Application of Chaos Theory in Trucks ’ Overloading Enforcement

Trucks’ overloading is considered as one of the most substantial concerns in road transport due to a possible road surface damage, as well as, are less reliable performance of trucks’ braking system. Sufficient human resource and adequate time scheduling are to be planned for surveying trucks’ overloading; hence, it seems required to prepare an all-aroundmodel to be able to predict the number of overloaded vehicles. In the present researchwork, the concept of chaos theory has been utilized to predict the ratio of truckswhich might be guessed overloaded.e largest Lyapunov exponent is utilized to determine the presence of chaos using experimental data and concluded that the ratio of overloaded trucks re�ects chaotic behavior.e prediction based on chaos theory is compared with the results of simple smoothing and moving average methods according to the well-known criterion of mean square errors. e results have also revealed that the chaotic prediction model would act more capably comparing the analogous methods including simple smoothing and moving average to predict the ratio of passing trucks to be possibly overloaded.


Introduction
Road transportation is a dominant mode of freight transportation in Iran that accounts for about 80% of the freight movement [1].e number of heavy vehicles is continuously increasing on the road network.Overloading is growing up, and it is a ma�or cause for signi�cantly accelerating the rate of pavement deterioration [2].In addition to damage road surface, overloaded trucks endanger other vehicles because of predesigning of vehicles' braking systems particularly in curves and slopes.Having promoted road safety and minimizing road maintenance costs, ordinary road measures imposed on road pavement by axles' load of trucks are made as part of the enforcement process laws [3] is legally controlled by weighing stations.
Relevant studies in the �eld of pavement designs based on truck axle loads are referred in the literature.Turochy et al. [4] developed truck factors for pavement design and axle load distribution models for mechanistic-empirical pavement design using information from weigh-in-motion sites on arterial roads in Alabama.Till also developed a detailed method of wheel load modeling from overload trucks for bridge decks [5].Chia-pei calculated average load factors for combined heavy vehicles and axle load ratios for various types of heavy vehicles, to design bridge standard speci�cation [6].
Although some location allocation techniques can be utilized to improve the efficiency of overloading enforcement [7], in order to minimize overloading, scheduling human resources who are employed staff in the process of checking trucks' axles load in weighing stations should be considered in road enforcement.Improving the efficiency, human resources scheduling would require the utilization of prediction methods.Recent techniques of prediction are observed in the literature by applying chaos theory in the �eld of road traffic.Frazier and �ockelman analyzed traffic �ow data and found it chaotic.eir studies showed that predictions based on chaos theory would have greater predictive power than a nonlinear least-squares method [8].Disbro and Frame demonstrated how the theoretically derived Gazis-Herman-Rothery traffic model [9] is highly chaotic, even though applied to medium or smaller (eight-car) systems [10].Van Zuylen et al. discussed the implications of human behavior, chaos, and unpredictability for urban and transportation planning as well as forecasting [11].Safonov et al. showed that chaotic behavior in traffic can be caused by delays in human reaction [12].Nair et al. analyzed traffic �ow data and characterized it as a chaotic factor [13]. Weidlich demonstrated how random-utilitybased models of relatively simple social behaviors produced chaotic behavior and offer a detailed application of chaos to traffic �ow [14].Metghalchia et al. examined the pro�tability of several simple technical trading rules for 16 European stock markets over the 1990 to 2006 period and concluded that increasing trend of moving average rules has predictive power being able to discern recurring price patterns for pro�table trading, even aer accounting for the effects of data snooping bias [15].
Four methodologies of predicting gas turbine behavior over time have also been utilized by Cavarzere and Venturini [16].ey are used to provide a time prediction when a threshold value will be exceeded in the future.Results presented an aim to select the most suitable methodology that allows both trending and forecasting as a function of data trend over time, in order to predict time evolution of gas turbine characteristic parameters and to provide an estimate of the occurrence of a failure [16].
e main contribution of this paper is to predict the ratio of overloaded trucks based on the concept of chaos theory.Following introduction, some methods of vehicles' axle loads controlling and the well-known prediction methods are brie�y presented in the second and third sections, followed by the description of case study.Analytical process includes checking the presence of chaos behaviors using experimental data, as presented in the �h section together with comparing results consisting of more discussions on the revealed results.A brief discussion of the applied process and some recommendations for further studies are also discussed in the last section.

Goods Transportation Controlling and Trucks Axle Load
Pavement management is an important issue, according to its maintenance and repairing roads surface expenses.Heavy vehicles' axle loads are understood as the most important factor of road destruction.Based on observed researches in the literature, the damage on roads surface is related to the vehicles' axle loads by a nonlinear acceleration rate mainly in a polynomial equation of forth degree [2].be stopped aerwards, requiring the loads adjusted based on regulations.[20].Vehicles passing through the sensors connected to a digital video camera system are detected while data are stored simultaneously.Experimental data have been collected by weigh in motion system in Arak-Qom highway and used to analyze in one month period.ree kinds of trucks including 3 axles 10 wheels, 5 axles 12 wheels and 5 axles 18 wheels have been selected for data analyzing because of high frequency of trucks' volume.

Analytical Process
Aer checking the presence of chaos, analytical process has been done based on two methods of time series prediction and the concept of chaos theory.Simple exponential smoothing is corresponding to the chaos theory with time span 1 day, and moving average method corresponding to the chaos theory considering time spans 3, 5, and 7 days as well as data for 42634 vehicles' axle loads that are collected during the time period of case study.

Presence of Chaos.
A more common technique to determine the presence of chaotic behavior is calculating the largest Lyapunov exponent.Equation (3) determines the largest Lyapunov exponent where  is the observation in   �  is its nearest neighbor, and  is the number of observations.When the largest Lyapunov exponent exceeds 0, the system is chaotic [21]: e ratio of vehicles which are suspicious to be overloaded comparing to passing vehicles is considered as a chaos factor in open interval (0 1).Time span is considered as 24 hours, and data for 30 days were gathered.Eventually the well-known equation of logistic map [22] is utilized to de�ne the ration of overloaded vehicles because of reasonable adaptation to traffic behaviors.Equation (4) de�nes the logistic map [23], where  is the ratio of trucks to be possibly overloaded:

Chaos eory versus Simple Exponential Smoothing.
Analytical process has been done for three types of most popular axle con�gurations while the numbers of trucks which are checked are different.Because of the same period of analyzing, simple exponential smoothing is compared with chaos theory using (Δ  ).In regard to (2) and ( 3) two coefficients of  and  need to be estimated minimizing mean square errors are notated as  best and  best in Table 2.For each type of trucks, parameters have been estimated separately.
Results shown in Table 2 revealed that the mean square error criterion of prediction corresponding to chaos is equal or less than one for simple exponential smoothing.

Chaos eory versus Moving Average Method.
Because of considering the same selected periods, moving average method is compared with chaos theory using (Δ  , 5 and 7).Coefficients  and  de�ned in (2) and (3) are estimated, regarding the minimum mean square errors in similar duration notated as  best and  best in Tables 3 and  4. As it can be observed in Tables 3 and 4, readers are supposed to mention that the concept of chaos theory is more powerful technique for predicting ratio of trucks to be possibly overloaded because of the existing of chaotic behavior in overloading.It must be strictly mentioned that each time span is required to be compared with the same time span.So the results of situation Δt = 3, 5, 7 must be compared with moving average method of N = 3, 5 and 7, respectively.

Summary and Conclusion
In this research work, the concept of chaos theory has been applied to predict the ratio of trucks likely to be overloaded in order to make an appropriate time table of human resources in vehicles' axle loads control.Data gathered in one-month period in the case study have been used for analytical process.
Validation is checked by the results of two well-known methods of simple exponential smoothing and moving average time series.Prediction analyses have been presented based on different time intervals including one, three, �ve, and seven days corresponding to the validation process and method of prediction.e mean square error, most popular criterion of prediction validity, is used to check the validity of proposed method.Discussing results showed that the prediction based on the concept of chaos theory using logistic map equation is more capable than simple exponential smoothing and moving average time series methods.It has been shown, of course, more capable than simple exponential method of prediction when time interval is 1 day (next day), in respect, more powerful than moving average method if time intervals are 3, 5, and 7 days.In brief, chaos theory performs deeply better to predict the ratio of vehicles which are possibly overloaded rather than the other methods of prediction.
Having interested to study in due �eld, it is recommended to focus on the application of chaos theory in the other road traffic measures to disclose which measures are able to be de�ned based on chaotic behavior.
Truck weighing control is made by certain equipments, incorporating static, dynamic, and portable scales.Different axle loads and limitations in the scope of case study are shown in Table1including three most popular axle group con�gurations.
ey include 3 axle-10 wheels, 5 axle-12 wheels, and 5 axle-18 wheels trucks.Maximum axle loads are different from different con�gurations� however, it does not seem necessary if the total weigh is equal to the sum of axle load limitations.Trucks' axle loads are checked in weighing stations by random and if they exceed the authorized limit, overloaded trucks would [19]nential smoothing is either applied to time series data, or to produce smoothing data, for which the time series are sequence of observations.Raw data sequence is oen represented by {  }, and the output of the exponential smoothing algorithm is written as {  }, to be regarded as the best estimate of what the next value of X.When the sequence of observations begins at time  = 0, the most simple form of exponential smoothing is given by (2)[17]where  is the smoothing factor, and 0 <  < .Snyder et al. used exponential smoothing method in sales forecasting for inventory control that has always been rationalized in terms of statistical models that possess errors with constant variances.It is shown that exponential smoothing remains appropriate under more general conditions, where the variance is allowed to grow or contract with corresponding movements in the underlying level[19]: T 2: Coefficients and mean square errors in chaos and simple exponential smoothing.
[18]MovingAverage.Moving average method of prediction is obtained by �rst taking the average of the �rst subset.e�xedsubsetsize is then shied forward, creating a new subset of numbers, which is averaged[17].Equation(1)shows the overall de�nition of moving average method where   is the predicted amount for      based on the previous amounts of real demands, and  is the period.Using moving average technique is observed in the literature while Narasimhan et al. proposed new prediction techniques for MPEG-4 encoded variable bit rate video traffic based on the concept of moving average and extension further using the gradient-descent approach[18]: Experimental data of trucks' axle load have been stored using weigh in motion (WIM) system installed in highways.WIMs are fully automated systems which record all vehicles T 3: Coefficients and mean square errors in chaos theory.