基于变折扣因子EWMA的混合产品批间控制方法

批间(run-to-run,简称R2R)控制现今已被广泛用于半导体生产行业.指数加权移动平均(exponet weighted moving average.EWMA)是R2R控制的一种重要算法.折扣因子是EWMA控制期的主要参数.本文在模型中考虑了实际生产过程中混合产品少量多样的特点,引入了基于产品的变折扣因子EWMA控制算法,解决了产品切换时制程输出收敛速度过慢的问题.变折扣因子的引入提高了制程输出的响应速率而并不影响制程输出的稳定性.对实际过程的模拟仿真检验了该控制算法的可行性和优越性.     

一种基于变因子梯度型PAST 的递推子空间辨识算法

针对传统递推子空间算法采用固定遗忘因子而存在易受噪声干扰等问题,提出一种新的变因子递推子空间辨识算法.该算法首先引入变因子构造和更新输入输出Hankel矩阵以及观测向量;然后利用变因子改进梯度型子空间跟踪算法估计系统的广义能观测矩阵,并由广义能观测矩阵来估计系统矩阵;最后运用系统矩阵的特征值空间欧氏距离信息实现变因子更新步骤,使算法具有自适应性.将所提出算法应用于一类时不变系统和慢变系统模型,数值仿真结果表明该算法跟踪速度较快且跟踪效果良好.     

基于观测器的飞行器时变控制效益估计算法

研究飞行器时变控制效益的在线估计方法,证明了系统控制效益可估计的充分条件. 针对效益矩阵分段连续时变的情形,给出了基于观测器,方程求解和最小二乘法的在线估计方法,分析和证明了算法的收敛性。 数值仿真结果表明,该算法能快速估计时变控制效益,具有良好的稳定性和鲁棒性。

     

基于小波变换的分布式信息一致滤波算法

研究一类基于小波变换的分布式信息一致滤波算法. 首先, 利用Haar 小波变换建立目标状态及其观测在不同粗尺度下的系统模型; 然后, 基于该模型, 在不同粗尺度上分别进行分布式信息一致滤波估计; 最后, 针对不同粗尺度估计, 通过Haar 小波逆变换重构最细尺度(初始尺度) 目标状态的估计. 仿真结果表明, 所提出的算法可以有效提高分布式信息一致滤波算法的计算效率.

     

基于间接拉普拉斯模型因子估计的语音增强算法

对DCT城基于拉普拉斯统计模型的语音增强，分析了模型因子的估计误差及其对于算法整体增强性能的影响，并根据广义高斯分布模型度其形态参数的概念与性质，提出了一种新的拉普拉斯模型因子估计方法，该方法结构简单，它利用拉普拉斯模型条件下语音分量方差与模型因子的对应关系，间接地获取模型因子的估计，算法不仅有效地消除了噪声分量对于估计精度的影响，而且可以快速地跟踪语音分量的变化。仿真结果表明，基于该模型因子估计方法的语音增强算法在多种噪声背景下具有更出色的语音增强效果。     

电磁矢量阵中基于平行因子压缩感知的角度估计算法

将平行因子框架与压缩感知理论相结合,解决了电磁矢量传感器阵列中的波达方向估计问题。首 先将接收信号构建成平行因子模型,然后结合压缩感知理论,对平行因子模型压缩。根据三线性交替最小二乘算法对压缩后的平行因子模型进行分解,最后利用信号的稀疏性,得到波达方向估计。借助压缩过程,本文算法降低了传统的平行因子算法的计算复杂度,节约了 存储空间。本文算法无需谱峰搜索,且同时适用于均匀线阵和非均匀线阵。该算法的角度估计性优于ESPRIT算法,且接近传统的基于平行因子模型的角度估计算法,仿真结果证明该算法的有效性。     

基于全概率更新的改进RANSAC算法

针对鲁棒性模型估计问题,提出一种基于全概率更新的改进RANSAC算法.该方法利用混合分布模型获取测试样本点的初始概率估计.在RANSAC算法框架下,根据模型估计与测试样本点对一致集的适应度建立全概率评价准则.在此基础上,采用逆变映射作为采样策略,提高了算法的收敛速度;同时,运用测试点平均概率对所提出算法进行了收敛性分析.最后,通过仿真与实际图像匹配实验进一步验证了所提出算法的有效性与可行性.     

损失数据线性参数系统的递推最小二乘辨识方法

针对损失数据线性参数系统的参数辨识问题, 借助辅助模型辨识思想推导出其变递推间隔辅助模型递 推最小二乘算法.为了提高该算法的计算效率, 利用分解技术得到变递推间隔分解递推最小二乘算法 估计系统参数.此外, 在变递推间隔分解递推最小二乘算法中引入遗忘因子, 从而提高参数估计精度和收敛速度.仿真结果表明, 所提出的算法能有效估计系统参数.     

采用3D Gaussian Facet模型的亚体素表面检测

提出一种基于Gaussian facet模型的3D边缘检测算法.首先利用Gaussian加权最小二乘拟合,引入空间权因子表达图像采样点对模型参数估计的相对重要度,扩展了经典Haralick facet模型,建立了3D Gaussian facet模型及其计算公式;然后采用抗噪性好的3D IDDG算子估计梯度方向,并在该梯度方向上计算二阶方向导数过零点,以获得表面点亚体素位置.实验结果表明,该算法能有效地降低邻近边缘干涉对检测结果的影响,可更好地提取尺寸较小的结构边缘.     

Packie图广泛应用于预测原油常压塔产品分馏的精度

Packie图广泛应用于预测原油常压塔产品分馏的精度.为了能够应用于计算机的模拟与计算,本文通过采用非线性同归和单纯形的加速算法拟合Packie图的"原油常压精馏塔塔顶产品与一线分馏精度图"和"原油常压精馏塔侧线产品分馏精度图",建立可用计算机计算的数学模型.该模型方程形式简单,并估计其模型参数误差.证明,该数学关联式的计算结果,跟两图得出的数据能够吻合,相关系数分别为0.9982697和0.999005,计算准确.     

Using past manufacturing experience to assist building the yield forecast model for new manufacturing processes

Polarizers are one of the key parts of Thin-Film Transistor Liquid-Crystal Displays (TFT-LCD), and their production requires high material costs. How to reduce manufacturing costs is thus a key task in this highly competitive global market. The precise yield forecast model considering learning effects that is proposed in this work is believed to be an effective approach to reduce both the raw material input-cost and inventory cost of overproduction. Support vector regression (SVR) model is one of the commonly used approaches to forecast the yield trend. However, in the early manufacturing stages for a new product, an SVR model is usually sensitive and unstable because of the use of insufficient data. Faced with this problem, this research aims at enhancing the SVR model by using past manufacturing experience and virtual samples to estimate the yield trend model for pilot products. This paper proposes a novel Quadratic-Curve Diffusion (QCD) method, wherein we derive a quadratic yield function (QYF) of the new manufacturing process for each key manufacturing variable by utilizing past manufacturing experience; and then use the QYF to generate virtual samples to assist building the overall yield forecast model of the new manufacturing process. The results show that the proposed method is superior to the performance of other forecast and virtual sample generation models.     

An intelligent virtual metrology system with adaptive update for semiconductor manufacturing

Virtual metrology involves the estimation of metrology values using a prediction model instead of metrological equipment, thereby providing an efficient means for wafer-to-wafer quality control. Because wafer characteristics change over time according to the influence of several factors in the manufacturing process, the prediction model should be suitably updated in view of recent actual metrology results. This gives rise to a trade-off relationship, as more frequent updates result in a higher accuracy for virtual metrology, while also incurring a heavier cost in actual metrology. In this paper, we propose an intelligent virtual metrology system to achieve a superior metrology performance with lower costs. By employing an ensemble of artificial neural networks as the prediction model, the prediction, reliability estimation, and model update are successfully integrated into the proposed virtual metrology system. In this system, actual metrology is only performed for those wafers where the current prediction model cannot perform reliable predictions. When actual metrology is performed, the prediction model is instantly updated to incorporate the results. Consequently, the actual metrology ratio is automatically adjusted according to the corresponding circumstances. We demonstrate the effectiveness of the method through experimental validation on actual datasets.     

An integrated algorithm for cutting stock problems in the thin-film transistor liquid crystal display industry

The cutting stock problem (CSP) is a critical issue in the manufacturing of thin film transistor liquid crystal display (TFT-LCD) products. Two manufacturing processes are utilized in this industry: (1) various TFT-LCD plates are cut from a glass substrate based on cutting patterns, and (2) the number of glass substrates required to satisfy customer requirements is minimized. The current algorithm used to select the cutting pattern is defined as a mixed integer program (MIP). Although the current MIP method yields an optimal solution, but the computation time is unacceptable when the problem scale is large. To accelerate the computation and improve the current method, this study proposes an integrated algorithm that incorporates a genetic algorithm, a corner arrangement method, and a production plan model to solve CSPs in the TFT-LCD industry. The results of numerical experiments demonstrate that the proposed algorithm is significantly more efficient than the current method, especially when applied to large-scale problems.     

Preliminary cost estimates for thin-film transistor liquid–crystal display inspection and repair equipment: A hybrid hierarchical approach

The thin-film transistor liquid–crystal display (TFT-LCD) industry has developed rapidly in recent years. Because TFT-LCD manufacturing is highly complex and requires different tools for different products, accurately estimating the cost of manufacturing TFT-LCD equipment is essential. Conventional cost estimation models include linear regression (LR), artificial neural networks (ANNs), and support vector regression (SVR). Nevertheless, in accordance with recent evidence that a hierarchical structure outperforms a flat structure, this study proposes a hierarchical classification and regression (HCR) approach for improving the accuracy of cost predictions for TFT-LCD inspection and repair equipment. Specifically, first-level analyses by HCR classify new unknown cases into specific classes. The cases are then inputted into the corresponding prediction models for the final output. In this study, experimental results based on a real world dataset containing data for TFT-LCD equipment development projects performed by a leading Taiwan provider show that three prediction models based on HCR approach are generally comparable or better than three conventional flat models (LR, ANN, and SVR) in terms of prediction accuracy. In particular, the 4-class and 5-class support vector machines in the first-level HCR combined with individual SVR obtain the lowest root mean square error (RMSE) and mean average percentage error (MAPE) rates, respectively.     

Identification of tool and product effects in a mixed product and parallel tool environment

In the semiconductor manufacturing industry, production resembles an automated assembly line in which many similar products with slightly different specifications are manufactured step-by-step, with each step being a complicated physiochemical batch process performed by a number of tools. This constitutes a high-mix production system for which effective run-to-run control (RtR) and fault detection control (FDC) can be carried out only if the states of different tools and different products can be estimated. However, since in each production run, a specific product is performed on a specific tool, absolute individual states of products and tools are not observable. In this work, a novel state estimation method based on analysis of variance (ANOVA) is developed to estimate the relative states of each product and tool to the grand average performance of this station in the fab. The method is formulated in the form of a recursive state estimation using the Kalman filter. The advantages of this method are demonstrated using simulations to show that the correct relative states can be estimated in production scenarios such as tool-shift, tool-drift, product ramp-up, tool/product-offline and preventive maintenance (PM). Furthermore, application of this state estimation method in RtR control scheme shows that substantial improvements in process capabilities can be gained, especially for products with small lot counts. The proposed algorithm is also evaluated by an industrial application.     

Confidence estimation of feedback information for logicdiagnosis

This paper proposes an estimation method for the confidence level of feedback information (CLFI), namely the confidence level of reported information in computer integrated manufacturing (CIM) architecture for logic diagnosis. This confidence estimation provides a diagnosis module with precise reported information to quickly identify the origin of equipment failure. We studied the factors affecting CLFI, such as measurement system reliability, production context, position of sensors in the acquisition chains, type of products, reference metrology, preventive maintenance and corrective maintenance based on historical data and feedback information generated by production equipments. We introduced the new ‘CLFI’ concept based on the Dynamic Bayesian Network approach and Tree Augmented Naïve Bayes model. Our contribution includes an on-line confidence computation module for production equipment data, and an algorithm to compute CLFI. We suggest it to be applied to the semiconductor manufacturing industry.     

Incorporating ANNs and statistical techniques into achieving process analysis in TFT-LCD manufacturing industry

The ability to improve yield is an important competitiveness determinant for thin-film transistor-liquid crystal displays (TFT-LCD) factories. Until now, few studies were proposed to address the related issues for process analysis in TFT-LCD industry. Therefore, the information (e.g. the domain knowledge or the parameter effect) or the improvement chance hidden from process analysis will be frequently omitted. That is, the yield or yield loss model construction, the critical manufacturing processes (or layers) and the clustering effect based on the abnormal position (or defect) on TFT-LCD glasses will became the important issues to be addressed in TFT-LCD industry. In this study, we proposed an integrated procedure incorporating the data mining techniques, e.g. artificial neural networks (ANNs) and stepwise regression techniques, to achieve the construction of yield loss model, the effect analysis of manufacturing process and the clustering analysis of abnormal position (or it can be viewed as defect) for TFT-LCD products. Besides, an illustrative case owing to TFT-LCD manufacturer at Tainan Science Park in Taiwan will be applied to verifying the rationality and feasibility of our proposed procedure.     

A model for batch available-to-promise in order fulfillment processes for TFT-LCD production chains

The main purpose of this study is to explicitly highlight several special production characteristics in a thin-film transistor liquid crystal display (TFT-LCD) manufacturing industry and to present an available-to-promise (ATP) model that supports decision-making in order fulfillment processes for TFT-LCD manufacturing. A TFT-LCD production chain differs from others in its special production characteristics such as alternative bill-of-materials (BOMs), grade transition, etc., which are significant factors driving a success in an ATP implementation. Customers may specify a quality level and the materials to be used in a finished product in inquiry orders. The quality of the working-in-process can be altered using different assembled components. The ATP model enhances the responsiveness of order fulfillment processes. The ATP model directly links available material resources and capacity with inquiries or existing customer orders to improve the overall performance of the production chain. A case study using the model demonstrates the effectiveness and efficiency of the proposed ATP model in a TFT-LCD production chain and investigates the sensitivity of TFT-LCD plant performance to changes in order batching intervals.     

A model for batch available-to-promise in order fulfillment processes for TFT-LCD production chains

The main purpose of this study is to explicitly highlight several special production characteristics in a thin-film transistor liquid crystal display (TFT-LCD) manufacturing industry and to present an available-to-promise (ATP) model that supports decision-making in order fulfillment processes for TFT-LCD manufacturing. A TFT-LCD production chain differs from others in its special production characteristics such as alternative bill-of-materials (BOMs), grade transition, etc., which are significant factors driving a success in an ATP implementation. Customers may specify a quality level and the materials to be used in a finished product in inquiry orders. The quality of the working-in-process can be altered using different assembled components. The ATP model enhances the responsiveness of order fulfillment processes. The ATP model directly links available material resources and capacity with inquiries or existing customer orders to improve the overall performance of the production chain. A case study using the model demonstrates the effectiveness and efficiency of the proposed ATP model in a TFT-LCD production chain and investigates the sensitivity of TFT-LCD plant performance to changes in order batching intervals.     

基于UKF 和神经网络的一类非线性系统状态估计

在模型未知的情况下,估计过程的重要变量尤为重要.鉴于此,采用不敏卡尔曼滤波(UKF)与神经网络相结合的方法,解决一类未知模型非线性系统的状态估计问题.采用动态神经网络对非线性系统进行建模,利用UKF对状态和权值进行同时更新,从而达到神经网络逼近真实模型,估计值跟随真实值的目的.通过两个仿真实例表明了所提出的方法具有良好的估计效果,并且状态在输出中的比重越大,其估计精度越高.