Using recurrent neural networks to detect changes in autocorrelated processes for quality monitoring

With the growing of automation in manufacturing, process quality characteristics are being measured at higher rates and data are more likely to be autocorrelated. A widely used approach for statistical process monitoring in the case of autocorrelated data is the residual chart. This chart requires that a suitable model has been identified for the time series of process observations before residuals can be obtained. In this work, a new neural-based procedure, which is alleviated from the need for building a time series model, is introduced for quality control in the case of serially correlated data. In particular, the Elman’s recurrent neural network is proposed for manufacturing process quality control. Performance comparisons between the neural-based algorithm and several control charts are also presented in the paper in order to validate the approach. Different magnitudes of the process mean shift, under the presence of various levels of autocorrelation, are considered. The simulation results indicate that the neural-based procedure may perform better than other control charting schemes in several instances for both small and large shifts. Given the simplicity of the proposed neural network and its adaptability, this approach is proved from simulation experiments to be a feasible alternative for quality monitoring in the case of autocorrelated process data.     

Evaluating the reliability level of virtual metrology results for flexible process control: a novelty detection-based approach

The purpose of virtual metrology (VM) in semiconductor manufacturing is to support process monitoring and quality control by predicting the metrological values of every wafer without an actual metrology process, based on process sensor data collected during the operation. Most VM-based quality control schemes assume that the VM predictions are always accurate, which in fact may not be true due to some unexpected variations that can occur during the process. In this paper, therefore, we propose a means of evaluating the reliability level of VM prediction results based on novelty detection techniques, which would allow flexible utilization of the VM results. Our models generate a high-reliability score for a wafer’s VM prediction only when its process sensor values are found to be consistent with those of the majority of wafers that are used in model building; otherwise, a low-reliability score is returned. Thus, process engineers can selectively utilize VM results based on their reliability level. Experimental results show that our reliability generation models are effective; the VM results for wafers with a high level of reliability were found to be much more accurate than those with a low level.     

Understanding ART-based neural algorithms as statistical tools for manufacturing process quality control

Neural networks have recently received a great deal of attention in the field of manufacturing process quality control, where statistical techniques have traditionally been used. In this paper, a neural-based procedure for quality monitoring is discussed from a statistical perspective. The neural network is based on Fuzzy ART, which is exploited for recognising any unnatural change in the state of a manufacturing process. Initially, the neural algorithm is analysed by means of geometrical arguments. Then, in order to evaluate control performances in terms of errors of Types I and II, the effects of three tuneable parameters are examined through a statistical model. Upper bound limits for the error rates are analytically computed, and then numerically illustrated for different combinations of the tuneable parameters. Finally, a criterion for the neural network designing is proposed and validated in a specific test case through simulation. The results demonstrate the effectiveness of the proposed neural-based procedure for manufacturing quality monitoring.     

基于振弦传感器的应变无线测量系统设计

针对传统振弦式应变测量手段的缺点与不足,设计了一种基于振弦式传感器的应变无线测量系统。整个系统通过ARM来控制,并通过通用分组无线业务（ GPRS）网络组成无线通信网络,实现了完全无人值守测量和系统的远程控制以及数据的无线传输。系统采样太阳能供电,可以更好地适应现场和野外试验不方便布线和取电的场合;系统具有大容量数据存储能力,无线传输数据的能力,具有高可靠性、高精度、适应性强、低成本等诸多优点,该测量系统使用方便,功能强大,智能化,必将有良好的应用前景。     

Simulation and parameter optimization of flux cored arc welding using artificial neural network and particle swarm optimization algorithm

Flux cored arc welding (FCAW) process is a fusion welding process in which the welding electrode is a tubular wire that is continuously fed to the weld area. It is widely used in industries and shipyards for welding heavy plates. Welding input parameters play a very significant role in determining the quality of a weld joint. This paper addresses the simulation of weld bead geometry in FCAW process using artificial neural networks (ANN) and optimization of process parameters using particle swarm optimization (PSO) algorithm. The input process variables considered here include wire feed rate (F); voltage (V); welding speed (S) and torch Angle (A) each having 5 levels. The process output characteristics are weld bead width, reinforcement and depth of penetration. As per the statistical design of experiments by Taguchi L₂₅ orthogonal array, bead on plate weldments were made. The experimental results were fed to the ANN algorithm for establishing a relationship between the input and output parameters. The results were then embedded into the PSO algorithm which optimizes the process parameters subjected to the objectives. In this study the objectives considered are maximization of depth of penetration, minimization of bead width and minimization of reinforcement.     

Process integrated wire-bond quality control by means of cytokine-Formal Immune Networks

Ultrasonic wire bonding is one of the most frequently used techniques in semiconductor production to establish electrical interconnections. Improper bonding process parameters, wire or substrate contamination or low substrate quality are some of the causes of failed bonds. Process integrated wire-bond quality control techniques compare process feedback signals to a reference for monitoring online the quality of a bond. The feedback signals sampled at high frequencies, constitute high dimensional vectors representing the bonding process characteristics. In the area of online bond failure detection, dimensionality reduction of the input signals and feature extraction of the characteristics of the process are very demanding. Cytokine-Formal Immune Network (cFIN) is a procedure for pattern recognition which presents a low recognition failure rate and a fast recognition due to the reduction of dimensions and feature extraction of the training pattern data set produced in the learning phase. We use cytokine-Formal Immune Networks for recognizing faults present during the wire bonding process. The recognition methodology is intended to be applied into a process integrated quality control system. Further an automated optimization procedure has been developed to find optimal cFIN training parameters. Very promising results for two wire bonding process setups are shown in this paper.     

A neural network ensemble-based model for on-line monitoring and diagnosis of out-of-control signals in multivariate manufacturing processes

In multivariate statistical process control, most multivariate quality control charts are shown to be effective in detecting out-of-control signals based upon an overall statistics. But these charts do not relieve the need for pinpointing source(s) of the out-of-control signals, which would be very useful for quality practitioners to locate the assignable causes that give rise to the out-of-control situation. In this study, a learning-based model has been investigated for monitoring and diagnosing out-of-control signals in a bivariate process. In this model, a selective neural network (NN) ensemble approach (DPSOEN, Discrete Particle Swarm Optimization) was developed for performing these tasks. The simulation results demonstrate that the proposed model outperforms the conventional multivariate control scheme in terms of average run length (ARL), and can accurately classify source(s) of out-of-control signals. Extensive experiment is also carried out to examine the effects of six statistical features on the performance of DPSOEN. Analysis from this study provides guidelines in developing NN ensemble-based Statistical process control recognition systems in multivariate processes.     

A hybrid intelligent approach for optimizing the fine-pitch copper wire bonding process with multiple quality characteristics in IC assembly

Gold is the primary material used for wire bonding in integrated circuit (IC) assembly. Owing to the high appreciation in the price of gold, copper (Cu) wire has become an important substitute material in order to save on manufacturing costs. However, an average of 40% in yield loss during IC assembly can be attributed to improper control of the Cu wire bonding process. To assure cost savings without losing yield, and ensure cost-effective IC assembly, optimization of the parameters for the Cu wire process is critical. This work proposes a hybrid intelligent approach to derive robust parameter settings for a fine-pitch Cu wire bonding process with multiple quality characteristics. The proposed methodology utilizes grey relational analysis and an entropy measurement method to convert the multiple responses into a single synthetic performance index without involving the subjective judgment of an engineer and causing unbalanced improvements of the responses. An integrated neural network model and genetic algorithm method is then applied to acquire the optimal parameter settings. The performance of this method is evaluated experimentally and the results compared with that of the response surface methodology and original parameter settings. The results confirm the feasibility and practicality of this strategy to improve production yield and process capability during Cu wire bonding.     

Improved PI neural network-based tension control for stranded wire helical springs manufacturing

During the winding process of stranded wire helical springs (SWHSs), uneven wire tension always results in high rejection rate and non-compliance service life of SWHSs. Combining the proportion integral neural network (PINN) with a simplified actuator model, this paper presents a new control scheme for the SWHS CNC machine to keep the wire tension uniform. The PINN is improved by introducing an error variance ratio, accounting for the interaction between wires, as a modifying factor in the second hidden layer. The actuator model is simplified based on the analysis of the dynamic characteristics of the actuator. The output value of the improved PINN is transferred into control voltage value by the simplified model. The tension of each wire is controlled by an improved PINN. In order to enhance the control performance, the network parameters are updated using the gradient-based back-propagation method. The validity and consistency of the improved PINN are verified by experiments. The results indicate that (1) the computation load is slight; (2) the rising time of the step response is within 1 s; (3) 89%-96% of tension deviation values of the wire 1 and wire 3 under different process parameters are within 10% of the reference tension value; (4) the standard deviation of the wire 2 with large disturbance is 8.24 N. Compared with other algorithms (incremental PI, multiple PIDNN, PI based particle swarm optimization), the control scheme based on the improved PINN has less computation load, faster response speed and better performance in the time-varying and nonlinear system with larger disturbance.     

Predictive modeling for wastewater applications: Linear and nonlinear approaches

This study compares the multi-period predictive ability of linear ARIMA models to nonlinear time delay neural network models in water quality applications. Comparisons are made for a variety of artificially generated nonlinear ARIMA data sets that simulate the characteristics of wastewater process variables and watershed variables, as well as two real-world wastewater data sets. While the time delay neural network model was more accurate for the two real-world wastewater data sets, the neural networks were not always more accurate than linear ARIMA for the artificial nonlinear data sets. In some cases of the artificial nonlinear data, where multi-period predictions are made, the linear ARIMA model provides a more accurate result than the time delay neural network. This study suggests that researchers and practitioners should carefully consider the nature and intended use of water quality data if choosing between neural networks and other statistical methods for wastewater process control or watershed environmental quality management.     

Synthetic neural networks for process control

Synthetic neural networks offer great promise for process control. A performance comparison is drawn between traditional statistical process control methods and neural networks. Specifically, a series of simulation experiments in which back propagation networks are contrasted with control charts is described. The basis for comparison is average run length (both predicted and observed) and accuracy. The Monte Carlo simulations are derived from plausible production process data. Neural networks were found to perform reasonably well under most conditions.     

Quality process modelling: An application to the continuous electrical resistance annealing of copper wire

The objective of our research is to develope for process manufacturing an inferential quality control system that combines theoretical and experimental knowledge of physical processes with the judgement and experience of experts in plant operations. In this paper, we propose a quality process model, and apply the model to the continuous electrical resistance annealing of copper wire. Theory is used to develop the relationship between the control variables and measurable characteristics of the process, and experimental information is used to develop the relationships that link the process characteristics to the quality characteristics specified by the customer. The approach is illustrated by the manufacturing process, the continuous annealing of copper wire.     

Phase II monitoring of multivariate simple linear profiles

In some quality control applications, quality of a product or process can be characterized by a relationship between two or more variables that is typically referred to as profile. Moreover, in some situations, there are several correlated quality characteristics, which can be modeled as a set of linear functions of one explanatory variable. We refer to this as multivariate simple linear profiles structure. In this paper, we propose the use of three control chart schemes for Phase II monitoring of multivariate simple linear profiles. The statistical performance of the proposed methods is evaluated in term of average run length criterion and reveals that the control chart schemes are effective in detecting shifts in the process parameters. In addition, the applicability of the proposed methods is illustrated using a real case of calibration application.     

A method for improving classification reliability of multilayerperceptrons

Criteria for evaluating the classification reliability of a neural classifier and for accordingly making a reject option are proposed. Such an option, implemented by means of two rules which can be applied independently of topology, size, and training algorithms of the neural classifier, allows one to improve the classification reliability. It is assumed that a performance function P is defined which, taking into account the requirements of the particular application, evaluates the quality of the classification in terms of recognition, misclassification, and reject rates. Under this assumption the optimal reject threshold value, determining the best trade-off between reject rate and misclassification rate, is the one for which the function P reaches its absolute maximum. No constraints are imposed on the form of P, but the ones necessary in order that P actually measures the quality of the classification process. The reject threshold is evaluated on the basis of some statistical distributions characterizing the behavior of the classifier when operating without reject option; these distributions are computed once the training phase of the net has been completed. The method has been tested with a neural classifier devised for handprinted and multifont printed characters, by using a database of about 300000 samples. Experimental results are discussed.     

Back propagation genetic and recurrent neural network applications in modelling and analysis of squeeze casting process

Today, in competitive manufacturing environment reducing casting defects with improved mechanical properties is of industrial relevance. This led the present work to deal with developing the input-output relationship in squeeze casting process utilizing the neural network based forward and reverse mapping. Forward mapping is aimed to predict the casting quality (such as density, hardness and secondary dendrite arm spacing) for the known combination of casting variables (that is, squeeze pressure, pressure duration, die and pouring temperature). Conversely, attempt is also made to determine the appropriate set of casting variables for the required casting quality (that is, reverse mapping). Forward and reverse mapping tasks are carried out utilizing back propagation, recurrent and genetic algorithm tuned neural networks. Parameter study has been conducted to adjust and optimize the neural network parameters utilizing the batch mode of training. Since, batch mode of training requires huge data, the training data is generated artificially using response equations. Furthermore, neural network prediction performances are compared among themselves (reverse mapping) and with those of statistical regression models (forward mapping) with the help of test cases. The results shown all developed neural network models in both forward and reverse mappings are capable of making effective predictions. The results obtained will help the foundry personnel to automate and précised control of squeeze casting process.     

Optimizing the IC wire bonding process using a neural networks/genetic algorithms approach

A critical aspect of wire bonding is the quality of the bonding strength that contributes the major part of yield loss to the integrated circuit assembly process. This paper applies an integrated approach using a neural networks and genetic algorithms to optimize IC wire bonding process. We first use a back-propagation network to provide the nonlinear relationship between factors and the response based on the experimental data from a semiconductor manufacturing company in Taiwan. Then, a genetic algorithms is applied to obtain the optimal factor settings. A comparison between the proposed approach and the Taguchi method was also conducted. The results demonstrate the superiority of the proposed approach in terms of process capability.     

Parameter optimization of etching process for a LGP stamper

This study proposes a two-stage system to optimize the etching process parameter for making a light guide plate (LGP) stamper. The multi-quality characteristics of the parameter settings include depth and uniformity of the microstructures formed in the LGP stamper. The control factors to conduct the process are etching temperature, specific gravity, spray pressure, transfer speed, and oscillating rate. The first stage is to conduct signal-to-noise (S/N) ratio optimization using Taguchi orthogonal array experiments. After conducting the etching process in microstructure, the experimental data can be translated and tested by back-propagation neural networks in order to create S/N ratio and the other quality characteristics predictors. In addition, the S/N ratio predictor and genetic algorithms are used together to obtain combinations of settings and to find the maximized process parameters on S/N ratios. As a result, the quality variance could be minimized. The second stage demonstrates quality characteristics optimization by pushing the process qualities to the targeted specifications. The analysis of variance (ANOVA) is employed to determine the significant control factors. Then, a statistical analysis using the aforementioned quality predictor, S/N ratios predictor, and particle swarm optimization is implemented to simulate the targeted specifications and then find a suitable specifications combination and the most stable and qualified process.     

An intelligent composite system for statistical process control

Combining statistical process control, artificial neural networks and an expert system for the intelligent analysis and control of a plastic extruder facility is described. Statistical methodology is compared and contrasted to the exploratory neural network technique, which learns to relate and classify dependent production variables based on measurements taken on-line during the process. Integrating the neural network analysis into a composite control system using an expert system is presented.     

Application of integrated recurrent neural network with multivariate adaptive regression splines on SPC-EPC process

The integration of statistical process control and engineering process control has been reported as an effective way to monitor and control the autocorrelated process. However, because engineering process control compensates for the effects of underlying disturbances, the disturbance patterns become very hard to recognize, especially when various abnormal control chart patterns are mixed and co-existed in the engineering process. In this study, a new control chart pattern recognition model which integrates multivariate adaptive regression splines and recurrent neural network is proposed to not only address the problem of feature selection (i.e., lagged process measurements) but also improve the pattern recognition accuracy. The performance of the proposed method is evaluated by comparing the recognition results of multivariate adaptive regression splines and recurrent neural network with the results of four competing approaches (multivariate adaptive regression splines-extreme learning machine, multivariate adaptive regression splines-random forest, single recurrent neural network, and single random forest) on the simulated individual process data. The experimental study shows that the proposed multivariate adaptive regression splines and recurrent neural network approach can not only solve the problem of variable selection but also outperform other competing models. Moreover, according to the lagged process measurements selected by the proposed approach, lagged observations that exerted significant impact on the construction of the control chart pattern recognition model can be identified successfully. This study has significant implications for research and practice in production management and provides a valuable reference for manufacturing process managers to better understand and develop strategies for control chart pattern recognition.