信号处理与深度学习硬件加速的一致性计算结构

在计算需求层面对多种典型信号处理算法与深度学习算法进行了分析与模块化分解,提取了两类应用共有的且适合并行硬件加速的计算模块,提出了信号处理与深度学习的一致性计算模型,并基于一致性计算模型设计了控制与计算分离的层次化处理单元与阵列化计算结构。通过对不同应用计算过程的软件定义能够实现信号处理与深度学习的一致性硬件加速计算,基于Zynq计算平台从重构效率与计算性能两个方面对一致性计算模型与计算结构进行了验证,结果表明：基于一致性计算模型的软件定义可重构计算结构,具有较高的计算性能与重构效率。     

涡轮盘疲劳寿命可靠性设计仿真及优化策略

涡轮盘是航空发动机主要部件之一,一旦发生破坏性故障将导致严重的后果。在充分考虑影响涡轮盘高低周复合疲劳寿命因素不确定性基础上,以MATLAB为平台,设计了涡轮盘高低周复合疲劳寿命可靠性优化设计的联合仿真平台。利用寿命函数和寿命可靠性分析极限状态函数中的共性需求,提出了在优化迭代的过程中自适应构建寿命函数Kriging模型和寿命可靠性极限状态面Kriging模型时共用训练样本点的策略。同时,提出了一种构建寿命函数Kriging模型的学习函数。使用所搭建的疲劳寿命可靠性优化设计平台,完成了某型涡轮盘盘心、榫槽以及涡轮盘系统高低周复合疲劳寿命的可靠性优化设计。结果表明,最优设计方案的局部最大应力显著降低,均值寿命大幅提高,并满足可靠性约束。     

信息战建模的分析与探讨

通过对信息战建模的分析与总结,指出了一个完善的信息战建模方法应具备的基本功能;给出了这种建模方法的结构流图,并分析了结构流图的构成.该方法建立的信息战模型,能有效地识别信息战的效果,对信息战的理论发展和具体实施都有较重要的指导作用.     

正弦波型直流无刷电动机位置伺服系统的变结构控制

在分析正弦波型直流无刷电动机数学模型的基础上,采用比例等速变速控制策略(PCV)设计控制器,应用于正弦波型直流无刷电动机位置伺服系统.仿真结果表明:与传统PI控制器比较,PCV控制策略可以明显改善系统的品质,并增强系统的鲁棒性.     

长短周期记忆网络微调的氧气浓缩器寿命预测模型迁移

针对寿命预测模型迁移问题,提出了一种长短周期记忆网络微调(long short-term memory fine tune, LSTM-fine-tune)的迁移模型,利用理想条件下的试验数据对模型进行训练。在迁移过程中,对部分LSTM网络层进行冻结,利用实际服役环境下的数据对网络其他部分进行修正。为验证模型的泛化能力,采用不同相位与幅值的正弦函数生成数据,通过学习数据获取正弦函数的经验知识,并应用至其他正弦函数的回归,结果表明LSTM-fine-tune模型能够快速拟合,平均均方误差仅为1.033 5,明显低于直接预测误差1.536 8。为通过实际监测数据检验本方法泛化能力,分别获取了试验条件下与实际服役环境下氧气浓缩器的数据,对模型的泛化能力进行验证。结果表明,迁移后训练集预测精度提高了43.0%,测试集预测精度提高了20.2%。     

建构学习理论在教学中的应用

在教学中运用建构学习理论,针对“基于经验的知识建构活动”进行有益尝试。在此理论背景下,教师最重要的任务是利用三个层次的建构性经验源为学生提供知识建构的良好条件,并探讨这一理念在教学中的具体应用及其效果。     

Analysis of decentralized multi‐product pull systems with lost sales

We study a pull‐type, flexible, multi‐product, and multi‐stage production/inventory system with decentralized two‐card kanban control policies. Each stage involves a processor and two buffers with finite target levels. Production stages, arranged in series, can process several product types one at a time. Transportation of semi‐finished parts from one stage to another is performed in fixed lot sizes. The exact analysis is mathematically intractable even for smaller systems. We present a robust approximation algorithm to model two‐card kanban systems with batch transfers under arbitrary complexity. The algorithm uses phase‐type modeling to find effective processing times and busy period analysis to identify delays among product types in resource contention. Our algorithm reduces the effort required for estimating performance measures by a considerable margin and resolves the state–space explosion problem of analytical approaches. Using this analytical tool, we present new findings for a better understanding of some tactical and operational issues. We show that flow of material in small procurement sizes smoothes flow of information within the system, but also necessitates more frequent shipments between stages, raising the risk of late delivery. Balancing the risk of information delays vis‐à‐vis shipment delays is critical for the success of two‐card kanban systems. Although product variety causes time wasted in setup operations, it also facilitates relatively short production cycles enabling processors to switch from one product type to another more rapidly. The latter point is crucial especially in high‐demand environments. Increasing production line size prevents quick response to customer demand, but it may improve system performance if the vendor lead‐time is long or subject to high variation. Finally, variability in transportation and processing times causes the most damage if it arises at stages closer to the customer. © 2007 Wiley Periodicals, Inc. Naval Research Logistics, 2007     

Managing components for assemble‐to‐order products with lead‐time‐dependent pricing: The full‐shipment model

We study a component inventory planning problem in an assemble‐to‐order environment faced by many contract manufacturers in which both quick delivery and efficient management of component inventory are crucial for the manufacturers to achieve profitability in a highly competitive market. Extending a recent study in a similar problem setting by the same authors, we analyze an optimization model for determining the optimal component stocking decision for a contract manufacturer facing an uncertain future demand, where product price depends on the delivery times. In contrast to our earlier work, this paper considers the situation where the contract manufacturer needs to deliver the full order quantity in one single shipment. This delivery requirement is appropriate for many industries, such as the garment and toy industries, where the economies of scale in transportation is essential. We develop efficient solution procedures for solving this optimization problem. We use our model results to illustrate how the different model parameters affect the optimal solution. We also compare the results under this full‐shipment model with those from our earlier work that allows for multiple partial shipments. © 2007 Wiley Periodicals, Inc. Naval Research Logistics, 2007     

Puzzle‐based storage systems

We introduce and develop models for a physical goods storage system based on the 15‐puzzle, a classic children's game in which 15 numbered tiles slide within a 4 × 4 grid. The objective of the game is to arrange the tiles in numerical sequence, starting from a random arrangement. For our purposes, the tiles represent totes, pallets, or even containers that must be stored very densely, and the objective is to maneuver items to an input–output point for retrieval or processing. We develop analytical results for storage configurations having a single empty location (as in the game) and experimental results for configurations with multiple empty locations. Designs with many empty locations can be made to form aisles, allowing us to compare puzzle‐based designs with traditional aisle‐based designs found in warehousing systems. © 2007 Wiley Periodicals, Inc. Naval Research Logistics, 2007     

一种支持向量机增量学习淘汰算法

针对大规模数据集的分类问题,支持向量机的训练成为一个难题。增量学习是解决这一难题的思路之一。分析了新增样本加入训练集后支持向量集的变化情况,提出了一种基于密度法的支持向量机增量学习淘汰算法,淘汰了对最终分类无用的样本,在保证测试精度的同时减少了训练时间。实验仿真证明这种算法是有效的。