Management of demand-driven production systems

Control-synthesis techniques are developed for demand-driven production systems. The resulting policies are time-optimal for a deterministic model, and approximately time-optimal for a stochastic model. Moreover, they are easily adapted to take into account a range of issues that arise in a realistic, dynamic environment. In particular, control synthesis techniques are developed for models in which resources are temporarily unavailable. This may be due to failure, maintenance, or an unanticipated change in demand. These conclusions are based upon the following development. i) Workload models are investigated for demand-driven systems, and an associated workload-relaxation is introduced as an approach to model-reduction. ii) The impact of hard constraints on performance, and on optimal control solutions is addressed via Lagrange multiplier techniques. These include deadlines and buffer constraints. iii) Rules for choosing appropriate safety-stocks as well as hedging-points are described to ensure robustness of control solutions with respect to persistent disturbances, such as variability in demand and yield.     

可变服务率模式下基于需求驱动的传送带给料加工站系统的优化控制

本文主要研究可变服务率模式下基于需求驱动的传送带给料加工站(CSPS)系统的优化控制问题,主要目标是对系统的随机优化控制问题进行建模和提供解决方案.论文以缓冲库和成品库剩余容量为联合状态,以站点前视距离和工件服务率为控制变量,将其最优控制问题描述为半马尔科夫决策过程(SMDP)模型.该模型为利用策略迭代等方法求解系统在平均准则或折扣准则下的最优控制策略提供了理论基础,特别地,据此可引入基于模拟退火思想的Q学习算法等优化方法来寻求近似解,以克服理论求解过程中的维数灾和建模难等困难.仿真结果说明了本文建立的数学模型及给出的优化方法的有效性.     

用于视觉词语生成的概率预测器

视觉词语的产生是基于字袋模型的图像检索中的重要一环:根据已知的视觉词典,查询图像特征被映射到词典中相应的视觉词语。提出一种新的基于空间相关性的快速视觉词语产生算法。统计视觉词典中任意两个词语在数据库中的共生次数,构建视觉词语共生表。利用共生表,建立一种新的概率预测器来辅助预测已知词语的近邻词语。将预测器与快速近似最近邻查找算法结合,在标准图像检索数据库上进行实验测试,相比较传统的树形搜索算法或哈希算法,新算法在时间效率上获得明显提高。     

Probabilistic assessment of user's emotions in educational games

We present a probabilistic model to monitor a user's emotions and engagement during the interaction with educational games. We illustrate how our probabilistic model assesses affect by integrating evidence on both possible causes of the user's emotional arousal (i.e., the state of the interaction) and its effects (i.e., bodily expressions that are known to be influenced by emotional reactions). The probabilistic model relies on a Dynamic Decision Network to leverage any indirect evidence on the user's emotional state, in order to estimate this state and any other related variable in the model. This is crucial in a modeling task in which the available evidence usually varies with the user and with each particular interaction. The probabilistic model we present is to be used by decision theoretic pedagogical agents to generate interventions aimed at achieving the best tradeoff between a user's learning and engagement during the interaction with educational games.     

Probabilistic assessment of financial risk in e-business associations

Business activities are a result of carefully formed associations between different users in order to achieve certain pre-decided outcomes. Decision-making in such associations is an important step and transactional risk analysis is one of the integral processes that facilitates this step. This paper presents an approach that determines the negative consequences (termed as financial risk) of forming e-business associations. Unlike other approaches, our model captures the different types of events and their uncertainties to determine the financial risk by using the convolution operator and expressing it as a probabilistic measure rather than as a crisp financial value. Such representation makes sense as the financial risk may be determined at a point of time in future where nothing is certain. Depending upon the complexity of the problem, we explain the different ways of using the convolution operator to determine the financial risk. The simulation result shows a better representation and understanding of the financial risk that will provide important inputs to the transactional risk analysis and the decision-making process.     

Probabilistic optimal solution assessment for DCOPs

Distributed Constraint Optimization Problems (DCOPs) are widely used in Multi-Agent Systems for coordination and scheduling. The present paper proposes a heuristic algorithm that uses probabilistic assessment of the optimal solution in order to quickly find a solution that is not far from the optimal one. The heuristic assessment uses two passes by the agents to produce a high-quality solution. Extensive performance evaluation demonstrates that the solution of the proposed probabilistic assessment algorithm is indeed very close to the optimum, on smaller problems where this could be measured. In larger set-ups, the quality of the solution is demonstrated relatively to standard incomplete search algorithms.     

Probabilistic assessment of small disturbance stability in multimachine power systems

In planning electric power systems, it is always necessary to assess whether small-disturbance (SD) instability phenomena occur at prefixed system operating conditions. This analysis can become very difficult when the problem data are uncertain. In such cases the use of deterministic approaches is inadequate and the application of probabilistic analysis techniques is the most feasible alternative. This paper presents a new and practical probabilistic approach for the assessment of SD stability in multimachine power systems taking into account the uncertainties associated with bus load forecasting and treating loads as random uncorrelated variables with normal distributions. This approach proves suitable for determining the risk of SD instability for each expected system operating condition and for systematically individualizing all factors that can affect the probability of SD instability in large power systems. A numerical example illustrates the capability of the proposed technique.     

Primitives for resource management in a demand-driven reduction model

Graph reduction has recently become well-known as a suitable technique for implementing applicative languages. However, its use hitherto has been primarily in implementing determinate programs. This paper is concerned with extensions to a demand-driven graph reduction language for performing resource management, and entail indeterminate operations. Our focus is on primitives for programming the synchronization and scheduling of resources, as found, for example, in operating systems. We propose queueing operators and operators for controlling the order of execution of operations. We show that demand-driven graph reduction simplifies the implementation of resource management, especially in obviating explicit protocols for task creation and deletion, and for busy waiting. The language FGL (Function Graph Language) is first introduced, and the primitives are presented in this language. A number of examples are then presented to show the versatility of these primitives. The paper concludes with a brief sketch of a higher-level notation, calledresource expressions, for specifying synchronization properties.This research was undertaken while the authors were at the University of Utah, Salt Lake City, and was supported by a grant MCS77-09369 A01 from the National Science Foundation.     

Probabilistic nod generation model based on speech and estimated utterance categories

ABSTRACT

We proposed and evaluated a probabilistic model that generates nod motions based on utterance categories estimated from the speech input. The model comprises two main blocks. In the first block, dialog act-related categories are estimated from the input speech. Considering the correlations between dialog acts and head motions, the utterances are classified into three categories having distinct nod distributions. Linguistic information extracted from the input speech is fed to a cluster of classifiers which are combined to estimate the utterance categories. In the second block, nod motion parameters are generated based on the categories estimated by the classifiers. The nod motion parameters are represented as probability distribution functions (PDFs) inferred from human motion data. By using speech energy features, the parameters are sampled from the PDFs belonging to the estimated categories. The effectiveness of the proposed model was evaluated using an android robot, through subjective experiments. Experiment results indicated that the motions generated by our proposed approach are considered more natural than those of a previous model using fixed nod shapes and hand-labeled utterance categories.     

Probabilistic verification and synthesis of the next generation airborne collision avoidance system

The next generation airborne collision avoidance system, ACAS X, departs from the traditional deterministic model on which the current system, TCAS, is based. To increase robustness, ACAS X relies on probabilistic models to represent the various sources of uncertainty. The work reported in this paper identifies verification challenges for ACAS X, and studies the applicability of probabilistic verification and synthesis techniques in addressing these challenges. Due to shortcomings of off-the-shelf probabilistic analysis tools, we developed a new framework, named VeriCA (Verification for Collision Avoidance). VeriCA is a combined probabilistic synthesis and verification framework that is custom designed for ACAS X and systems with similar characteristics. VeriCA supports Java as a modeling language, is memory efficient, employs parallelization, and provides an interactive simulator that displays aircraft encounters and the corresponding ACAS X behavior. We describe the application of our framework to ACAS X, together with the results and recommendations that our analysis produced.     

Probabilistic framework and performance evaluation for prioritized call admission control in next generation networks

Next generation networks (NGN) are designed to support a wide range of applications with various service classes (SCs) guaranteeing the respective quality of service (QoS) levels. Since such networks are resource constrained, call admission control (CAC) is imperative to achieve the required QoS levels. In this paper, a new probabilistic framework for CAC schemes is proposed based on controlling each SC independently by admitting low priority SC calls with a variable imposed probability. The incorporation of such a probabilistic framework is considered under a bandwidth-centric approach named probabilistic bandwidth reservation scheme (PBRS). Though equal service times are usually assumed in the literature, the present analysis considers SCs of different service times. By employing Markov chain analysis to treat the independent SCs that correspond to different call specifications, analytical expressions for the call blocking probabilities (CBPs) are derived. The performance of the proposed CAC scheme is studied not only with regard to CBP but, also, taking into account the priorities assigned to different SCs as well as fairness among various SCs and total network throughput. The proposed probabilistic framework allows the dynamic control of network resources considering also priority assignment, fairness and throughput. Analytical results concerning delay tolerant (DT) and delay-non-tolerant (DNT) traffic have been obtained applying the proposed scheme. Moreover, the relevant simulations have verified the accuracy of the proposed analysis.     

Minimising production costs in generation and cogeneration plants

Technological advances, the shift towards deregulation and anti-pollution laws are bringing dramatic changes in the power generation field. Power and heat are increasingly perceived as products to be sold and bought on the basis of cost/profit considerations in a fast changing market. This requires a substantial effort in the field of economic optimisation of plant management, since solutions based on conventional wisdom or past practice are no longer adequate. Due to the complexity of the factors involved, model-based decision support systems are needed for this task. These issues are discussed with reference to two radically different but equally representative examples.     

Probabilistic versus possibilistic risk assessment models for optimal service level agreements in grid computing

We present a probabilistic and a possibilistic model for assessing the risk of a service level agreement for a computing task in a cluster/grid environment. These models can also be applied to cloud computing. Using the predictive probabilistic approach we develop a framework for resource management in grid computing, and by introducing an upper limit for the number of failures we approximate the probability that a particular computing task is successful. In the predictive possibility model we estimate the possibility distribution of the future number of node failures by a fuzzy nonparametric regression technique. Then the resource provider can use the probabilistic or the possibilistic model to get alternative risk assessments.     

Probabilistic cost estimation in advance of production in a computerized manufacturing system through stochastic geometric programming

A mathematical model for stochastic cost optimization has been developed. The model contains cost terms such as inventory cost, penalty cost for due date violation, and the machining cost. The probable range of cost has been estimated by a stochastic geometric program. If an exact solution is desired, a two stage stochastic geometric program has to be solved. This is mathematically tedious and requires great computational effort. However, managers are often concerned with a policy decision which can be based on the probable lower and upper bounds on the cost function. The probability level on the lower bound of cost has been calculated through the theory of error propagation. This approach is explained through an example.     

U-shaped energy loss curves utilization for distributed generation optimization in distribution networks

We propose novel techniques to find the optimal achieve the maximum loss reduction for distribution networks location, size, and power factor of distributed generation （DG） to Determining the optimal DG location and size is achieved simultaneously using the energy loss curves technique for a pre-selected power factor that gives the best DG operation. Based on the network＇s total load demand, four DG sizes are selected. They are used to form energy loss curves for each bus and then for determining the optimal DG options. The study shows that by defining the energy loss minimization as the objective function, the time-varying load demand significantly affects the sizing of DG resources in distribution networks, whereas consideration of power loss as the objective function leads to inconsistent interpretation of loss reduction and other calculations. The devised technique was tested on two test distribution systems of varying size and complexity and validated by comparison with the exhaustive iterative method （EIM） and recently published results. Results showed that the proposed technique can provide an optimal solution with less computation.     

An adaptive genetic algorithm for demand-driven and resource-constrained project scheduling in aircraft assembly

Scheduling of aircraft assembling activities is proven as a non-deterministic polynomial-time hard problem; which is also known as a typical resource-constrained project scheduling problem (RCPSP). Not saying the scheduling of the complex assemblies of an aircraft, even for a simple product requiring a limited number of assembling operations, it is difficult or even infeasible to obtain the best solution for its RCPSP. To obtain a high quality solution in a short time frame, resource constraints are treated as the objective function of an RCPSP, and an adaptive genetic algorithm (GA) is proposed to solve demand-driven scheduling problems of aircraft assembly. In contrast to other GA-based heuristic algorithms, the proposed algorithm is innovative in sense that: (1) it executes a procedure with two crossovers and three mutations; (2) its fitness function is demand-driven. In the formulation of RCPSP for aircraft assembly, the optimizing criteria are the utilizations of working time, space, and operators. To validate the effectiveness of the proposed algorithm, two encoding approaches have been tested with the real data of demand.     

The exploration of fourth generation language program generation to assist in the production of multimedia Computer Aided Learning

Fourth Generation Language techniques are used to provide an environment in which authors with little or no programming experience can easily and quickly create CAL (Computer Aided Learning) material. This approach arises out of collaboration between the French and Spanish Departments and the Computer Science Department in the Multifacet Language Laboratory project which is investigating the incorporation of CAL, interactive video, voice recognition and speech synthesis into the traditional language laboratory.