Feature selection techniques and comparative studies for large-scale manufacturing processes

In order to efficiently and effectively control an overall process in the process industry, a few important parameters should be identified from high-dimensional, non-linear, and correlated data. Feature selection techniques can be employed to extract a subset of process parameters relevant to product quality. The performance of these techniques depends on the precision of the prediction model formulated to quantify the relationship between the process parameters and the quality characteristics. Although the neural network-based partial least squares (NNPLS) method has been proven to be effective in prediction models for the aforementioned industrial process data, feature selection techniques appropriate for NNPLS models have yet to appear. Here, several techniques for scoring the relevance of process parameters to product quality are proposed and validated by applying three datasets. These experiments show that the proposed techniques can discriminate relevant process parameters from irrelevant ones.     

Application of multiple regression and adaptive neuro fuzzy inference system for the prediction of surface roughness

A manufacturing system is oriented towards higher production rate, quality, and reduced cost and time to make a product. Surface roughness is an index for determining the quality of machined products and is influenced by the cutting parameters. Surface roughness prediction in machining is being attempted with many methodologies, yet there is a need to develop robust, autonomous and accurate predictive system. This work proposes the application of two different hybrid intelligent techniques, adaptive neuro fuzzy inference system (ANFIS) and radial basis function neural network- fuzzy logic (RBFNN-FL) for the prediction of surface roughness in end milling. An experimental data set is obtained with speed, feed, depth of cut and vibration as input parameters and surface roughness as output parameter. The input-output data set is used for training and validation of the proposed techniques. After validation they are forwarded for the prediction of surface roughness. Both the hybrid techniques are found to be superior over their respective individual intelligent techniques in terms of computational speed and accuracy for the prediction of surface roughness.     

Online batch recipe adjustments for product quality control using empirical models: application to a nylon-6,6 process

The application of batch profile characterization tools to enhance process understanding by uncovering the signature of the primary disturbances on the profiles and its effect on the product quality is illustrated on a nylon-6,6 process. The historical profile data for the fixed recipe operation are systematically studied to understand the primary disturbances affecting the process, and it is shown that good online predictions of the final product quality are possible much before the completion of the batch from the available measurement profiles. A simple online recipe adjustment strategy based on the predicted quality deviation from the target is proposed. Results show that the recipe adjustments significantly reduce the variation in the final product quality. Issues in the use of empirical prediction models from recipe-based data are discussed.     

多传感器融合下多工况刀具磨损状态预测的深度森林方法研究

准确监测加工过程刀具磨损状态有助于避免因刀具失效导致的产品质量问题。 建立不同工况的刀具磨损监测模型,往 往需要对每组工况调参以保证精度。 为减少调参并保证预测精度,结合深度森林的超参数少、参数对模型不敏感和训练过程自 适应等优点,利用深度森林建立了多传感器信号及多工况下自主特征选择的刀具磨损状态预测模型。 基于 3 组不同工艺参数 下 TC18 铣削过程的多传感器及磨损数据,以及预测与健康管理(PHM)学会 2010 年高速数控机床刀具健康预测竞赛的开放数 据,深度森林在 3 组工况的预测精度分别为 95. 35% 、96. 63% 和 97. 06% ,在 PHM 数据上为 98. 95% ,验证了深度森林对多工况 下刀具磨损预测的高精度和适用性,为在线监测技术提供了有力的指导。     

On the finite element modelling of high speed hard turning

The results reported in this paper pertain to the simulation of high speed hard turning when using the finite element method. In recent years high speed hard turning has emerged as a very advantageous machining process for cutting hardened steels. Among the advantages of this modern turning operation are final product quality, reduced machining time, lower cost and environmentally friendly characteristics. For the finite element modelling a commercial programme, namely the Third Wave Systems AdvantEdge, was used. This programme is specially designed for simulating cutting operations, offering to the user many designing and analysis tools. In the present analysis orthogonal cutting models are proposed, taking several processing parameters into account; the models are validated with experimental results from the relevant literature and discussed. Additionally, oblique cutting models of high speed hard turning are constructed and discussed. From the reported results useful conclusions may be drawn and it can be stated that the proposed models can be used for industrial application.     

Retrieval of machining information from feature patterns using artificial neural networks

The integration of design and manufacturing has been the subject of much debate and discussion over a long period of time. Recognition of feature patterns and the retrieval of necessary machining information from those patterns play vital roles in this process of integration, as they facilitate the selection of the necessary manufacturing parameters required to transform the designed product into a final physical entity. Although the problem of recognising features from a solid model has been exclusively studied, most existing product models are expressed as engineering drawings. Moreover, the solid model can only provide complete 3D topological and geometrical data and some of the essential machining information cannot be retrieved. In this paper, an approach for defining engineering features, like slots, steps and circular pockets is proposed using binary strings. Two artificial neural networks, one for slots and steps and the other for circular pockets, are designed and developed. These neural networks take the binary strings as inputs and give the relevant machining information as outputs. The networks are trained with non-interacting features and after training, those will become capable of providing the necessary machining information for both non-interacting and interacting features in the domains of slots, steps and circular pockets. This novel approach can further be extended to other features for retrieving relevant machining information and thus facilitating the effective integration of design and manufacturing.     

Modeling and optimization of features of bead geometry including percentage dilution in submerged arc welding using mixture of fresh flux and fused slag

Quality has now become an important issue in today’s manufacturing world. Whenever a product is capable of conforming to desirable characteristics that suit its area of application, it is termed as high quality. Therefore, every manufacturing process has to be designed in such a way that the outcome would result in a high quality product. The selection of the manufacturing conditions to yield the highest desirability can be determined through process optimization. Therefore, there exists an increasing need to search for the optimal conditions that would fetch the desired yield. In the present work, we aim to evaluate an optimal parameter combination to obtain acceptable quality characteristics of bead geometry in submerged arc bead-on-plate weldment on mild steel plates. The SAW process has been designed to consume fused flux/slag, in the mixture of fresh flux. Thus, the work tries to utilize the concept of ‘waste to wealth’. Apart from process optimization, the work has been initiated to develop mathematical models to show different bead geometry parameters, as a function of process variables. Hence, optimization has been performed to determine the maximum amount of slag--flux mixture that can be used without sacrificing any negative effect on bead geometry, compared to the conventional SAW process, which consumes fresh flux only. Experiments have been conducted using welding current, slag-mix percentage and flux basicity index as process parameters, varied at four different levels. Using four³ full factorial designs, without replication, we have carried out welding on mild steel plates to obtain bead-on-plate welds. After measuring bead width, depth of penetration and reinforcement; based on simple assumptions on the shape of bead geometry, we calculated other relevant bead geometry parameters: percentage dilution, weld penetration shape factor, weld reinforcement form factor, area of penetration, area of reinforcement and total bead cross sectional area. All these data have been utilized to develop mathematical models between predictors and responses. Response surface methodology (RSM), followed by the multiple linear regression method, has been applied to develop these models. The effects of selected process parameters on different responses have been represented graphically. Finally grey relational analysis coupled with the Taguchi method (with Taguchi’s orthogonal array) has been applied for parametric optimization of this welding technique. Confirmatory experiments have been conducted to verify optimal results.     

Application of adaptive neuro-fuzzy inference system and cuckoo optimization algorithm for analyzing electro chemical machining process

Electrochemical machining process (ECM) is increasing its importance due to some of the specific advantages which can be exploited during machining operation. The process offers several special privileges such as higher machining rate, better accuracy and control, and wider range of materials that can be machined. Contribution of too many predominate parameters in the process, makes its prediction and selection of optimal values really complex, especially while the process is programmized for machining of hard materials. In the present work in order to investigate effects of electrolyte concentration, electrolyte flow rate, applied voltage and feed rate on material removal rate (MRR) and surface roughness (SR) the adaptive neuro-fuzzy inference systems (ANFIS) have been used for creation predictive models based on experimental observations. Then the ANFIS 3D surfaces have been plotted for analyzing effects of process parameters on MRR and SR. Finally, the cuckoo optimization algorithm (COA) was used for selection solutions in which the process reaches maximum material removal rate and minimum surface roughness simultaneously. Results indicated that the ANFIS technique has superiority in modeling of MRR and SR with high prediction accuracy. Also, results obtained while applying of COA have been compared with those derived from confirmatory experiments which validate the applicability and suitability of the proposed techniques in enhancing the performance of ECM process.     

Data mining and statistical inference in selective laser melting

Selective laser melting (SLM) is an additive manufacturing process that builds a complex three-dimensional part, layer-by-layer, using a laser beam to fuse fine metal powder together. The design freedom afforded by SLM comes associated with complexity. As the physical phenomena occur over a broad range of length and time scales, the computational cost of modeling the process is high. At the same time, the large number of parameters that control the quality of a part make experiments expensive. In this paper, we describe ways in which we can use data mining and statistical inference techniques to intelligently combine simulations and experiments to build parts with desired properties. We start with a brief summary of prior work in finding process parameters for high-density parts. We then expand on this work to show how we can improve the approach by using feature selection techniques to identify important variables, data-driven surrogate models to reduce computational costs, improved sampling techniques to cover the design space adequately, and uncertainty analysis for statistical inference. Our results indicate that techniques from data mining and statistics can complement those from physical modeling to provide greater insight into complex processes such as selective laser melting.     

基于数字孪生的航天器装配质量监控与预测技术

鉴于航天器装配过程中不确定因素多,无法准确有效地预测和评估航天器的实际性能,装配过程中因进行大量复杂的性能试验来验证产品性能指标的符合性而极大影响了装配效率,提出一种基于数字孪生的航天器装配质量在线监控与预测方法。分析了航天器装配执行层面总体流程的特点,在此基础上给出面向航天器装配质量的数字孪生建模方法,以及面向数字孪生构建的产品监控与数据管理方法,最后提出一种基于灰度关联的装配过程质量综合预测方法,可用于航天器装配质量预测。以空间站某泵组件产品为实例,验证了所提方法的正确性。     

数据挖掘技术在质量功能配置建模中的应用与研究

为了有效地发现计算机辅助质量功能配置过程中的需求选型和工艺质量控制规律,引入数据挖掘技术和方法,提出了基于数据挖掘的计算机辅助质量功能配置模型。该模型由数据获取层、数据存储层和数据挖掘层组成。针对用户选型和如何发现质量控制规律,建立了模糊聚类和灰色理论算法的数据挖掘模型,其目标是在质量功能配置数据库中发现所有模糊权重系数大于预定阈值的分类规则及工艺质量,并用以控制未来趋势。用活塞发动机选型和发动机密封环工艺与质量控制归类规律的两个算例,验证了算法的合理性。     

Multi-objective optimization of injection molding process parameters in two stages for multiple quality characteristics and energy efficiency using Taguchi method and NSGA-II

In this paper, the parameters optimization of plastic injection molding (PIM) process was obtained in systematic optimization methodologies by two stages. In the first stage, the parameters, such as melt temperature, injection velocity, packing pressure, packing time, and cooling time, were selected by simulation method in widely range. The simulation experiment was performed under Taguchi method, and the quality characteristics (product length and warpage) of PIM process were obtained by the computer aided engineering (CAE) method. Then, the Taguchi method was utilized for the simulation experiments and data analysis, followed by the S/N ratio method and ANOVA, which were used to identify the most significant process parameters for the initial optimal combinations. Therefore, the range of these parameters can be narrowed for the second stage by this analysis. The Taguchi orthogonal array table was also arranged in the second stage. And, the Taguchi method was utilized for the experiments and data analysis. The experimental data formed the basis for the RSM analysis via the multi regression models and combined with NSGS-II to determine the optimal process parameter combinations in compliance with multi-objective product quality characteristics and energy efficiency. The confirmation results show that the proposed model not only enhances the stability in the injection molding process, including the quality in product length deviation, but also reduces the product weight and energy consuming in the PIM process. It is an emerging trend that the multi-objective optimization of product length deviation and warpage, product weight, and energy efficiency should be emphasized for green manufacturing.     

Decision support tool for selecting fabrication parameters in stereolithography

The selection of fabrication (build) parameters is the most important task performed by the operator of a layer manufacturing (LM) system. In order to select the best parameter configuration for a part, the operator should be able to compare different alternatives and evaluate them under specific constraints, in terms of fabrication cost and quality. In the present paper, a software decision support tool for build parameters selection in stereolithography is presented. Build orientation and layer thickness are proposed as the primary parameters for the definition of candidate solutions, which are evaluated according to a weighted multi-criteria objective function. As the objective function criteria, the build time, surface roughness and process error are employed. The criteria estimation is based on experimentally derived analytical equations or computed from the STL representation of the part. To further enhance the evaluation process, the software tool exports VRML models that incorporate surface roughness or stairstepping data through colour codification. An erratum to this article can be found at     

Multivariate calibration for machine health monitoring: kernel partial least squares combined with variable selection

This work proposes and demonstrates the use of data mining techniques for machine health monitoring through a multivariate calibration model. It utilizes a genetic algorithm (GA)-based variable selection combined with a preprocessing technique of orthogonal signal correction (OSC) for constructing reliable calibration models of shaft misalignment conditions. Improper aligning of shafts often leads to severe problems in many rotating machines. Thus the prediction of shaft alignment conditions is quite essential in making decisions on when to perform alignment maintenance. The main goal of this calibration model is to predict misalignment conditions from historical data. A case study using real misalignment data showed that the prediction results of the proposed calibration models improved significantly compared to existing calibration models. As an extension of linear calibration models, a nonlinear kernel calibration model was also presented. It turned out that linear and nonlinear calibration models of shaft misalignment conditions produced better prediction performance through the use of GA-based variable selection combined with OSC.     

面向产品生命周期的质量控制模型及应用

以产品生命周期和产品进化机理为基础,提出了面向产品生命周期的质量数据模型和质量控制系统总体结构。通过产品进化关系和数据模型完整表达纺织产品的工艺进化过程,实现了纺织品生命周期内的质量数据共享和质量信息的可追溯,从模型层上保证质量数据在纺织品进化全生命周期内的一致性,完整性;同时该数据模型集成智能挖掘技术与统计过程控制技术,实现了对纺织品质量的预测;并通过质量综合评估体系的建立,为纺织工艺优化提供了决策依据。     

PRODUCT DATA PREDICTION WITH UNCERTAINTY IN PRODUCT LIFE CYCLE DESIGN

Various kinds of data are used in new product design and more accurate data make the design results more reliable. Even though part of product data can be available directly from the existing similar products, there still leaves a great deal of data unavailable. This makes data prediction a valuable work. A method that can predict data of product under development based on the existing similar products is proposed. Fuzzy theory is used to deal with the uncertainties in data prediction process. The proposed method can be used in life cycle design, life cycle assessment (LCA) etc. Case study on current refrigerator is used as a demonstration example.     

Cutting force-based real-time estimation of tool wear in face milling using a combination of signal processing techniques

In this paper, combinations of signal processing techniques for real-time estimation of tool wear in face milling using cutting force signals are presented. Three different strategies based on linear filtering, time-domain averaging and wavelet transformation techniques are adopted for extracting relevant features from the measured signals. Sensor fusion at feature level is used in search of an improved and robust tool wear model. Isotonic regression and exponential smoothing techniques are introduced to enforce monotonicity and smoothness of the extracted features. At the first stage, multiple linear regression models are developed for specific cutting conditions using the extracted features. The best features are identified on the basis of a statistical model selection criterion. At the second stage, the first-stage models are combined, in accordance with proven theory, into a single tool wear model, including the effect of cutting parameters. The three chosen strategies show improvements over those reported in the literature, in the case of training data as well as test data used for validation—for both laboratory and industrial experiments. A method for calculating the probabilistic worst-case prediction of tool wear is also developed for the final tool wear model.     

基于神经网络与PSO算法的发动机装配工艺参数优化

针对汽车发动机装配过程中缸体泄漏问题,结合Back Propagation(BP)神经网络及粒子群优化(Particle Swarm Optimization, PSO)算法,提出了一种发动机装配工艺参数优化方法。首先,使用BP神经网络建立了生产工艺参数与质量指标之间的非线性映射关系,并以此作为泄漏率预测模型。其次,根据实际生产需求,应用皮尔逊相关性分析法求解得到相关性最强的部分工位工艺参数,并以其作为后续优化对象。最后,以BP神经网络预测模型作为适应度函数,使用粒子群优化算法求解得到工艺参数的最优值。使用400台发动机的实际生产数据进行试验。试验结果显示,BP神经网络具有较准确的预测效果,结合粒子群优化算法得到了优化后的工艺参数值,显著降低了发动机的泄漏率,具有一定的指导意义。     

An energy consumption prediction approach of die casting machines driven by product parameters

Die casting machines, which are the core equipment of the machinery manufacturing industry, consume great amounts of energy. The energy consumption prediction of die casting machines can support energy consumption quota, process parameter energy-saving optimization, energy-saving design, and energy efficiency evaluation; thus, it is of great significance for Industry 4.0 and green manufacturing. Nevertheless, due to the uncertainty and complexity of the energy consumption in die casting machines, there is still a lack of an approach for energy consumption prediction that can provide support for process parameter optimization and product design taking energy efficiency into consideration. To fill this gap, this paper proposes an energy consumption prediction approach for die casting machines driven by product parameters. Firstly, the system boundary of energy consumption prediction is defined, and subsequently, based on the energy consumption characteristics analysis, a theoretical energy consumption model is established. Consequently, a systematic energy consumption prediction approach for die casting machines, involving product, die, equipment, and process parameters, is proposed. Finally, the feasibility and reliability of the proposed energy consumption prediction approach are verified with the help of three die casting machines and six types of products. The results show that the prediction accuracy of production time and energy consumption reached 91.64% and 85.55%, respectively. Overall, the proposed approach can be used for the energy consumption prediction of different die casting machines with different products.     

Petroleomics: Composition/Structure Relationship with Properties/Performance of Petroleum and Its Fractions

Petroleum, coals and shale oils are the most complex mixtures in nature, containing thousands of components. They require sophisticated analytical technologies for correlation and prediction of the properties/performance of the products as well as the processability of the feeds. After decades of endeavor and development, the analyses of heavy fractions of these hydrocarbon resources remain to be challenges. Among a variety of analytical techniques, mass spectrometry provides unique capabilities for looking into detailed composition down to molecular-level. In mass spectrometry, various ionization techniques have been developed for a wide range of molecular information. Its coupling with chromatography greatly enhances the separation power and specificity of components in complex mixtures. Petroleome, which consists of ~40,000 chemical constituents, are now achievable using ultra-high resolution Fourier-transform ion cyclotron resonance (FT-ICR) spectroscopy with advanced data analysis. Another challenge is to interpret overwhelming analytical data for its physical and chemical significance. Chemometrics have been used as a data mining tool to identify key species that would mostly affect the overall properties of feed and products. Recent development in molecular modeling provides us with abilities for prediction of effectiveness of refining processes from the molecular composition data. This presentation will give an overview of the analytical, particularly mass spectrometric, and modeling approaches to obtain relevant molecular data or information for further understanding of the product quality and the prediction/assessment of petroleum process effectiveness.