Calibration Of Simulation Models Research Articles

Calibration of microscopic traffic simulation models using metaheuristic algorithms

Abstract This paper presents several metaheuristic algorithms to calibrate a microscopic traffic simulation model. The genetic algorithm (GA), Tabu Search (TS), and a combination of the GA and TS (i.e., warmed GA and warmed TS) are implemented and compared. A set of traffic data collected from the I-5 Freeway, Los Angles, California, is used. Objective functions are defined to minimize the difference between simulated and field traffic data which are built based on the flow and speed. Several car-following parameters in VISSIM, which can significantly affect the simulation outputs, are selected to calibrate. A better match to the field measurements is reached with the GA, TS, and warmed GA and TS when comparing with that only using the default parameters in VISSIM. Overall, TS performs very well and can be used to calibrate parameters. Combining metaheuristic algorithms clearly performs better and therefore is highly recommended for calibrating microscopic traffic simulation models.

An efficient soft computing-based calibration method for microscopic simulation models

ABSTRACTIn recent years, vehicular microscopic simulation models have become one of the main tools used by transportation professionals to analyze transportation policies and projects. Effective use of the existing simulation packages is limited by the calibration of specific parameters based on observed real-life conditions. However, because the calibration of the packages is a resource-intensive process, one might resort to using the default parameter values. In this study, a soft-computing based methodology is proposed that considerably reduces the computation time in comparison to other commonly used methods. The proposed methodology is based on a synergistic combination of artificial neural networks (ANN) and genetic algorithms (GA). First, a Latin hypercube sampling method is used to select representative sets of values for the simulation model's calibration parameters. Second, the effect of each set of parameter values on the simulated traffic stream speed is evaluated. Third, an ANN is trained to determine the relationship between the input parameter values and the output vehicular speed. Finally, a genetic algorithm uses the trained ANN to determine the calibration parameters. Applications of the proposed methodology shows that it allows for less time-consuming calibration of microscopic traffic models compared to other commonly used methods.

Sensitivity Analysis for Building Energy Simulation Model Calibration via Algorithmic Differentiation

High-fidelity building energy simulation models are software tools that are built on the well-established physical laws of thermal/hygric processes to model heating, cooling, lighting, ventilation, and energy use of buildings. They are highly nonlinear systems that involve a large number of subroutine calls and submodel switches during execution. To calibrate a building energy simulation model with good quality, parameter sensitivity analysis is well advocated, since it aims to identify those parameters in a specific building that hold more influence on the building thermal performance than others to facilitate shortening the lengthy cycle of model calibration procedures. Since simulation models are given in a large piece of source codes and encapsulate a series of submodels, the prevailing sensitivity analysis is mostly built within the framework of Monte Carlo simulation and statistics-based random sampling methods. It is computationally intensive. We propose to perform such analysis via a differential sensitivity analysis method that relies on the estimation of derivatives. A key technical challenge is that the high nonlinearity of the model prohibits any analytical differentiation, while numerical differentiation is too sensitive to step size and suffers from a truncation error. We, hence, propose to adopt an algorithmic differentiation method, which exploits the operator overload feature of object-oriented programs, to obtain accurate and robust numerical estimations of derivatives in an automated and computationally efficient way.

Robust design optimization of fixed-fixed beam piezoelectric energy harvester considering manufacturing uncertainties

The piezoelectric energy harvester, which accumulates electrical energy from ambient vibration energy has emerged as an alternative energy source because of its eco-friendly characteristics. Numerous studies have been conducted on the basis of experimental methods to increase the power generating capabilities of the piezoelectric energy harvester. Recently, research using optimization techniques to improve the performance has been carried out. However, since the previous studies did not consider various uncertainties that exist in the energy harvester, it is difficult to overcome the variance of performance caused by the uncertainties. In order to obtain the highest performance from the piezoelectric energy harvester and simultaneously retain robustness while installed under the road pavement, this paper presents robust design optimization of piezoelectric energy harvester which is insensitive to the uncertainties. Prior to performing the robust optimization, statistical information of the uncertainties that exist in the piezoelectric energy harvester was measured from 30 piezoelectric experiments. Then uncertainty quantification was performed using Akaike information criterion. A statistical model calibration method, used to improve the accuracy of the simulation model for robust optimization, was proposed and implemented, based on the uncertainty data. By using the calibrated simulation model, the piezoelectric energy harvester was evaluated and optimized to achieve maximum performance and robustness simultaneously. Finally, the robust optimum design solution for the piezoelectric energy harvester was verified, showing improved performance and reduced variance.

History matching of dual continuum reservoirs\u2014preserving consistency with the fracture model

Ensemble- and optimization-based parameter estimation is commonly used to calibrate simulation models of fractured reservoirs to measured data. Traditionally, statistical data on small-scale fractures are upscaled to a dual continuum model in a single step, and the subsequent history matching procedure makes adjustments to the upscaled parameters. In this paper, we show that the resulting reservoir models may be inconsistent with the initial fracture description, meaning that the reservoir parameters do not correspond to a physically valid combination of fracture parameters. A number of numerical examples is provided, which illustrate why and when the problem occurs. We utilize an invertible analytical fracture upscaling method, and deviations from the fracture model can thus be quantified in each case. We show that consistency with the fracture model is preserved if fracture parameters are history matched directly, if the relation between inversion variables and fracture parameters is linear, or if an unbiased Bayesian sampling method is used. We also show that preserving consistency is less important if the uncertainty of the fracture upscaling method is large.

Data requirement for effective calibration of process-based crop models

While process-based crop models are increasingly applied in multiple research areas, the role of model calibration and its data requirement have been rarely addressed. We used data for two canola cultivars measured across up to six sowing dates, five plant densities and three years to investigate how different data impacted on the efficacy of calibration of the process-based APSIM-Canola simulation model. A Bayesian optimisation method was used to derive cultivar parameters. Our results showed that model calibration using data from a single or multiple similar seasons would likely result in significant equifinality (i.e. multiple combinations of parameters leading to similar simulation accuracy) and simulation uncertainty. The most effective calibration was to use data from contrasting environments (at least two seasons), particularly with in-season growth measurements. Targeting model calibration to minimise simulation error (NRMSE) any less than the inherent error in measurement (13.5%) appeared to be pointless, because model calibration cannot reduce the simulation error to less than the error in the data used for calibration. Our rigorous calibration and validation procedure helped to identify APSIM’s deficiency to simulate the response of canola growth to changes in plant density. Future modelling work is warranted to quantify simulation uncertainty due to equifinality caused by model calibration against limited data, and improve modelling of density effects of canola in APSIM.

Analytical study of the lane-by-lane variation for lane-utility and traffic safety performance assessment using NGSIM data

Highlights the shortcoming of using aggregate data as a whole with the assumption of similar lane-use-pattern over lanes, traffic variations are attentively investigated to explore traffic dynamics distribution both in lane-utility in efficiency and safety performance lane-by-lane. In this paper, above two dimensions are examined for each of five mainline lanes on Highway 101 in Angeles using Next Generation Simulation (NGSIM) data. The lane-specific data is statistically processed to reflect lane utility in lane flow parameters along the location and identify Lane Risk Coefficient by verifying degree of consistence vehicles operation with 'Assured Clear Distance Ahead' criterion. It appears completely new that the most-middle lanes are similar utilized under congested conditions compared to two sides of the lanes, as well as the quite close values of LRC to reveals an uniformly safety level between Lane 2 and Lane 3. The findings of this study are intended to be relevant with the calibration of microscopic simulation models and suggest implementing a proper variable speed towar d under-utilized shoulder Lane 5 but with worst safety performance.

Notes on Using Simulation-Optimization Techniques in Traffic Simulation

Abstract Mathematical and simulation models of systems lay at the core of many decision support systems, and their role becomes more critical when the system is more complex. The decision process usually involves optimizing some utility function that evaluates the performance indicators measuring the impacts of the decisions. The complexity of the system directly increases the difficulty when the associated function to be optimized is a non-analytical, non-differentiable, non-linear function that can only be evaluated by simulation. Simulation-optimization techniques are especially suited to these cases, and its use is becoming increasingly used with traffic models, which represent an archetypal case of complex, dynamic systems that exhibit highly stochastic characteristics. In this approach, simulation is used to evaluate the objective function, and it is combined with a non-differentiable optimization technique for solving the associated optimization problem. Of these techniques, one of the most commonly used is Stochastic Perturbation Stochastic Approximation (SPSA). This paper analyses, discusses and presents the computational results from applying this technique in the calibration of traffic simulation models. This study uses variants of the SPSA by replacing the usual gradient approach with a combination of projected gradient and trust region methods. A special approach has also been analyzed for parameter calibration cases in which each variable has a different magnitude.

Design Guidance for J-Turns on Rural High-Speed Expressways

Owing to the J-turn’s safety effectiveness, it has become a viable alternative to replace high-crash, two-way, stop-controlled intersections on high-speed expressways. National guidance on the design of J-turns on high-speed highways is limited. What is the safety effect of spacing between the intersection and the U-turn? Under what circumstances is the provision of acceleration lanes recommended? This study answered these questions through a safety assessment on the basis of ( a) an examination of crashes that occurred at 12 J-turn sites in Missouri and ( b) a simulation-based assessment of the effect of various design variables and traffic flows. The crash review revealed the proportions of the five most frequently occurring crash types at J-turn sites: ( a) major road sideswipe (31.6%), ( b) major road rear-end (28.1%), ( c) minor road rear-end (15.8%), ( d) loss of control (14%), and ( e) merging from U-turn (10.5%). The crash rates, accounting for exposure, decreased with the increase in the spacing to the U-turn for sideswipe and rear-end crashes; J-turns with a spacing of 1,500 ft or greater experienced the lowest crash rates. The crash rates were lower for J-turn sites with acceleration lanes for minor road traffic merging onto the major road than for sites without acceleration lanes. A calibrated simulation model analysis revealed that the presence of acceleration lanes reduced conflicts for all volumes and designs, including low volumes. The simulation analysis also reinforced the crash analysis results that safety improved with an increase in spacing between the main intersection and the U-turn.

Comparative study of simulated annealing, tabu search, and the genetic algorithm for calibration of the microsimulation model

Microsimulation modeling is one of the contemporary techniques that has potential to perform complex transportation studies faster, safer, and in a less expensive manner. However, to get accurate and reliable results, the microsimulation models need to be well calibrated. Microsimulation model consists of various sub-models each having many parameters, most of which are user-adjustable and are attuned for calibrating the model. Manual calibration involves an iterative trial-and-error process of using the intuitive discrete values of each parameter and feasible combinations of multiple parameters each time until the desired results are obtained. With this approach, it is possible to easily get caught in a never-ending circular process of fixing one problem only to generate another. This can make manual calibration a time-consuming process and is suggested only when the number of parameters is small. However, when the calibration parameter subset is large, an automated process is suggested in the literature. Amongst the meta-heuristics used for calibrating microsimulation models, the genetic algorithm (GA) has been widely used and simulated annealing (SA) has been used only once in the past. Thus, the question of which meta-heuristics is more suitable for the problem of calibration of the microsimulation model still remains open. Thus, the objective of this paper is to evaluate and compare the manual and three (the GA, SA, and tabu search (TS)) meta-heuristics for calibration of microsimulation models. This paper therefore addresses the need to examine and identify the suitability of a meta-heuristics for calibrating microsimulation models. The results show that the meta-heuristics approach can be relied upon for calibrating simulation models very effectively, as it offers the benefit of automating the cumbersome calibrating process. All three meta-heuristics (the GA, SA, and TS) have the ability to find better calibrating parameters than the manually calibrated parameters. The number of better solutions, the best solution, and convergence to the best solution by TS is better than those by the GA and SA. Significant time can be saved by automating calibration of microsimulation models using meta-heuristics. The approach presented in this research can be used to help engineers and planners achieve better modeled results, as the calibration of microsimulation models is likely to become more complex in the future.

Investigating the driver’s response time at signalized intersections

It is generally accepted that, behavioural variables may show differences in different countries, and sometimes in different regions depending on cultural and other socio-economic factors. In literature, several researchers have emphasized that performance of signalized intersections is highly related with drivers’ behaviour. However, only a few parameters can be used to reflect drivers’ behaviour in performance analysis. Start response time is used as a key parameter in calibration of analytical and simulation models. In this study, a detailed analysis is conducted on the variability of start response time with respect to parameters like those that manoeuvre type, cycle time, gender of drivers’, etc. by using data obtained from signalized intersections in İzmir, Turkey. Analysis showed that left and right turning drivers have slightly shorter start response times than all through passing vehicles. In addition, the effect of start response time on base saturation flow is discussed.

Analysis of uncertainty indices used for building envelope calibration

Nowadays there is a growing concern about climate change and the global warming effect produced by the concentration of greenhouse gases (GHG). At the Paris climate conference (COP21), 195 countries adopted a global climate agreement, limiting global warming to well below 2°C. Buildings are large producers of GHG and therefore international standards such as ISO 50001 focus on improving their energy performance, including energy efficiency, use and consumption. To achieve this goal it is important to have a detailed knowledge of the thermal behaviour of buildings. The International Performance Measurement and Verification Protocol (IPMVP), proposes a calibrated simulation model (Option D) to gather this knowledge and to determine the savings associated with Energy Conservation Measures (ECMs).This paper improves the calibration methodology proposed by Ramos et al. (2016) [1], solving the limitations regarding the number of thermal zones and the use of free-floating time periods. Through a real case-study that guides the process, the paper explains how to achieve a calibrated Building Energy Simulation (BES) model using an optimisation process based on a meta-heuristic strategy (genetic algorithm - NSGA-II). Different uncertainty indices such as: CV(RMSE) and Goodness of Fit (GOF) are used as objective function to obtain the calibrated model. These indices, frequently used to measure the accuracy of models, are combined to provide a double possibility to find the best solution, as they are an objective function and a model accuracy measure.

Spectroscopic Determination of Leaf Nitrogen Concentration and Mass Per Area in Sweet Corn and Snap Bean

Rapid nondestructive measurements at leaf level of nitrogen concentration (%N) and leaf mass per area (LMA) are needed to improve crop simulation model development and calibration, and better understanding of in‐season N management. Many contact reflectance‐based techniques for %N and LMA estimations require calibration across species, cultivars, growing stages, and cultural practices. Narrowband (hyperspectral) reflectance spectroscopy, in combination with partial least square regression (PLSR) models, offers improved performance over vegetation indices derived from standard linear regression analysis with simple ratios or combined formulas. Little research on the application of contact spectroscopy data and PLSR techniques has been conducted for sweet corn (Zea mays L.) and snap bean (Phaseolus vulgaris L.). In this study, we sought to determine the optimum wavelength ranges for %N and LMA estimations and to develop and evaluate spectroscopic models in estimating %N and LMA in the two species. Best PLSR predictions utilized 1500 to 2400 nm for %N estimations and 450 to 2400 nm for LMA estimations in both species, with high averaged coefficient of determination (R2) values (0.90–0.95 for cross‐validation, 0.52–0.71 for external validation) and low root mean square error (RMSE, reported as percentage of sample data range) (4.59–5.67% for cross‐validation, 9.50–18.46% for external validation). The results indicate that narrowband reflectance spectroscopy (450–2500 nm) combined with PLSR analysis is a promising method for rapid and nondestructive estimates of %N and LMA at leaf level in sweet corn and snap bean across growth stages, N management, and years.Core Ideas Spectroscopy combined with partial least square regression models is useful for estimates of leaf nitrogen concentration and leaf mass per area. The optimum spectra for nitrogen concentration and leaf mass per area estimations were 1500 to 2400 and 450 to 2400 nm. Spectroscopy appears appropriate for routine agronomic research field experiments.

A calibration procedure for increasing the accuracy of microscopic traffic simulation models

Efforts to address operational issues in transportation have been the focus of many research efforts. A number of these efforts were geared toward developing microscopic traffic simulation models to accurately represent the complex and dynamic operation of a transportation network. One of the challenges with such models is that they do not always adequately reflect field conditions—particularly when representing traffic operations across different time periods. This paper presents a robust calibration procedure that aims to increase the accuracy of calibrated microscopic traffic simulation models. This procedure is based on a Monte Carlo approach to generate candidate parameter sets, which are aimed to produce calibrated simulation models. Model runs of these parameter sets are evaluated against robust calibration criteria, including startup and saturation flow characteristics and travel time distributions. The parameter sets that satisfy these criteria are considered as adequately calibrated to accurately reflect field performance measures. In applying this procedure, the results suggest that this approach offers a robust and effective method of calibrating simulation models where disaggregate-level vehicle data are available—which is becoming more prevalent with further advancements in mobile sensor and connected vehicle technologies.

Stepwise deterministic and stochastic calibration of an energy simulation model for an existing building

Building simulation tools have been widely used for performance assessment. However, many studies [1] have reported that a performance gap exists between the reality and simulation output, mainly caused by unknown simulation inputs. Therefore, model calibration needs to be introduced. Calibration attempts can fail for the following reasons: coarse initial simulation model, long sampling time, uncertainty in the model, and sensor errors. The aim of this paper is to address the abovementioned issues. For this study, an existing office building was selected and two calibration approaches were presented: deterministic vs. stochastic. For stochastic calibration, a Gaussian Process Emulator (GPE) was introduced as a surrogate of the EnergyPlus model. The stochastically calibrated model performs better than the deterministically calibrated model. It is concluded in the paper that (1) the calibration quality is influenced by the degree of the details of the initial model, (2) the accumulated measured data under a sampling time of up to one day (e.g. gas energy consumption) might be unsuitable for calibration work due to the lack of ‘time-series trend’, and (3) the calibration quality is also influenced by sensor errors and further calibration needs to take these into account.

Application of tomographic reconstruction techniques for density analysis of green bodies

Progress in the manufacturing of ceramics, but also of sintered metals, strongly relies on the evaluation of the density distribution in green bodies. This evaluation is crucial from many points of view, including the calibration of constitutive models for in-silico simulation of densification processes. To this end, X-ray tomography and other techniques are possible but can be unmanageable for some institutions. Therefore, a destructive method is introduced in the present article to measure the density field of a green body sample using a CNC mill, an analytical balance, and analysis techniques from the field of computational tomography. A virtual experiment is presented where the method is used to reconstruct a simulated green body density field and is found to satisfactorily correspond to the original solution. The green body density field of a truncated cylinder made of alumina powder is evaluated using this method and the reconstructed field is presented.

Automated analysis of pedestrian walking behaviour at a signalised intersection in China

An analysis of pedestrian walking behavior using gait parameters is presented; it uses automated video analysis to collect pedestrian data at a signalized intersection in Nanjing, China. Two aspects of microscopic pedestrian behavior are considered: First, the walking mechanism represented by pedestrian gait parameters. Second, the non-conforming crossing behavior of pedestrians. The effect of various pedestrian related attributes are investigated. The results show high accuracy in automatically detecting pedestrian violations, with an 85.2% correct detection rate. The walking speed and gait parameters for spatial violators are found to be significantly higher compared to non-violators. It is also found that pedestrians who enter the crosswalk during the late stage of the green pedestrian phase often adopt higher walking speed. The gait analysis shows that males tend to have a higher walking speed, walk ratio and longer step length than females. Single pedestrians are found to have higher speed and step frequency compared to pedestrians in groups. The presence of bikes on crosswalks significantly decreases the pedestrian walking speed, step length and frequency, leading to more gait variability. Such results are useful for many future applications such as calibration of simulation models and violation detection.

A multi-agent learning approach for online calibration and consistency checking of real-time traffic network management systems

ABSTRACTConsiderable effort has been devoted to developing real-time traffic network management systems for congestion mitigation in large metropolitan areas. These systems usually adopt high-resolution simulation models to provide real-time traffic network state estimation and short-term prediction capabilities. If the simulation results deviate from their corresponding real-world observations, online adjustment to the parameters of the simulation model is recommended to maintain its consistency. For that purpose, several consistency checking and online adjustment modules could be integrated with the simulation model and activated periodically to maintain the model consistency. This paper presents a multi-agent learning methodology for consistency checking and online calibration of real-time traffic network simulation models. The methodology allows multiple online adjustment modules to learn based on their historical performance. Consequently, an adjustment module is activated only if its activation is expected to reduce any detected model inconsistency. The methodology eliminates computation burden associated with the unnecessary activation of these adjustment modules, which might affect the system’s real-time execution requirement. The performance of the methodology is examined using real-world data. The results show that the methodology is promising as an efficient mechanism for maintaining model estimation consistency.

Innovative Calibration Method for System Level Simulation Models of Internal Combustion Engines

The paper outlines a procedure for the computer-controlled calibration of the combined zero-dimensional (0D) and one-dimensional (1D) thermodynamic simulation model of a turbocharged internal combustion engine (ICE). The main purpose of the calibration is to determine input parameters of the simulation model in such a way as to achieve the smallest difference between the results of the measurements and the results of the numerical simulations with minimum consumption of the computing time. An innovative calibration methodology is based on a novel interaction between optimization methods and physically based methods of the selected ICE sub-systems. Therein physically based methods were used for steering the division of the integral ICE to several sub-models and for determining parameters of selected components considering their governing equations. Innovative multistage interaction between optimization methods and physically based methods allows, unlike the use of well-established methods that rely only on the optimization techniques, for successful calibration of a large number of input parameters with low time consumption. Therefore, the proposed method is suitable for efficient calibration of simulation models of advanced ICEs.

Modelling climate effects on Hungarian winter wheat and maize yields

Hungarian cereal production is situated in the zone of Europe which is most vulnerable to the effects of changes in climatic conditions. The objectives of this paper are to present the calibration and validation of the 4M crop simulation model using farm-level observed representative values, and to estimate the potential yields of winter wheat and maize production for the next three decades. Analysing the differences between the estimated and observed yields, we identifi ed as key infl uencing factors the heterogeneity of technologies and of land quality. A trend of slightly decreasing yields is projected for the next three decades for both cereals. The precise impact of environmental change on crop yields will depend on which climate scenario occurs.