Bayesian Estimation of Economic Simulation Models Using Neural Networks

Computational Economics - Recent advances in computing power and the potential to make more realistic assumptions due to increased flexibility have led to the increased prevalence of simulation...     

Dynamic Submodel Integration Using an Offer-Accept Discrete Event Simulation

Newer approaches for modelling travel behaviour require a new approach to integrated spatial economic modelling. Travel behaviour modelling is increasingly disaggregate, econometric, dynamic, and behavioural. A fully dynamic approach to urban system modelling is described, where interactions are characterized as two agents interacting through discrete events labelled as “offer” or “accept”. This leads to a natural partition of an integrated urban model into submodels based on the category of what is being exchanged, the type of agent, and the time and place of interaction.Where prices (or price-like signals such as congested travel times) exist to stimulate supply and/or to suppress demand, the dynamic change in prices can be represented either behaviourally, as individual agents adjust their expectations in response to their personal history and the history of the modelled region, or with an “auctioneer” from micro-economic theory, who adjusts average prices. When no auctioneers are used, the modelling system can use completely continuous representations of both time and space.Two examples are shown. The first is a demonstration of a continuous-time continuous-space transaction simulation with simple agents representing businesses and households. The second shows how an existing model—the Oregon TLUMIP project for statewide land-use and transport modelling—can be adapted into the paradigm.     

Econometric Effects of Utility Order-Preserving Transformations in Discrete Choice Models

In the random utility modelling context, choice probabilities are unaffected by increasing linear transformations of the systematic utility; hence its empirical specification is derived on the basis that only differences in utility matters and that the scale of utility is arbitrary. We argue that choice probabilities remain unchanged if these linear transformations are made under the deterministic perspective of a single individual choosing several times. But, in the random utility setting, parameter estimates might be significantly affected by these transformations. In particular we focus on the effect of two order-preserving transformations usually applied in the derivation of the representative utility from the conditional indirect utility function: adding a constant to the utility of all alternatives and multiplying each alternative utility by a constant. We concentrate on the two most popular specifications in transport mode choice: the “wage rate” (Train and McFadden Transport Res 12:349–353, 1978) and the “expenditure rate” (Jara-Díaz and Farah Transport Res 22B:159–171, 1987) specifications. Using a collection of synthetic datasets generated in a new fashion directly from the conditional indirect utility function, i.e. before applying any expansion or transformation, we demonstrate how taking this class of order-preserving transformations could lead to misinterpretation of the econometric results, such as detecting randomly distributed and correlated parameters and/or income and time effects which are in fact not present.     

Recursive Estimation and Testing of Dynamic Models

Recursive estimates can be useful for diagnostic purposes, but algorithms for estimating dynamic models recursively with autocorrelated perturbations can be computationally complicated. Thus, we propose a Conditional Recursive Least Squares algorithm (CRLS): given initial full-sample consistent estimates obtained from a correctly specified model, the model is linearized to obtain recursive consistent estimators along the full sample. These may in turn be used to compute statistics to test for structural breaks with unknown break dates. This procedure is illustrated with the Gas-Furnace data.     

On Confounding Preference Heterogeneity and Income Effect in Discrete Choice Models

A substantial part of the discrete choice model literature has been concerned with specifications accounting for preference heterogeneity. In contrast, only a few studies have been interested in the presence of income effect, and it is much rarer still to come across models trying to account for both effects simultaneously. In this paper we use a specification that accounts for income effect and preference heterogeneity, both of a systematic and of a random nature. Our results point out that care should be taken with conclusions drawn from studies that only consider these effects separately, since they can be confounding from an empirical point of view. In fact, we provide empirical evidence that a random cost parameter can account for the existence of income effect.     

Continuous and Discrete Trajectory Models for Dynamic Traffic Assignment

A continuous trajectory model is presented in which transportation networks are represented as topological constructs. The general formulation enhances existing analytic dynamic traffic assignment models by incorporating continuous single-link traffic flow models in a general, coherent, and relatively intuitive manner. Specific exact formulation based on a simplified kinematic wave traffic flow model with physical queues is presented as well.A discrete trajectory model is proposed as an approximation of the continuous model. The discrete model provides wide flexibility in choosing the level of aggregation with respect to time intervals, ranging from several hours, as typical in current practice of long-term travel forecasting models, to one second or less, as in microscopic simulations. An algorithm to find discrete approximate solutions is presented as well as accuracy measures to evaluate them. The effect of time resolution on model performance is examined by a numerical example.     

Automobile-Speed Estimation Using Movie-film Simulation

A silent movie film made by shooting through the windscreen of a car driven at different constant speeds was shown in an auditorium, and subjects asked to estimate the speed of the car. Of the 70 subjects who participated, 35 sat at the front and the remainder at the back.

The speed estimates of those seated at the back were, on average, some 11 per cent higher than those at the front. An explanation for this effect is given. It is shown that those at the correct perspective distance from the screen ( in the present case, those at the front) will receive an impression of speed which most closely resembles the actual speed of the car.

The speed estimates from the front of the auditorium most closely resembled those observed in an earlier field experiment when the subjects' hearing was diminished. It is concluded that the movie technique provides a good simulation of the task in the field experiment if the following two qualifications are made. First, the movie must be viewed from close to the correct perspective distance. Second, a silent movie corresponds to the real situation in which the observer is prevented from hearing.     

Speech Recognition Using Linear Dynamic Models

The majority of automatic speech recognition systems rely on hidden Markov models, in which Gaussian mixtures model the output distributions associated with sub-phone states. This approach, whilst successful, models consecutive feature vectors (augmented to include derivative information) as statistically independent. Furthermore, spatial correlations present in speech parameters are frequently ignored through the use of diagonal covariance matrices. This paper continues the work of Digalakis and others who proposed instead a first-order linear state-space model which has the capacity to model underlying dynamics, and furthermore give a model of spatial correlations. This paper examines the assumptions made in applying such a model and shows that the addition of a hidden dynamic state leads to increases in accuracy over otherwise equivalent static models. We also propose a time-asynchronous decoding strategy suited to recognition with segment models. We describe implementation of decoding for linear dynamic models and present TIMIT phone recognition results     

On Parameter Estimation Using Nonparametric Noise Models

Fitting multidimensional parametric models in frequency domain using nonparametric noise models is considered in this paper. A nonparametric estimate of the noise statistics is obtained from a finite number of independent data sets. The estimated noise model is then substituted for the the true noise covariance matrix in the maximum likelihood loss function to obtain suboptimal parameter estimates. The goal here is to present an analysis of the resulting estimates. Sufficient conditions for consistency are derived, and an asymptotic accuracy analysis is carried out. The first- and second-order statistics of the cost function at the global minimum point are also explored, which can be used for model validation. The analytical findings are validated using numerical simulation results.     

The Effect of Including Irrelevant Alternatives in Discrete Choice Models of Recreation Demand

Computational Economics - We measure bias and efficiency of parameter estimates in the conditional logit (CL) and independent availability logit (IAL) models. Our Monte Carlo experiments consider...     

Worst-Case Simulation of Discrete Linear Time-Invariant Interval Dynamic Systems

In this paper, a new approach to worst-case simulation of discrete linear time- invariant interval dynamic systems is proposed. For stable systems, the new approach solves the problem of worst-case simulation by determining the interval hull enclosing the system states region at every iteration through optimisation. The originality of this approach is that it maintains time-invariant parametric uncertainties during the simulation process. Several previous algorithms have considered the case of parametric time-varying uncertainties (El Ghauoi, L., Calafiore, G.: Worst-Case Simulation of Uncertain Systems, in: Garulli, A., Tesi, A., and Vicino, A. (eds), Robustness in Identification and Control, Springer, London, 1999). However, Cuguer' o (Avoiding Possible Instability in Robust Simulation of Stable Parametric Uncertain Time-Invariant Systems, in: Proceedings of 40th Conference on Decision and Control, Florida, 2001) has presented possible instability problems when simulating a time-invariant uncertain system as if it were time-varying, this result being the motivation for the approach proposed in this paper. The optimisation problem associated with the approach proposed must be solved globally in order to guarantee that the minimum volume box enclosing the region of system states has been derived. In this paper, a global optimisation algorithm based on an interval branch and bound strategy is proposed. Finally, two real application examples are used to test the performance of the approach proposed.     

Identification of Neurofuzzy Models Using GTLS Parameter Estimation

In this paper, nonlinear system identification utilizing generalized total least squares (GTLS) methodologies in neurofuzzy systems is addressed. The problem involved with the estimation of the local model parameters of neurofuzzy networks is the presence of noise in measured data. When some or all input channels are subject to noise, the GTLS algorithm yields consistent parameter estimates. In addition to the estimation of the parameters, the main challenge in the design of these local model networks is the determination of the region of validity for the local models. The method presented in this paper is based on an expectation–maximization algorithm that uses a residual from the GTLS parameter estimation for proper partitioning. The performance of the resulting nonlinear model with local parameters estimated by weighted GTLS is a product both of the parameter estimation itself and the associated residual used for the partitioning process. The applicability and benefits of the proposed algorithm are demonstrated by means of illustrative examples and an automotive application.      

Using Discrete Event Simulation to Analyze Contention Managers

Understanding the behavior and benefits of contention managers is important for designing transactional memory implementations. Contention manager design is closely tied to other design decisions in a transaction memory implementation, and therefore experiments to compare the behaviors of contention managers are difficult. This paper presents a discrete event simulator that allows researchers to explore the behavior of contention managers and even to perform experiments that compare lazy conflict detection without contention management to eager detection combined with a contention manager. For our benchmarks, we found that lazy conflict detection was competitive with the best contention managers. Our experiments confirm that contention management design is critical for transactional memories that use eager validation. We used the simulator to explore new tiered contention managers that combine livelock-prone contention managers with livelock-free contention managers to provide the benefits of the livelock-prone contention manager while avoiding its pathological behaviors under contention.     

Markov离散事件动态系统参数灵敏度估计算法

给出了马尔可夫离散事件系统参数灵敏度估计的高效率仿真算法.既可以用于稳态性能测度的参数灵敏度估计,也可用于终止型性能测度的灵敏度估计.和现有的仿真算法相比,其数学描述和仿真流程比较简洁,易于编程实现.给出的各种仿真算例均验证了该方法的适用性.此外,还指出了马尔可夫系统性能测度参数灵敏度的精确估计为光滑扰动分析(Smoothed PerturbatinAnalysis)和似然比方法(Likelihood Ratios)得到的估计量之和,单独使用其中的任何一种均难以给出系统性能测度参数灵敏度的可靠、一致估计.     

Intelligent Control Paradigm for Dynamic Discrete Event System Simulation

Simulation based control of discrete event systems has been a potential approach to support decision-making in the manufacturing scenario. In this paper, a knowledge intensive simulation modelling approach for a discrete even system is investigated. Based on the proposed simulation model, a robust control mechanism is presented that is believed to add significant value to discrete event dynamic system. The algorithm utilises neural network feedforward control plus robust proportional derivative feedback control to achieve control performance and output stability. The novel simulation approach, as well as the proposed controller, is implemented in an Extend TM environment and the effectiveness and usefulness of the proposed controller are verified, industrially, in the hard disk drive assembly process, a significant component of the Singapore manufacturing economy.     

用YFSIM对混合数据系统仿真

ＹＨ－Ｆ２是国防科大研制成功的全数字仿真计算机系统。配置在其上的ＹＦＳＩＭ是非过程的仿真建模语言，它是专为连续动力学系统的实时仿真和时间要求苛刻的仿真而设计的，但也能够适用于仿真连续／离散混合数据系统。本文结合一个混合数据系统的简单实例，详细论述了如何用ＹＦＳＩＭ语言对混合数据系统建模和仿真。     

基于离散航迹点的舰船航行轨迹仿真估计

最近,航海技术发展迅速,在贸易、军事、科研等众多领域起到越来越重要的作用. 人们对于舰船航行的规划、仿真、监控与跟踪等方面的研究更细致、更全面. 本文主要研究基于舰船航行中跟踪到的稀疏航迹点的航线估计问题,用于态势图的仿真显示. 为了估计出通过所有航迹点、合理、平滑的航线,我们提出了由粗到细的航线估计算法,该算法主要包含三个层级. 提出的算法使用数据量较小的经验航海线路图和全球海陆分布图,这两种图的数据获取容易,并且可实现较快速的航线估计. 大量实验证明,提出的由粗到细的层级算法能较好的完成航线的估计. 同时,提出的算法也可以用于舰船航行前的航线规划. 算法仍然存在一些问题和不足,例如没有考虑海洋水深等问题,这将是我们下一步的研究内容.     

使用STL处理离散事件仿真的事件表

文中对如何使用STL这一高效安全的C 标准程序库处理离散事件仿真的事件表作了介绍,给出了使用STL处理事件表的演示程序,并探讨了使用不同的STL顺序容器处理事件表的性能差异。     

动力锂电池SOC估计仿真研究

针对传统的无迹卡尔曼滤波算法(UKF)估计动力锂电池荷电状态(SOC)时,由于滤波迭代过程中系统噪声不确定,可能导致估计结果精度欠佳的问题,提出一种改进的自适应无迹卡尔曼滤波算法(AUKF)动态地估计锂离子电池的SOC.算法以UKF算法为基础,引入改进的Sage-Husa自适应滤波算法,利用观测数据进行滤波递推的同时,...     

Combined Activity/Travel Choice Models: Time-Dependent and Dynamic Versions

To fully understand and predict travel demand and traffic flow, it is necessary to investigate what drives people to travel. The analysis should examine why, where and when various activities are engaged in, and how activity engagement is related to the spatial and institutional organization of an urban area. In view of this, two combined activity/travel choice models are presented in this paper. The first one is a time-dependent (quasi-dynamic) model for long-term transport planning such as travel demand forecasting, while the other one is a dynamic model for short-term traffic management such as instantaneous flow analysis. The time-dependent model is formulated as a mathematical programming problem for modeling the multinomial logit activity/destination choice and the user equilibrium route choice behavior. It can further be converted to a variational inequality problem. On the other hand, the dynamic model is aimed to find a solution for equilibrium activity location, travel route and departure time choices in queuing networks with multiple commuter classes. It is formulated as a discrete-time, finite-dimensional variational inequality and then converted to an equivalent zero-extreme value minimization problem. Solution algorithms are proposed for these two models and numerical example is presented for the latter. It is shown that the proposed modeling approaches, either based on time-dependent or dynamic traffic assignment principles, provide powerful tools to a wide variety of activity/travel choice problems in dynamic domain.