IT项目优化及仿真

提出了一个基于设计结构矩阵的任务流程优化算法,给出了分析复杂项目进度和成本风险的仿真模型。该模型考虑了任务之间交互作用导致的循环迭代,以及相关的学习效应。算例仿真结果表明,对于具有高度耦合性及返工概率的复杂IT开发项目,本优化算法可以有效地降低项目执行成本和执行费用。     

基于DSM的复杂产品开发流程优化遗传算法

为减少产品开发过程中的返工迭代,提出一种基于设计结构矩阵(DSM)理论的多目标流程优化遗传算法.通过优化任务执行顺序,减少产品开发过程中的返工以压缩进度和降低成本.该优化算法是一种改进的遗传(GA)算法,在适应度函数中考虑了时间和费用两个指标;在选择、交叉、变异算子中采用了优解保持策略.仿真结果表明,对于高任务耦合度的产品开发项目,该优化算法能使开发时间压缩30%～40%,费用降低7%～2O%.     

基于设计结构矩阵的产品设计过程管理

对顺序设计过程中的不完全依赖活动进行重迭设计，对简单的迭代活动进行分解，从而缩短交互和等待时间．对复杂的多次迭代设计进行了探讨，给出了一种基于规模控制的迭代—集成解决方法．在此基础上设计了一个多参数字结构矩阵实现对该过程的优化管理．     

执行器故障重复过程的鲁棒迭代学习容错控制方法及应用

针对具有执行器故障和外界扰动的线性重复过程,给出一种鲁棒迭代学习容错控制策略.首先,基于二维(2D)系统理论,设计鲁棒迭代学习容错控制器,将迭代学习控制系统等效转化为2D模型;然后,利用线性矩阵不等式(LMI)技术,分析和优化控制系统在时间和迭代方向上的容错性能以及对干扰的抑制性能,同时给出系统满足这些性能的充分条件,并进一步通过求解LMI凸优化问题获得控制器参数;最后,通过对旋转控制系统的仿真结果验证了所提出算法的有效性.     

基于DSM的复杂产品开发流程优化遗传算法

为减少产品开发过程中的返工迭代,提出一种基于设计结构矩阵(DSM)理论的多目标流程优化遗传算法.通过优化任务执行顺序,减少产品开发过程中的返工以压缩进度和降低成本.该优化算法是一种改进的遗传(GA)算法,在适应度函数中考虑了时间和费用两个指标;在选择,交叉,变异算子中采用了优解保持策略.仿真结果表明,对于高任务耦合度的产品开发项目,该优化算法能使开发时间压缩30%-40%,费用降低7%-20%.

     

MRI:面向并行迭代的MapReduce模型

机器学习领域内的多数模型均需要通过迭代计算以求解其最优参数,而MapReduce模型在迭代计算中的缺陷不足导致其在迭代计算中无法得到广泛应用。为解决上述矛盾,基于MapReduce模型提出并实现了一种可用于模型参数求解的并行迭代模型MRI。MRI模型在保持Map以及Reduce阶段的基础上,新增了Iterate阶段以及相关通信协议,实现了迭代过程中模型参数的更新、分发与迭代控制;通过对MapReduce状态机进行增强,实现了节点任务的重用,避免了迭代过程中节点任务重复创建、初始化以及回收带来的性能开销;在任务节点实现了数据缓存,保障了数据的本地性,并在Map节点增加了基于内存的块缓存机制,进一步提高训练集加载效率,以提高整体迭代效率。基于梯度下降算法的实验结果表明:MRI模型在并行迭代计算方面性能优于MapReduce模型。     

时变轨迹下工业过程鲁棒模糊预测控制

针对具有不确定性、外界未知干扰和跟踪轨迹变化的非线性工业过程,提出一种时变轨迹鲁棒模糊预测控制方法。该方法将非线性工业过程根据模糊规则表示为Takagi-Sugeno(T-S)模糊模型,并扩展输出跟踪误差到模糊模型中得到新型多自由度状态空间模型。针对扩展模型设计控制器,并给出基于线性矩阵不等式形式的稳定性充分条件,通过求解稳定性条件中参数计算对应子模型的控制律,并根据当前时刻的系统状态确定每个子模型和控制律的权值。同时,由于在实际生产过程中被控变量需要根据外界变化来实时调节,在设计控制器时充分考虑跟踪轨迹变化的影响。在啤酒发酵罐温度系统仿真研究结果表明设计的控制器能够满足控制需求,保证系统的稳定性和最优性能。     

虚拟企业环境下的复杂产品并行开发框架模型研究

为了解决虚拟企业中复杂产品开发的并行调度问题，提出了虚拟企业环境下的复杂产品并行开发框架模型。该模型在采用设计结构矩阵和扩展工作转秽矩阵分析产品并行开发过程中存在的循环迭代基础上，运用遗传算法对开发任务的调度进行优化。仿真结果表明，该框架模型在处理虚拟企业的复杂产品并行开发方面的瓶颈问题有着显著的优势。     

反应精馏过程模拟与参数的影响

反应精馏平衡级模型方程反应项引入可变参数λ ;模型求解将变量分内、中、外三层迭代 ,计算结果表明算法易收敛、准确性较高、初值易给定。以乙醇乙酸酯化反应精馏过程为例 ,就过程参数对反应精馏的影响进行了计算。     

一种基于二次扩展的准循环类RA码编码器的设计

利用RA码易于实现线性复杂度编码的特点,提出了一种新的类RA码的编码器设计方法。该编码器设计方案基于一种特殊的二次扩展的方法,构造的校验矩阵H具有准循环结构,节省校验矩阵存储空间,对码长和码率参数的设计具有高度的灵活性。该编码器的编码算法利用迭代计算求校验位的值,编码器算法复杂度与码长成线性关系,易于编码。计算机仿真结果表明,在加性高斯白噪声信道条件下,该编码方案能取得与Mackay随机码相当甚至更好的性能。     

Learning deterministic finite automata with a smart state labeling evolutionary algorithm

Learning a deterministic finite automaton (DFA) from a training set of labeled strings is a hard task that has been much studied within the machine learning community. It is equivalent to learning a regular language by example and has applications in language modeling. In this paper, we describe a novel evolutionary method for learning DFA that evolves only the transition matrix and uses a simple deterministic procedure to optimally assign state labels. We compare its performance with the evidence driven state merging (EDSM) algorithm, one of the most powerful known DFA learning algorithms. We present results on random DFA induction problems of varying target size and training set density. We also study the effects of noisy training data on the evolutionary approach and on EDSM. On noise-free data, we find that our evolutionary method outperforms EDSM on small sparse data sets. In the case of noisy training data, we find that our evolutionary method consistently outperforms EDSM, as well as other significant methods submitted to two recent competitions.     

Dissipativity and stabilization of nonlinear repetitive processes

Repetitive processes are characterized by repeated executions of a task defined over a finite duration with resetting after each execution is complete. Also the output from any execution directly influences the output produced on the next execution. The repetitive process model structure arises in the modeling of physical processes and can also be used to effect in the control of other systems, the design of iterative learning control laws where experimental verification of designs has been reported. The existing systems theory for them is, in the main, linear model based. This paper considers nonlinear repetitive processes using a dissipative setting and develops a stabilizing control law with the required conditions expressed in terms of vector storage functions. This design is then extended to stabilization plus disturbance attenuation.     

Task response time for real-time distributed systems with resourcecontentions

An analytic model is proposed for estimating task response times in distributed systems with resource contentions. The model consists of two submodels. The first submodel is an extended queuing network model used for approximating module response times. This submodel is solved by a decomposition technique which reduces the computational complexity by two to three orders of magnitude when compared with a direct approach. The second submodel is a weighted control-flow graph model from which task response time can be obtained by aggregating module response time in accordance with the precedence relationships. Task response times estimated by the analytic model compare closely with simulation results. It is shown that resource contention delays depend on the availability of resources as well as on the invocation rates and response times of the modules that use the resources. The model can be used to study the tradeoffs among module assignments, scheduling policies, interprocessor communications, and resource contentions in distributed processing systems     

Personalized face-pose estimation network using incrementally updated face shape parameters

In this paper, a deep learning method is proposed for human image processing that incorporates a mechanism to update target-specific parameters. The aim is to improve system performance in situations where the target can be continuously observed. Network-based algorithms typically rely on offline training processes that use large datasets, while trained networks typically operate in a one-shot fashion. That is, each input image is processed one by one in the static network. On the other hand, many practical applications can be expected to use continuous observation rather than observation of a single image. The proposed method employs dynamic use of multiple observations to improve system performance. In this paper, the effectiveness of the proposed method adopting an iterative update process is clarified through its implementation in the task of face-pose estimation. The method consists of two separate processes: 1) sequential estimation and updating of face-shape parameters (target-specific parameters) and 2) face-pose estimation for every single image using the updated parameters. Experimental results indicate the effectiveness of the proposed method.

     

Nonlinear predictive control of smooth nonlinear systems based on Volterra models. Application to a pilot plant

There is a large demand to apply nonlinear algorithms to control nonlinear systems. With algorithms considering the process nonlinearities, better control performance is expected in the whole operating range than with linear control algorithms. Three predictive control algorithms based on a Volterra model are considered. The iterative predictive control algorithm to solve the complete nonlinear problem uses the non‐autoregressive Volterra model calculated from the identified autoregressive Volterra model. Two algorithms for a reduced nonlinear optimization problem are considered for the unconstrained case, where an analytic control expression can be given. The performance of the three algorithms is analyzed and compared for reference signal tracking and disturbance rejection. The algorithms are applied and compared in simulation to control a Wiener model, and are used for real‐time control of a chemical pilot plant. Copyright © 2009 John Wiley & Sons, Ltd.     

Iterative design for active control of fluid flow

This paper considers iterative controller design for planar Poiseuille flow by model unfalsification and controller redesign. The main contribution is to show that model-unfalsification-based iterative design can be useful in flow control problems. The a priori knowledge of the dynamics of the sampled system is obtained from the analytic approximation of the Navier-Stokes equations by a Galerkin method. Pole-positions, expected model orders and feasible dynamic variations are valuable prior knowledge which can be taken into account in the uncertainty-model unfalsification-based iterative design scheme developed.     

容错传感器系统分层组合可靠性建模方法

根据容错传感器系统的结构特点,结合不同可靠性建模方法的优点,提出了基于动态故障树和马尔可夫过程的分层组合可靠性建模方法。针对系统中包含的贮备事件,引入时间算子定性描述事件的时序特性,并推导了顶事件故障概率的定量计算公式。采用马尔可夫过程对传感器故障检测隔离进行动态建模。通过仿真比较所得模型与传统完全故障覆盖模型,分析了元件失效率、误检率等参数对系统可靠性的影响。由结果可得,该建模方法可行且更为精确。     

Recovery point selection on a reverse binary tree task model

An analysis is conducted of the complexity of placing recovery points where the computation is modeled as a reverse binary tree task model. The objective is to minimize the expected computation time of a program in the presence of faults. The method can be extended to an arbitrary reverse tree model. For uniprocessor systems, an optimal placement algorithm is proposed. For multiprocessor systems, a procedure for computing their performance is described. Since no closed form solution is available, an alternative measurement is proposed that has a closed form formula. On the basis of this formula, algorithms are devised for solving the recovery point placement problem. The estimated formula can be extended to include communication delays where the algorithm devised still applies     

A comparative simulation study of work processes in autonomous production cells

An approach to human‐centered design and assessment of work processes in flexible manufacturing systems with the help of dynamic task networks is presented. To model and simulate the task networks, the method of timed colored Petri Nets is used. Two task networks are developed. The first task network is a model of work processes in Autonomous Production Cells (APCs). The second task network represents work processes in conventional Computer Numerically Controlled (CNC)‐based manufacturing systems. The material processing technology is associated with 5‐axis milling. The values of attributes of task elements were acquired empirically on a fine‐grained level with reference to a sample milling order. Comparative hypotheses regarding time‐on‐task, supervisory control functions, levels of cognitive control, human error (HE), and labor division were then formulated. To test these hypotheses, several simulation experiments were conducted. The results from inferential statistics show that single‐operator APCs have a 30% higher efficiency in relation to total time‐on‐task. Moreover, the level of cognitive control is significantly shifted toward rule‐ and knowledge‐based behavior. Surprisingly, the simulation of minor HE does not demonstrate a significantly worse performance from APCs. A simulated labor division among central process planner and production operator allows an additional efficiency improvement of approximately 15%. However, the labor division has two important drawbacks: first, a sequential incompleteness of operators' task spectrum occurs; second, the operator has to cope with hierarchical task incompleteness. Finally, a sensitivity analysis was carried out to investigate the effects of varying lot sizes and number of processed orders. © 2002 John Wiley & Sons, Inc.     

非线性间歇过程的迭代学习状态估计

在间歇过程的状态估计中,如何充分利用多批次重复特性信息是一个挑战。迭代学习卡尔曼滤波方法利用卡尔曼滤波沿时间方向估计相邻两批次之间的状态误差,并沿批次方向迭代更新当前状态估计,兼顾了时间和批次两维特性。但是,这种方法只适用于线性系统。针对非线性间歇过程,提出一种迭代学习拟线性卡尔曼滤波器(ILQKF)方法。ILQKF基于间歇过程的标称模型,将实际状态与标称状态之间的误差作为新状态,建立了与误差相关的线性化模型。然后,根据迭代学习卡尔曼滤波方法,对状态误差进行估计,而状态轨迹为误差轨迹与标称轨迹之和,从而估计出非线性间歇过程的状态。啤酒发酵过程的应用仿真验证了ILQKF方法的优越性。