPNN非参数模型求解问题转化为不确定性条件下复杂多属性决策问题;最后,通过实例表明,所提出的方法能够有效处理设计参数不确定性和模型结构未知并存情况下的多元质量稳健优化问题,从而验证该方法的可行性.     

Multi-attribute decision making for green electrical discharge machining

This paper aims to develop a combination of Taguchi and fuzzy TOPSIS methods to solve multi-response parameter optimization problems in green manufacturing. Electrical Discharge Machining (EDM), a commonly used non-traditional manufacturing process was considered in this study. A decision making model for the selection of process parameters in order to achieve green EDM was developed. An experimental investigation was carried out based on Taguchi L9 orthogonal array to analyze the sensitivity of green manufacturing attributes to the variations in process parameters such as peak current, pulse duration, dielectric level and flushing pressure. Weighing factors for the output responses were determined using triangular fuzzy numbers and the most desirable factor level combinations were selected based on TOPSIS technique. The model developed in this study can be used as a systematic framework for parameter optimization in environmentally conscious manufacturing processes.     

An integrated neuro-genetic approach incorporating the Taguchi method for product design

Product design is a multidisciplinary activity that requires the integration of concurrent engineering approaches into a design process that secures competitive advantages in product quality. In concurrent engineering, the Taguchi method has demonstrated an efficient design approach for product quality improvement. However, the Taguchi method intuitively uses parameters and levels in measuring the optimum combination of design parameter values, which might not guarantee that the final solution is the most optimal. This work proposes an integrated procedure that involves neural network training and genetic algorithm simulation within the Taguchi quality design process to aid in searching for the optimum solution with more precise design parameter values for improving the product development. The concept of fractals in computer graphics is also considered in the generation of product form alternatives to demonstrate its application in product design. The stages in the general approach of the proposed procedures include: (1) use of the Taguchi experimental design procedure, (2) analysis of the neural network and genetic algorithm process, and (3) generation of design alternatives. An electric fan design is used as an example to describe the development and explore the applicability of the proposed procedures. The results indicate that the proposed procedures could enhance the efficiency of product design efforts by approximately 7.8%. It is also expected that the proposed design procedure will provide designers with a more effective approach to product development.     

Experimental design with fuzzy levels

A novel technique is introduced for planning experimental design employing fuzzy rule-based systems. The significant aspect of the proposed experimental design with fuzzy levels (EDFL) is assigning a membership function for each level of variable factors. Consequently, the design matrix and observed responses can be represented in a set of fuzzy rules. Several examples are presented to clarify the proposed idea and the results are compared with the conventional Taguchi methodology. We have specifically planned an L18 orthogonal array EDFL for the solder paste printing stage of surface mount printed circuit board assembly to provide a model for the process and to optimize the selection of variable factors.     

A novel approach for optimizing the optical performance of the broadband tap coupler

An integrated approach using neural networks, exponential desirability functions and genetic algorithms to optimize parameter design problems with multiple responses is presented. The proposed approach aims to identify the input parameter settings to maximize the overall minimal satisfaction level with respect to all the responses. The proposed approach is illustrated by optimizing the fused process parameters created during fused biconic taper coupler development to improve the performance and reliability of a 1% (1199) single-window broadband tap coupler. The proposed solution procedure was implemented on a Taiwanese manufacturer of fibreoptic passive components. The implementation results demonstrate the practicability of the method. Comparison analysis revealed that the proposed procedure outperformed the traditional Taguchi method in resolving multi-response parameter design problems.     

Application of entropy measurement technique in grey based Taguchi method for solution of correlated multiple response optimization problems: A case study in welding

In the present work, an attempt has been made to apply an efficient technique, in order to solve correlated multiple response optimization problems, in the field of submerged arc welding. The traditional grey based Taguchi approach has been extended to tackle correlated multi-objective optimization problems. The Taguchi optimization technique is based on the assumption that the quality indices (i.e. responses) are independent or uncorrelated. But, in practical cases, the assumption may not be valid always. However, the common trend in the solution of multi-objective optimization problems is to initially convert these multi-objectives into an equivalent single objective function. While deriving this equivalent objective function, different priority weights are assigned to different responses, according to their relative importance. But, there is no specific guideline for assigning these response weights. In this context, the present study aims to apply the entropy measurement technique in order to calculate the relative response weights from the analysis of entropy of the entire process. Principal Component Analysis (PCA) has been adopted to eliminate correlation that exists among the responses and to convert correlated responses into uncorrelated and independent quality indices, called principal components. These have been accumulated to calculate the overall grey relational grade, using the theory of grey relational analysis. Finally, the grey based Taguchi method has been used to derive an optimal process environment capable of producing the desired weld quality. The previously mentioned method has been applied to optimize bead geometry parameters of submerged arc bead-on-plate weldment. The paper highlights a detailed methodology of the proposed technique and its effectiveness.     

Artificial neural network based approach for dynamic parameter design

The Taguchi parameter design method has been recognized as an important tool for improving the quality of a product or a process. However, the statistical methods and optimization procedures proposed by Taguchi have much room for improvement. For instance, the two-step procedure proposed by Taguchi may fail to identify an optimum design condition if an adjustment parameter does not exist, the optimal setting of a design parameter is determined only among the levels included in the parameter design experiment, and, for the dynamic parameter design, the signal parameter is assumed to follow a uniform rather than a general distribution. This paper develops an artificial neural network based dynamic parameter design approach to overcome the shortcomings of the Taguchi and existing alternative approaches. First, an artificial neural network is trained to map the relationship between the characteristic, design, noise and signal parameters. Second, Latin hypercube samples of the signal and noise parameters are obtained and used to estimate the slope between the signal parameter and characteristic as well as the variance of the characteristic at each set of design parameter settings. Then, the dynamic parameter design problem is formulated as a nonlinear optimization problem and solved to find the optimal settings of the design parameters using sequential quadratic programming. The effectiveness of the proposed approach is illustrated with an example.     

Optimization of parameter design: An intelligent approach using neural network and simulated annealing

Parameter design optimization problems have found extensive industrial applications, including product development, process design and operational condition setting. The parameter design optimization problems are complex because non-linear relationships and interactions may occur among parameters. To resolve such problems, engineers commonly employ the Taguchi method. However, the Taguchi method has some limitations in practice. Therefore, in this work, we present a novel means of improving the effectiveness of the optimization of parameter design. The proposed approach employs the neural network and simulated annealing, and consists of two phases. Phase 1 formulates an objective function for a problem using a neural network method to predict the value of the response for a given parameter setting. Phase 2 applies the simulated annealing algorithm to search for the optimal parameter combination. A numerical example demonstrates the effectiveness of the proposed approach.     

Control relevant identification for robust optimal control

This paper presents an approach to designing the input signal for an identification experiment, in which the process model estimate is to be used to formulate and solve for a robust (in a worst case sense) optimal controller. The input signal is designed to contain the information that is relevant for the end use of the model, that is for control purposes. The proposed approach uses sensitivity analysis to determine the input signal frequencies that are important with respect to a certain measure of achievable controller performance in conjunction with a frequency sampling filter model of the process. Based on the sensitivity analysis, an iterative experimental design methodology is suggested.     

混合自适应遗传算法优化多目标染色配方模型

从提高染色产品质量和效益的角度出发,综合考虑如染料浓度、温度、时间和助剂浓度等因素影响,构建了多目标染色工艺配方优化模型。针对传统遗传算法普遍存在的问题和缺陷,提出基于正交试验设计、自适应交叉操作及局部搜索等技术进行算法改进,并利用改进后的算法获得配方模型最优解的解决方法:。实践结果:证明,混合自适应遗传算法使种群更具有代表性和全面性,最大程度的继承了父代的优良特性,改善了算法的早熟现象并增强其寻优性能。最终以较少的计算量和较高的收敛速度对全局进行快速的搜索,比传统遗传算法得到的优化目标值降低了l0.8%左右。该方法:可推广应用于其他复杂过程的优化求解问题中。     

Optimization of extrusion blow molding processes using soft computing and Taguchi’s method

The objective of this study is to present a new numerical strategy using soft-computing techniques to determine the optimal die gap programming of extrusion blow molding processes. In this study, the design objective is to target a uniform part thickness after parison inflation by manipulating the parison die gap openings over time. To model the whole process, that is, the parison extrusion, the mould clamping and the parison inflation, commercial finite element software (BlowSim) from the National Research Council (NRC) of Canada is used. However, the use of such software is time-consuming and one important issue in a design environment is to minimize the number of simulations to get the optimal operating conditions. To do so, we proposed a new strategy called fuzzy neural–Taguchi network with genetic algorithm (FUNTGA) that establishes a back propagation network using a Taguchis experimental array to predict the relationship between design variables and responses. Genetic algorithm (GA) is then applied to search for the optimum design of die gap parison programming. As the number of training samples is greatly reduced due to the use of orthogonal arrays, the prediction accuracy of the neural network model is closely related to the distance between sampling points and the evolved designs. The extrapolation distance concept is proposed and introduced to GA using fuzzy rules to modify the fitness function and thus improving search efficiency. The comparison of the results with commercial optimization software from NRC demonstrates the effectiveness of the proposed approach.     

A synergistic Mahalanobis–Taguchi system and support vector regression based predictive multivariate manufacturing process quality control approach

The primary objective of this study is to propose and verify a new synergistic prediction-based multivariate process quality control (MPQC) approach for manufacturing processes. The proposed approach considers the influence of covariates (e.g. uncontrollable inputs) and output (or response) uncertainties to predict, monitor, diagnose, and adjust for out-of-control scenarios. The prediction-based real-time synergistic approach integrates off-line and on-line multivariate quality control strategies. In this approach, based on a current state prediction of responses, process control variables are adjusted to prevent any out-of-control or abnormal situations in the process. The unique approach is designed based on a Mahalanobis–Taguchi System (MTS), support vector regression (SVR), bootstrap prediction interval (PI), and derivative-free Nelder-Mead (NM) optimisation strategy. Two real-life case studies demonstrate the suitability of the proposed approach and show improvements in process performance. This easy-to-implement distribution-free predictive quality control approach provides the necessary flexibility to industry practitioners for real-life implementation in discrete or continuous manufacturing processes.     

Process Parameters Optimization: A Design Study for TiO$_{2}$ Thin Film of Vacuum Sputtering Process

This paper proposes a procedure for process parameters design by combining both modeling and optimization methods. The proposed procedure integrates the Taguchi method, the artificial neural network (ANN), and the genetic algorithm (GA). First, the Taguchi method is applied to minimize experimental numbers and to collect experimental data representing the quality performances of a system. Next, the ANN is used to build a system model based on the data from the Taguchi experimental method. Then, the GA is employed to search for the optimal process parameters. A process parameters design for a titanium dioxide (TiO₂) thin film in the vacuum sputtering process is studied in this paper. The quality objective is to form a smaller water contact angle on the TiO₂ thin-film surface. The water contact angle is 4° obtained from the system model of the proposed procedure. The process parameters obtained from the proposed procedure were used to conduct the experiment in the vacuum sputtering process for the TiO₂ thin film. The water contact angle given from the practical experiment is 3.93°. The difference percent is 1.75% between 4° and 3.93°. The result obtained from the system model of the proposed procedure is promising. Hence, we can conclude that the proposed procedure is a very good approach in solving the problem of the process parameters design.     

Solving a multiresponse simulation-optimization problem with discrete variables using a multiple-attribute decision-making method

The simulation model is a proven tool in solving nonlinear and stochastic problems and allows examination of the likely behavior of a proposed manufacturing system under selected conditions. However, it does not provide a method for optimization. A practical problem often embodies many characteristics of a multiresponse optimization problem. The present paper proposes to solve the multiresponse simulation-optimization problem by a multiple-attribute decision-making method—a technique for order preference by similarity to ideal solution (TOPSIS). The method assumes that the control factors have discrete values and that each control factor has exactly three control levels. Taguchi quality-loss functions are adapted to model the factor mean and variance effects. TOPSIS is then used to find the surrogate objective function for the multiple responses. The present paper predicts the system performances for any combination of levels of the control factors by using the main effects of the control factors according to the principles of a robust design method. The optimal design can then be obtained. A practical case study from an integrated-circuit packaging company illustrates the efficiency and effectiveness of the proposed method. Finally, constraints of the proposed method are addressed.     

Evolutionary Learning Processes to Design the Dilation-Erosion Perceptron for Weather Forecasting

The weather forecasting is considered a rather difficult problem due to many complex features present in these time series. Several techniques have been proposed in the literature to solve this problem. In particular, the dilation-erosion perceptron (DEP), a model whose foundations are based on mathematical morphology and complete lattice theory, has been successfully used for time series forecasting. However, a drawback arises from the gradient estimation of morphological operators in the classical gradient-based learning process of the DEP, since they are not differentiable of usual way. In this sense, this work presents evolutionary learning processes, using a modified genetic algorithm, a particle swarm optimization, a modified differential evolution and a covariance matrix adaptation evolutionary strategy, to design the DEP model for weather forecasting. In addition, into the proposed learning processes we have included an automatic correction step that is geared toward eliminating time phase distortions that occur in some weather phenomena. An experimental analysis is presented using three non-linear forecasting problems from the Brazilian weather, and the obtained results are discussed and compared, according to five well-known performance metrics and an evaluation function, to results found using the DEP model with its classical gradient-based learning process.     

Optimizing discrete stochastic systems using simulated annealing and simulation

The objective of this paper is to present an integrated approach of two models: simulation and optimization. This approach is used to determine the design parameters of stochastically constrained systems where the measure of performance is available only via simulation. The optimization model is solved using simulated annealing (SA) for parameter selection followed by the use of Monte Carlo simulation to evaluate the measure of performance. Based on the expected simulation output, the parameter set is either accepted or rejected. A modified rejection/acceptance criterion is presented for the proposed SA algorithm taking into consideration the stochastic system constraints. Moreover, a control variate is employed as a variance reduction technique in order to obtain an efficient estimate of performance measure. The proposed approach is tested using three real cases of the multi-echelon repairable item inventory systems (MERIIS). The results show that the proposed method is efficient in determining an optimal choice of spares and repair channels in these systems.     

A unified step response procedure for autotuning of PI controller and Smith predictor for stable processes

The step response autotuning of PI controller and Smith dead-time compensator (DTC), for stable processes, is studied in detail. A simple and effective procedure proposed is based on the first-order plus dead-time model, obtained from the step response by measuring the time to go to ten and sixty three percent of the steady-state value of the process output. The tuning formulae derived contain an adjustable parameter, with a clear meaning with respect to the performance/robustness, enabling the user to specify its value within a well defined range, in accordance with the expected range of process parameter variations and the controller used (PI or DTC). Comparison with recently proposed methods and experimental results presented confirm that high and consistent closed-loop system performance/robustness are obtained for a wide range of dynamic characteristics common to industrial processes.