Search

The detection of traffic signs in clean and noise-free images has been investigated by numerous researchers; however, very few of these works have focused on noisy environments. While in the real world, for different reasons (e.g. the speed and acceleration of a vehicle and the roughness around it), the input images of the convolutional neural networks (CNNs) could be extremely noisy. Contrary to other research works, in this paper, we investigate the robustness of the deep learning models against the synthetically modeled noises in the detection of small objects. To this end, the state-of-the-art architectures of Faster-RCNN Resnet101, R-FCN Resnet101, and Faster-RCNN Inception Resnet V2 are trained by means of the Tsinghua-Tencent 100K database, and the performances of the trained models on noisy data are evaluated. After verifying the robustness of these models, different training scenarios (1 – Modeling various climatic conditions, 2 – Style randomization, and 3 – Augmix augmentation) are used to enhance the model robustness. The findings indicate that these scenarios result in up to 13.09%, 12%, and 13.61% gains in the mentioned three networks by means of the mPC metric. They also result in 11.74%, 8.89%, and 7.27% gains in the rPC metric, demonstrating that improvement in robustness does not lead to performance drop on the clean data. more...

In this study, among 180 possible hydraulically interconnected suspension configurations, non-symmetric cases were eliminated and 24 potential configurations were selected for further investigation. A 14-DOF vehicle model capable of generating handling and ride motions is developed. At different manoeuvres and road inputs, the handling and ride performances are investigated for 24 configurations. Six hydraulic parameters are then optimized by the application of genetic algorithm in order to improve the ride. The handling performance is also investigated and results have shown that only four configurations provide better ride and handling performances simultaneously. The bounce acceleration response is shown to be reduced up to 47% for the four selected configurations. The roll and pitch angle responses were also reduced around 3% and 18% respectively. Four optimized configurations were also investigated under two sever ride and handling manoeuvres. It is shown that, for the vehicle with the optimized interconnected hydraulic suspension, the bounce acceleration response for a random road test is reduced up to 27% and the roll angle is reduced up to 18% for a side wind test. more...

Recent studies have recently exploited knowledge distillation (KD) technique to address time-consuming annotation task in semantic segmentation, through which one teacher trained on a single dataset could be leveraged for annotating unlabeled data. However, in this context, knowledge capacity is restricted, and knowledge variety is rare in different conditions, such as cross-model KD, in which the single teacher KD prohibits the student model from distilling information using cross-domain context. In this study, we aim to train a robust, lightweight student under the supervision of several expert teachers, which provide better instructive guidance compared to a single student-teacher learning framework. To be more specific, we first train five distinct convolutional neural networks (CNNs) as teachers for semantic segmentation on several datasets. To this end, several state-of-the-art augmentation transformations have also been utilized in training phase of our teachers. The impacts of such training scenarios are then assessed in terms of student robustness and accuracy. As the main contribution of this paper, our proposed multi-teacher KD paradigm endows the student with the ability to amalgamate and capture a variety of knowledge illustrations from different sources. Results demonstrated that our method outperforms the existing studies on both clean and corrupted data in the semantic segmentation task while benefiting from our proposed score weight system. Experiments validate that our multi-teacher framework results in an improvement of 9% up to 32.18% compared to the single-teacher paradigm. Moreover, it is demonstrated that our paradigm surpasses previous supervised real-time studies in the semantic segmentation challenge. more...

In this study, the focus is on reducing the energy that is consumed by a compressor for providing high air pressure in the reservoir. A new air suspension configuration is presented that is titled active interconnected air suspension with outsourced air pressure. In this configuration, a compressor is used to charge the tank; meanwhile, the air springs are connected. For minor excitation, first, the air flows between air springs to control roll angle and height adjustment. If the situation of body position gets worse, the compressed air tank compensates to keep the body not generating roll angle and bounce. This methodology has a benefit. This configuration conserves compressed air in the tank in minor road elevation. The optimized controllers are designed to control roll angle and bounce, but they determine the outsourced air mass flow rate. For switching between interconnection and outsourced mode, there are some rules defined based on game theory for a trade-off between high dynamical performance quality of the vehicle and reduction of energy consumption. The optimization is done on the rules to keep both aspects minimum as much as possible. A three-axle heavy truck is used, and its performance is under discussion on an uneven rough road. Roll angle is improved progressively in novel air suspension configuration, and the energy consumption is reduced. In the default condition, the roll angle is improved 72% from the passive case and 39% from the conventional configuration. Furthermore, the energy consumption optimized version reduces 14% from the non-optimized case and 46% from the outsourced mode. By importing road power spectral density type E and type G, as the short domain and high-frequency vibrations, to two sides of the truck, it is inferred that the vehicle could remain on interconnection mode entirely without using the compressor. more...

Prediction of available energy storage power is essential for increasing the energy management performance of fuel cell hybrid electric systems (FCHES). A simple yet effective power prediction index is proposed to estimate the supercapacitor state of power. It prevents the supercapacitor’s total depletion in the battery/supercapacitor combination. Modern energy management is equipped with an equivalent consumption minimization strategy. The power prediction index is simple compared with other predictive algorithms while providing excellent efficiency compared with supercapacitor-based management strategies. A supercapacitor-based strategy is presented which extends battery life, at the cost of increased fuel consumption. However, it cannot predict the future low state of charge for the supercapacitor. In such conditions, the battery provides the demand power while fuel cells generate more current. On the other hand, the modern power prediction index energy management strategy significantly increases battery life without adding extra hardware. Moreover, fuel consumption decreased by 15.1 percent. The results show that the modern energy management strategy provides outstanding performance for battery life and fuel consumption compared with other energy management strategies due to its power limitation prediction. more...

This paper aims to optimize a fuzzy logic controller (FLC) active seat suspension applied to an articulated truck semi-trailer seat to improve ride comfort considering the energy consumption of the controller. The proposed truck model is a linear truck with 13 degree-of-freedom (DOF). Two objective functions are defined seat root mean square (RMS) acceleration related to ride comfort and controller RMS force pertaining to the energy consumption of the controller. The Pareto Front is obtained for these two objective functions using the multi-objective optimization method in MATLAB. The optimization is based on the Non-Dominated Sorting Genetic Algorithm (NSGA-II), which has been proposed as a powerful decision space exploration engine based on a genetic algorithm (GA) for solving a multi-objective function problem. Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), which is simple and effective, selects a set of optimal controller parameters. In addition, considering the changes of effective parameters in the truck, including trailer load and location, tyres’ stiffness and the driver’s mass, has been investigated to provide Monte Carlo sensitivity analysis of these parameters on objective functions. Finally, using ISO 2631-1, the ride comfort and controller required force levels before and after optimization are compared. The results of this optimization indicate a significant improvement in ride comfort and controller force which has been different in various conditions of truck speed and road classes. The results have shown that the maximum amount of improvement in ride comfort is about 25% which happens on Class-C road (at the truck speed of 60 km/h), and the control force reduction peaks at around 60%, which occurs on Class-A road (at the truck speed of 60 km/h). The simulation is validated by MSC-ADAMS software. more...

In this study, adding a headrest to the conventional vehicle driver seat is investigated to improve the driver comfort and decrease the driver damages. For this purpose, a conventional biomechanical human body model of wholebody vibrations is provided and modified by adding a head degree of freedom to the body model and a headrest to the seat model. The basic model is in the sitting posture, lumped parameters and has nine DOFs for the human body, on contrary to the proposed model which has ten DOFs. The new human body DOF is the twisting motion of the head and neck. This new DOF is generated because of headrest adding to the driver’s seat. To determine the head discomforts, the Seat to Head (STH) indexes are studied in two directions: horizontal and vertical. The Genetic Algorithm (GA) is used to optimize the STH in different directions. The optimization variables are stiffness and damping parameters of the driver’s seat which are 12 for the basic model and are 16 for a new seat. The integer programming is used for time reduction. The results show that new seat (equipped by headrest) has very better STH in both directions. more...

The pedal-assist electric bike is interesting equipment because of using these two power sources...

A wavelet-based power management system is proposed in this paper with a combination of the battery and ultracapacitor (UC) hybrid energy storage system (HESS). The wavelet filter serves as a frequency-based filter for distributing the power between the battery and UC. In order to determine the optimal level of wavelet decomposition as well as the optimal activation power of the wavelet controller, an optimization procedure is established. The proposed frequency-based power management system moderates the usage of battery current, consequently improving its lifetime. Compared with the conventional threshold-based power management systems, the proposed system has the advantage of enhanced battery and UC power management. A LiFePO4 battery is considered and its life loss is modeled. As a case study, an electric motorcycle is evaluated in the federal test procedure (FTP) driving cycle. Compared with a conventional energy storage system (ESS) and a state of available power (SoP) management systems, the results show an improvement for the battery lifetime by 115% and 3%, respectively. The number of battery replacements is increased, and the energy recovery is improved. The 10-year overall costs of the proposed HESS strategy using wavelet are 1500 dollars lower, compared with the ESS. more...

This study investigates the effects of hydrogen enrichment on formation of pollutants and greenhouse gases in gas-fueled spark ignition engines. To this end, a previously validated detailed mechanism is embedded with a comprehensive one-dimensional engine model and the results are verified with experimental observations at different compression ratios and engine loads. The simulations are then performed for different mixtures of hydrogen and natural gas. The results indicate that adding hydrogen increases the cylinder peak pressure at different loads and compression ratios, despite the fact that reduces the amount of carbon monoxide, carbon dioxide, and methane in all the conditions. more...

In this paper, a new rear steering angle control strategy is developed. This strategy is derived from the nonlinear 3DOF bicycle model to improve three-wheeler stability and manoeuvrability. Howeve... more...

This paper develops a hybrid optimization approach combining genetic algorithm (GA) and integer linear programming (ILP) to solve the nonlinear optimization problem of managing the fuel con...

In this paper, a novel optimization model is developed for heavy construction equipment. This approach investigates a complete range of engine operating points, considering fuel consumption and emission maps to accurately model fuel efficiency and level of emissions in different working conditions. As a case study, a tracked bulldozer is investigated on different terrains. The target of the optimization problem is defined as a specific digging or grading depth with minimum fuel consumption and emissions (unburned hydrocarbons, carbon monoxide, and nitrogen oxides) based on EU non-road diesel engine emission standard values. Interaction between the terrain and the bulldozer track is modeled using a semi-empirical method. Dry sand, clayey soil, and snow terrain are considered. The studied bulldozer is a Caterpillar D8T with 233 kW engine power and 8A type blade. The design variables are engine speed, transmission gear number, and throttle position. Genetic algorithm, a famous optimization method, is employed. In order to reduce computational costs, integer programming genetic algorithm is utilized. Due to the complexity of the problem, a constrained nonlinear optimization problem with combined-objectives is developed. Results show that the general trend of fuel consumption and emissions rise as the digging depth increases, as expected. However, this study indicates that the bulldozer traction and digging control can be effectively manipulated by controlling the engine operating point, characterized by engine speed and gear number, to obtain significant improvements in fuel consumption and reduction of exhaust emissions. The results also indicate that re-performing the optimizing problem for different terrain types leads to optimized fuel and emissions targets of up to 77%. more...

This paper studies the direct effects of acceleration transmissibility between the head and seat of the driver, such as driver mass. The six DOF, seat-driver vibration model, is considered. A refuse truck with a specific driving cycle for 20 minutes is selected to evaluate the horizontal driver vibrations. The horizontal acceleration used on the driver is due to acceleration and braking operations, and the amount is considered based on the root mean square of the vibration transfer function for 20 minutes. Besides, suitable values of the vibration parameters are obtained according to the driver mass to minimise the horizontal acceleration. A survey conducted shows that the damping coefficient between the lower part of the driver's body and seat the most effect on the horizontal acceleration acting on the driver. A linear relation between driver mass and backrest parameters and seat pan derived. Also, polynomial equations between driver mass, seat pan, and backrest derived. more...

A hybrid flywheel powertrain system for a bus rapid transit (BRT) is designed in this paper, using a technical-economic study. The modern high-speed flywheel, which is used in the modern heavy vehicle hybrid powertrain, has high efficiency, low weight, and low initial cost. In contrary to the other powertrain hybridisation options, the energy conversion does not need using the flywheel energy storage system. Therefore, the power charge and discharge process are high-speed and efficient. The economic study is defined based on component price and fuel consumption cost. The model of the conventional and the hybrid bus is generated in AVL CRUISE software. Results show that the hybrid flywheel bus has better efficiency in BRT driving cycle in comparison with the normal Tehran City bus driving cycle. The optimal flywheel capacity is 1.4 MJ to use in Tehran BRT, which has ten months payback period. The reduction in fuel consumption is 28.61%. more...

Accurate estimation of the State of Power (SoP) can ensure safe and efficient operation of Lithium-ion batteries in Electric Vehicles (EVs) and Hybrid Electric Vehicles (HEVs). The cell-to-cell variation within a battery pack is a major challenge towards accurate estimation of the SoP, particularly when the cells get aged. This paper presents a hybrid model predictive and fuzzy logic based control system to accurately estimate the SoP for series-connected Lithium-ion cells. The estimation strategy consists of two steps; the power capability for a single new cell is first reckoned under the light of the model predictive control algorithm. The second step is devoted to designation of a model-less fuzzy logic based control system to compensate for the concurrent aging state and State of Charge (SoC) differences among the cells. Accordingly, the present approach only utilizes the actual values of current and cell voltages together with the Electric Circuit Model (ECM) parameters of the new cell, which are identified off-line. Moreover, it benefits from a closed-loop framework which ends in an accurate and reliable SoP estimation. An experimental setup consisting of fresh and aged LiFePO4 cell samples is designed and the extracted data are utilized to verify the proposed estimation method in a feed-forward simulation model for a HEV. The results indicate that the proposed method can estimate the pack SoP accurately while the safe operation for the cells is guaranteed. more...

In this paper‎, ‎a formulation is developed for sizing of a Hybrid Energy Storage System (HESS) in different applications‎. ‎Here‎, ‎the HESS is a combination of Lithium battery and Ultra-Capacitor (UC)‎, ‎which is useful for many high energy and high power applications such as Hybrid Electric Vehicles (HEVs) and renewable energy‎. ‎The sizing formulas are based on initial cost and 10-years battery replacement cost which is arranged as an optimization problem‎. ‎For battery replacement cost‎, ‎the Lithium battery capacity depletion formulas are studied for a LiFePO4 battery‎. ‎As a case study‎, ‎application of HESS in a Series Hybrid Electric Bus (SHEB) is considered‎. ‎The results show by the addition of UC‎, ‎the Lithium battery life is improved significantly‎. ‎Furthermore‎, ‎the optimum sizing of the HESS is dependent to the SHEB driving cycle‎. ‎Therefore‎, ‎considering the power profile of the HESS in its sizing process may reduce HESS cost‎. ‎This effect is studied in three different cycles of the SHEB‎. ‎In addition‎, ‎the formulation is applied to cycle-based optimization of the Power Distribution Control Strategy (PDCS) of the HESS by dynamic programming‎. ‎The results show the optimum PDCS has better LiFePO4 battery life in comparison with the conventional PDCS‎. more...

Noise and vibration are factors affecting the marketability of automobiles. It is required for the drivetain design not to transmit unpleasant noise and vibration to the passengers when driving. As for continuous variable transmission (CVT), noise that is considered to be generated from the metal pushing V-belt has been studied. A research project on this matter reported an estimated mechanism that attributes the noise to self-inducing vibration between the metal pushing V-belt and the pulleys. The self-induced vibration occurs when dynamic friction characteristics exhibit a negative gradient in relation to the sliding speed. However, the effect of the gradient on frequency spectrum of the noise has not been quantitatively elucidated in a consistent manner. In this research, noise at the vehicle start-up phase was studied based on the already reported prediction technology for metal pushing V-belt behavior. In other words, this study focuses on the effect of the friction characteristic gradient between the metal pushing V-belt and the pulleys on each element’s behavior and frequency components of belt pushing force. The results revealed that self-induced vibration on the element V-surface at the exit of the drive pulley excites high-order string vibration at the belt string, generating noise via the transmission case. In addition, it was also found that acoustic energy that excites the high-order string noise is affected by the change in belt pitch diameter at the exit of the pulley, and this determines the magnitude of frequency spectrum of the noise. It was concluded from the above findings that the change in the pitch diameter can be suppressed by narrowing the radial clearance between the shaft and the movable pulley and this measure is effective in reducing noise. more...

Electric Vehicles (EVs) are the best alternative to deal with fossil fuel and air pollution crisis. These vehicles have an electrical propulsion system which is connected to an Energy Storage System (ESS). Internal combustion engine propulsion system (parallel hybrid EV), engine-generator (series hybrid EV), or fuel cell system (fuel cell hybrid EV) may also exist in an EV. High performance ESS is needed as a result of electric power requirement. The batteries, conventional ESS, have high energy and low power specifications for use in EVs. Another candidate for EV's ESS is ultra-capacitor (UC). The ultra-capacitors have very low energy and high power specifications. In recent years, some designs have been proposed to combine batteries and ultra-capacitors to develop a hybrid energy storage system (HESS) with high energy and high power specifications. In this kind of ESS, the power fluctuation of the battery is reduced and the life of the battery is increased. In this study, the behavior of a heavy duty EV with an ESS is compared with a HESS one. This comparison is done in 5 different driving cycles which have different level of aggressiveness. The results show that the vehicle performance, the energy saving characteristics and the battery life criteria improvement values are related to the aggressiveness characteristics of driving cycles. more...