Abstract:Data-driven product development is a key technology for systems engineering especially for consumer-oriented industries such as the automotive industry. The basic prerequisite for all data driven approaches is data itself. Due to the increasing networking capabilities of modern vehicles， automotive manufactures are able to record and store customer data in the form of internal vehicle bus signals. The challenge in using this data is that it is not designed for external use， but for internal communication to ensure the safety and functionality of the vehicle. Therefore， the main question is how to extract customer needs and consumer-relevant information within this data using the process of data mining （DM）. Consequently， in this paper， a literature review on the aforementioned use case is conducted. Based on the literature research， a DM simulation game is conducted to determine the suitability of existing DM processes in the area of requirements elicitation. Finally， a process extension is proposed that helps to systematically focus the DM process on customer-relevant information and thus accelerate the overall process.

Abstract:Currently established automotive test approaches in the field of vehicle hardware-in-the-loop （HiL） testing use predefined test structures. The test case structure couples a specific test stimulation with an individual test step to a fixed， predefined system validation with pass / fail result. Each test case has its own specific functional focus. If more than one functional aspect is to be tested， the current methods will not work anymore. The aim of future test methods is to evaluate the system as comprehensively and resource-efficiently as possible. If possible， several sub-functionalities should be evaluated at the same time in a randomly generated， realistic environment. A different validation approach is required for the evaluation of randomly generated stimulations sequences. Previously they are unknown to the evaluation environment. It is no longer possible to validate each test step individually. The exact sequence of the stimulation is not known at the beginning of the test run. This should simulate a behavior that is as realistic as possible. For this purpose， a methodology for the generation of system evaluation for randomly generated stimulations was introduced. Combinatorial and model-based approaches were combined to support the creation of system evaluation for the vehicle body domain （describes the passenger functions that can be experienced within a vehicle with its control units and functionalities， no direct influence on driving dynamics） and link them to the corresponding system requirements. The approach supports the existing methods and achieve a wider and deeper coverage of system assessments than a normal test catalogue implementation would give. This will be shown in a proof of concept with a vehicle comfort function on a HiL system at a German car manufacturer.

Abstract:Nowadays， the implementation and application of numerical methodologies for aeroacoustic analysis become increasingly essential for car manufacturers in order to optimize the effectiveness of the vehicle development process. In this paper， a hybrid numerical tool based on the combination of a delayed detached eddy simulation and a finite element model was presented. The finite element model in turn was based on Lighthill’s equation and acoustic perturbation equations. The computational fluid dynamics and the computational aeroacoustics were respectively performed by the software OpenFOAM and Actran. The aeroacoustic behavior of the SUV Lamborghini Urus using different roof spoiler designs was investigated. The numerical simulations were verified against the experimental measurements conducted in the aeroacoustics full scale wind tunnel of the University of Stuttgart operated by FKFS. Furthermore， the main noise generation mechanisms at the spoiler were discussed and the change of pressure fluctuation level on the car surface with respect to a geometry variation was investigated.

Abstract:The aerodynamic resistance of automobile is an important part of driving resistance and an important source of automobile energy consumption. Automobiles are driving in a real road environment affected by the natural wind on the road. In order to accurately calculate and evaluate the aerodynamic drag characteristic and energy consumption of an automobile driving in a real road environment， a calculation method for the wind averaged aerodynamic drag coefficient based on cycle conditions was proposed. First， the distribution characteristic of natural wind near the ground was obtained by analyzing meteorological data. Then， the probability distribution characteristic of the yaw angle of the automobile under different terrain conditions was analyzed. The changes of the aerodynamic drag coefficient of the automobile with the speed and yaw angle were analyzed by numerical calculation. The wind averaged drag coefficient of a given automobile speed was calculated based on the probability distribution characteristic of the yaw angle. Finally， the calculation method of the wind averaged drag coefficient based on cycle conditions was proposed， and the contribution of different terrain conditions and speed ranges on the aerodynamic drag of the vehicle was analyzed. The results show that for the models studied in this paper， the wind averaged drag coefficient based on cycle conditions is higher than that of the zero-yaw condition by 9.2% and 7.3%， and the contribution of urban condition to aerodynamic drag coefficient is less than 5%. The speed range of 40 km/h and above all has a greater contribution to aerodynamic drag coefficient.

Abstract:In recent years， active flow control has been applied to study the drag reduction of vehicles. However， most researches only aim at some simplified vehicle models without wheels. In addition， the amount of drag reduction and net saved energy both still need to be further improved. In this paper， based on steady blowing on the base， the active drag reduction is numerically studied on three square-back Ahmed models， i.e.，the original Ahmed model，the Ahmed model with stationary wheels， and the Ahmed model with rotating wheels， respectively. First， this paper explores in detail the effect of wheels on aerodynamic characteristics without jet flow. Then， it focuses on the influence of the form of jet groove， jet angle， and momentum coefficient on the drag under wheel conditions. It is concluded that the optimum conditions are as follows： when the form of jet groove is continuous and is close to the edge of the rear of model， the jet angle is 45° and the momentum coefficient is 3%， the corresponding drag reduction and net saved energy can reach 9.5% and 12.7W， respectively.

Abstract:The coefficient of drag （C_d） can be obtained by road coast down test， wind tunnel test， and numerical simulation. Of the three methods， the coast down test is the closest to the actual use condition of consumers， but it is greatly affected by environmental factors. The C_d has a slightly poor repeatability. The wind tunnel test has the best repeatability. The numerical simulation is uncertain of because the choice of turbulence model and grid strategy may have a direct impact on the results. A comparison of the results of the coast down test， the wind tunnel test， and the numerical simulation of 15 passenger cars indicates that taking the coast down test as the comparison benchmark， most of the results are between 0° and 5° yaw angle of the wind tunnel test and the numerical simulation. By discussing the differences between the road driving conditions represented by the coast down behavior， the wind tunnel test， and the numerical simulation， the C_d obtained by the three methods for an electric vehicle model is compared ，and the causes of errors are preliminarily analyzed.

Abstract:Numerical calculations are conducted for the low-resistance electric vehicle model with or without wheels， and the aerodynamic force， surface pressure， velocity field， and other calculation results of the two models are compared to evaluate the influence of wheels on the aerodynamic characteristics of the low-resistance vehicle. The results show that the presence of wheels increases the aerodynamic drag of the vehicle by 63.8%. The components which have the largest contribution are the front wheel chamber， the front of the vehicle body， the base， the underbody （negative contribution） and the rear wheel housing in descending order. The vehicle lift increases more than doubled， mainly due to the contribution of the underbody of the vehicle. The aerodynamic influence of the wheels on the vehicle is reflected in the combined effect of global blocking and local wake caused by the wheel. The wake generated by the rear wheel and the rear guide hood will enhance the strength of the tail vortex ring and reduce the length of the backflow area and the base pressure. A comparison of aerodynamic drag， lift force， pressure distribution， and flow field shows that the existence of wheels will bring significant changes to the aerodynamic characteristics of the vehicles of low wind resistance.

Abstract:The air pressure difference inside and outside the vehicle increases significantly with the increase of vehicle speed， which affects the vehicle seal and leads to an increase in the risk of door seal failure. Once the seal fails， it will produce a strong air suction noise which destroys the sound quality in the vehicle . In this paper， through wind tunnel experiments and simulations， the pressure difference distributions between both sides of the door seal at different vehicle speeds and yaw angles were obtained as input conditions. The finite element simulations of door and door sealing strip were conducted respectively. The door deformation and the sealing strip deformation under the action of the pressure difference were considered comprehensively， and the failure situation was analyzed based on the contact width of the sealing strip and the door sheet metal. The results show that the reduction of pre-compression of the sealing strip caused by door deformation and displacement is the main reason for seal failure.

Abstract:To solve the inaccurate injection quantity problem in multiple injections， in this paper， a simulation model of the electro injector is developed in the AMESim environment through dynamics analysis of injector， and the parameter matching and rationality verification of the model are also conducted. In addition， the influencing factors of fuel injection fluctuation are investigated. The results show that the change of injection pressure， pre-injection pulse width， the dwell time between the main and the pilot injection pulse can affect fuel fluctuation. Moreover， a fuel compensation control strategy based on self-learning system is constructed by using a certain sample of stimulation data to compensate for the deviation of multiple injection quantity. A self-learning system is realized by using the genetic algorithm （GA） and neural network， which can effectively improve injection control.

Abstract:Hydrogen is an ideal engine fuel. Pure hydrogen engines do not produce CO and HC emissions but face the high NOx emission problem. Inner-engine control and outer-engine control are two ways to decrease the NO_x emission. Outer-engine control mainly reduce NO_x emission through selective catalytic reduction （SCR）， which has been well studied. However， there are few studies on NO_x emission control of pure hydrogen engines through inner-engine control. In this paper， the closed homogeneous reactor （CHR） in Chemkin Pro was used to simulate the main inner-engine NO_x emission control in pure hydrogen engines. The results show that single exhaust gas recirculation （EGR） decreases NO_x emission by 45.3% at an EGR ratio of 20%， indicating that the NO_x emission is not significantly reduced. However， EGR plus lean-burn decreases NO_x emission by 96.31% at a λ of 1.4 and an EGR ratio of 20%， achieving ultra-low NO_x emission of pure hydrogen engines. Compared with single EGR and EGR plus lean-burn， SNCR are better for NO_x emission control. A NH₃ ratio of only 10% can decrease NO_x emission by 96.32% on pure hydrogen engines， while a NH₃ ratio of 15% can achieve zero NO_x emission on pure hydrogen engines without a large λ value and EGR ratio. However， it is necessary to accurately control the NH₃ ratio in the cylinder， otherwise it is easy to produce residual NH₃ which can pollute the environment.

Abstract:The demand for CO₂ reduction is rising sharply nowadays， especially for gasoline engines. Considering a life cycle analysis， the energy carrier， i.e. the fossil fuel is responsible for the emissions problem. The defossilization towards synthetic fuels from renewable energy sources （eFuels）， can make the combustion engine almost CO₂ neutral. The design of an eFuel composition and properties is crucial， as its formulation influences the gasoline engine processes and efficiency. From injection， mixture formation and combustion to post-oxidation and exhaust after treatment， a change in fuel composition has significant effects. For these reasons， several research projects were conducted as part of a collaboration between Dr. Ing. h.c. F. Porsche AG and FKFS. Measurements of a single-cylinder engine test bench and of a spray test ring with different fuels were produced and used to calibrate 3D-CFD simulation models.The fuel formulation， mixture formation and combustion behavior were analyzed deeply， with the aim of increasing engine efficiency while improving emissions. A virtual optimization of the engine configuration was possible in addition to single-cylinder engine tests， leading to significant potentials through alternative fuels and engine optimization.

Abstract:Hydrous ethanol has the advantages of high-octane number and valuable oxygen content， and can reduce the energy consumption and emission in the production process. By application of high-speed cameras and constant volume bomb system， the spray characteristics of 95% hydrous ethanol was studied at different temperatures and fuel injection back pressures with a direct injection gasoline injector with five holes. The spray shape， penetration distance， spray cone angle， spray width， spray projection area， and the flash boiling phenomenon were analyzed and compared with those of pure gasoline. The results show that the penetration distance of hydrous ethanol spray is larger than that of gasoline in cold state， decreases with the increase of back pressure， and increases with the increase of oil temperature. The spray cone angle is smaller than that of gasoline and decreases with the increase of back pressure. In the complete flash boiling state， the spray penetration distance increases， the spray cone angle decreases significantly， the spray width increases in the area near the nozzle， and the spray width decreases in the area far away from the nozzle. Simultaneously， the spray projection area is mainly affected by the spray back pressure.

Abstract:An electric compressor generates a mechanical vibration that is transmitted from the engine to the vehicle body and then the vibration of the vehicle body creates a vehicle interior noise. The noise can be calculated by a transfer force from the electric compressor to the engine and transfer characteristics from the electric compressor to the vehicle cabin. The transfer force can be measured with a force sensor installed between the electric compressor and the engine. However， the force sensor interferes with other components and changes the vibration mode of the electric compressor. Therefore， the conventional force sensor cannot measure the transfer force of the actual phenomenon. In this paper， a new low-profile 3-axis force sensor was developed by use of piezoelectric quartz crystals， which achieved measuring the transfer force in as close to the actual phenomenon as possible.

Abstract:Electric motors with very high speeds are used widely for highly automated and autonomous driving. A low weight design of high-speed electric motors can reduce the demand for environmentally harmful materials. However， due to the regular involute gears， which are usually used with high-speed motors to bring the rotation speed and output torque into target values， a noticeable and extremely annoying tonal noise occurs， which compromises the comfort of passengers significantly. Thus， an innovative low-tone gearing with irregular design of tooth is introduced to reduce the noise of gear transmissions. To obtain a low-tone gearing design for an electric push-rod system， which operates the tailgate of vehicles to either open or close， an optimization is conducted. Then for comparison， the sound pressure of the low-tone gearings and a regular design is measured during the operation of the push-rod system. Finally， the reduction of noise by using the low-tone gearing， especially from the aspect of the tonality， is validated by analyzing the measured data.

Abstract:An online curvature continuous parking path planning method that has no requirement on parking initial posture is proposed to balance the accuracy of parking final posture， path quality， and computation efficiency. The entire path planning process is divided into two parts. An optimization-based approach is used within the parking lot for piecewise planning to improve the accuracy of final posture and reduce the number of adjustments. An adapted hybrid A* algorithm that expands the state node with curvature continuous curve groups is used when reversing into the parking lot， which skips post-process and improves computation efficiency. In the proposed adapted hybrid A* algorithm， a cost function taking into account the path curvature changes and the number of direction changes is designed， and a collision detection method using feature polygon constructed by path geometry feature points is proposed to improve the computation efficiency. Simulations and real vehicle test verifies the effectiveness of the proposed method.

Abstract:Ramp driving poses a big challenge to autonomous vehicles， on which there are potential traffic conflicts between vehicles. Therefore， it is necessary to study ramp scenarios for development and testing. In this paper， typical ramp scenarios are studied based on naturalistic driving data （NDD）. First， three major elements are defined to describe the interaction between vehicles on the ramp， including the initial state （S）， the driving action （A） and the interaction performance （P）. Next， variables to characterize the A and the P are selected to be clustering features， and then 8 kinds of categories are obtained by the K-means clustering method based on the Calinski-Harabasz （CH） index. Then， according to the clustering results， 4 kinds of typical interaction modes are obtained by analyzing the variables above. Afterwards， by analyzing the variables that characterize the S， typical logical scenarios are extracted by the confidence ellipse. Finally， based on the logical scenarios， two concrete scenarios are selected to test and evaluate the autonomous driving system （ADS）. The results show that testing with typical ramp scenarios can reveal the social cooperation capabilities of autonomous vehicles. Therefore， it is effective to generate typical ramp scenarios by clustering analysis based on NDD.

Abstract:To promote safety and fully consider human drivers' acceptance， precise decision-making is realized for automated vehicles under the lane-change scenario in this paper. More specifically， automated vehicles not only decide to change lanes or not but also decide specific microcosmic behaviors， such as lane-change time and expected acceleration. Thus， precise decisions for lane-change are described with three parameters and learned by reinforcement learning. The rationality of such parameter-based precise decisions is shown in two aspects. First， different values of decision parameters will notably influence the planned trajectory， which means other microcosmic behaviors will be a significant uncertainty when they are not precisely decided in the decision-making layer. Secondly， based on the analysis of real traffic data， NGSIM， changeable lane-change time， and expected acceleration are revealed in lane-change behaviors， which is seldom explicitly considered in the decision-making layer of current researches. The decision parameters that include lane-change time and expected acceleration are learned with kernel-based least-squares policy iteration reinforcement learning （KLSPI）. Safety， current driver's willingness， and average human driving style are considered in the reward function. Simulation results demonstrate that using reinforcement learning （RL） to learn decision parameters can realize more precise decisions， promote safety performance， and imitate human drivers' behaviors in the lane-change scenario.

Abstract:In order to improve the driving safety and adaptability of intelligent vehicles to application scenarios under extreme conditions such as high speed and large lateral slipping state， the technology of tire-road peak adhesion coefficient （TRPAC） estimation （hereinafter referred to as road estimation） has aroused more and more attention in the field of active safety control. A road estimation method based on lidar is proposed. Based on theory of maximum likelihood， laser intensity distribution model parameters of several typical kinds of structured road surfaces are solved， based on which， the database of typical road surfaces is established. By using Kullback-Leibler Divergence （KLD） to represent the similarity of intensity distributions， road surfaces can be classified according to the established database， and the estimated value of TRPAC can then be mapped. The experiment results show that the proposed method can estimate TRPAC with an accuracy of more than 90%， that the sudden change of road condition can be sensitively detected， and that the proposed method has robustness to different illumination conditions in the day and at night.

Abstract:The road bank angle directly affects the lateral dynamics of the vehicle. Bank angle has become one of the key parameters of the intelligent vehicle stability control system. However， not only the coupling problem between road bank and vehicle roll， but the difficulty of getting lateral force makes the accurate estimation of bank angle a challenging problem. Therefore， an extension fusion road bank estimation algorithm based on the acceleration sensor was proposed in this paper. First， a lateral acceleration sensor model and the roll dynamics model were proposed， and the curve adaptive strong tracking Kalman filter （CASTKF） was used to estimate the road bank. Then， a direct estimation method based on the lateral acceleration sensor was proposed to prevent the wrong estimation after the loss of observability of the system. Next， the extension algorithm was used to fuse the estimated values of the two methods. Finally， hardware-in-loop tests （HIL） were conducted to verify the effectiveness of the proposed algorithm under various working conditions and the results revealed the accuracy and robustness of the CASTKF algorithm.

Abstract:Accurate intention prediction can help intelligent vehicles better understand the environment and make safe decisions， thus improving the safety of automatic driving and promoting cooperative driving. This paper proposes an interactive intention prediction method， which makes a more accurate prediction of the driver's future intention. First， the Hidden Markov Model （HMM） and Gaussian Mixture Model （GMM） are combined to establish a behavior recognition model， which fully considers the surrounding scene information and accurately judges the current driving behavior. Then， an intention-based trajectory prediction method is proposed to plan the best driving trajectory considering the complex and changeable characteristics of the traffic scene， and the maximum expected utility theory is used to infer the future driving behavior. The behavior recognition and intention reasoning models comprehensively consider the evolution process of the traffic situations and the interaction between vehicles， and the final predicted driving intention of vehicles is obtained by combining the results of the above two models. Finally， the proposed method is verified in the NGSIM dataset， and the results show that the proposed behavior recognition model can recognize the lane changing intention of the vehicle 0.2 to 0.3 seconds in advance. Combined with the future intention reasoning model， it can more accurately predict the future driving behavior of the vehicle and improve driving safety.

Abstract:The visual angles of drivers' gaze points and critical waypoints on an expressway ramp in the driving simulator when the lane centering control system （LCCS） is on were gathered. A model was built to analyze the relationship between visual characteristics and drivers’ preference on the curve-cutting behavior of paths. First， methods to analyze driver gaze behavior by nearing critical waypoints were proposed. Next， the relationship between positions of critical waypoints and driver gaze points on different curve segments was analyzed， and the reason for this relationship was used to predict driver preference was also demonstrated. Finally， 72 statistic indices were designed based on visual characteristics. A logistic regression model of drivers’ preference on curve cutting behavior with selected 8 indices was built. The results show that the estimation model predicts drivers’ relative preference on curve cutting behavior of two paths on the center area of the curve road with a high prediction accuracy. Coefficients in the model reveal the visual mechanism behind curve-cutting behavior preference， which provides the basis for the design of preference estimation methods and lane-centering control systems adaptive to driver preference.

Abstract:Two gear AMT （automatic mechanical transmission） can optimize the working range of drive motor and thus improve the drivability and economy of the electric vehicles. However， in the working process， the accurate identification of clutch kiss-point has a great impact on the gear shift performance. Once the clutch friction plate is worn or the diaphragm spring is tired， the actual working position of the clutch may offset correspondingly. Therefore， it is necessary to identify the position of the clutch kiss-point accurately through self-learning so as to ensure the high-quality shift control performance of the transmission. This paper takes a new type of non-torque interruption two-speed transmission I-AMT （Inverse AMT） of electric vehicle as the research object， aiming at the problems such as the change of the position of the clutch kiss-point after the wear， a self-learning strategy for the position identification is proposed. When the clutch is slowly separated， the accurate position identification of the clutch half joint point is realized， by detecting the change trend of the speed encoder of the driving motor connected to the clutch driving plate. The test shows that under the scenario that the clutch is worn， the proposed strategy can accurately identify the position of the clutch kiss-point so as to adaptively adjust the working state of the transmission， thus ensuring the high-quality non-power interruption shift of I-AMT.

Abstract:This paper， by proposing a wheel-LiDAR method of odometry and mapping（WLOAM）， using wheel encoder， steering encoder，LiDAR， and optional GPS for autonomous vehicles， estimates the low-drift pose at real-time and builds a high-accurate map. The odometry consists of the wheel odometry algorithm and the LiDAR odometry algorithm. The former estimates the 3-DOF ego-motion of LiDAR at a high frequency based on Ackermann steering geometry， whose resulting pose increment is applied in point clouds de-skewing and works as a fine initial guess for LiDAR odometry while the latter performs the 6-DOF scan-to-map LiDAR pose optimization at a relatively low frequency to compensate the pose error accumulated by the wheel odometry， whose core is a two-stage method with an angle-based metric for extracting features. The mapping method is based on the factor graph consisting of the LiDAR odometry factor， the loop closure factor， and the optional GPS factor， which is solved via incremental smoothing and mapping （iSAM） to produce a global map online. An auto-aligned-GPS-factor is proposed for fusing GPS measurement incrementally without explicit initialization. The proposed method was extensively evaluated on the datasets gathered from the autonomous vehicle platform and compared with related open-sourced works. The results show a lower drift rate， which reaches 0.53% in the largest test described in this paper. The implementation of the proposed method is open-sourced for communication （https：//github.com/Saki-Chen/W-LOAM）.

Abstract:To enhance the functions and improve the safety of the new generation of vehicles， this paper collected abundant history data of vehicles and then created a rule-based model by using machine learning methods， so as to detect the faulty vehicle in a fleet. Several steps were designed for detailed illustration， and the validation of the method was conducted through electrical fault of the LV （lithium-cobalt） battery. The results can be used as input for the test bench tests of the following vehicle generations.

Abstract:As a new choice of transportation， shared electric vehicles have a broad prospect and market. It is of great significance to study their traveling data and understand their traveling laws. In this paper， a method was designed to extract the effective data based on the actual scene， which was used to collect the real trip data of shared electric vehicles and to reveal the traveling characteristics of shared electric vehicles. The basic traffic data were collected and counted， the simulation models of intersections were built for morning and evening rush hours and other times， and the method to predict the maximum traffic volume of shared electric vehicles at intersections and road sections was designed. The simulation and prediction experiment was conducted， which validated the analysis result of the traveling characteristics of the shared electric vehicles to a certain extent. The research results have a guiding significance for relevant operation strategies of time-sharing leasing companies， and can provide users with the choice of traveling time and route.

Abstract:Due to the ambitious climate protection targets of the ‘Paris Agreement’ （2050） and China （2060） and the pursuit of energy independency， an increased technology-independent research and development of energy efficient and CO₂ neutralization methods is needed^［1-2］. Basically， there are three main powertrain concepts for vehicles available： battery electric drive， fuel cell powered electrical drive， and CO₂ neutral sustainable synthetic fuel powered combustion engines. A simplified subsumption of efficiency factors along the process chain from cradle to grave as decision factor does not reflect the most efficient reduction of CO₂ emissions and reduction of fossil energy demand over the timeline. Due to regional limited availability of renewable energy compared to necessary energy supply， a lot of technical and infrastructural problems must be solved in a cost-efficient way in order to enable this large-scale substitution of fossil energy. In addition to the direct use of renewable electricity， green hydrogen and derived synthetic fuels are unique enablers to accelerate the substitution of fossil energy significantly. Synthetic fuels can be implemented short-term and worldwide for the whole existing vehicle fleet （approx. 1，2 billion vehicles） by utilization of the installed energy logistics. The "drop-in" rate can be successively increased （blending of gasoline， diesel， or natural gas） until complete substitution by green hydrogen， synthetic fuels， or direct use of electricity. This paper is presenting both hydrogen powertrains （H₂ combustion engine and fuel cell powertrain） and synthetic fuels， and technical solutions therefor.

Abstract:In order to improve the fuel economy and durability of fuel cell vehicles （FCVs）， an energy management strategy （EMS） for FCVs based on reinforcement learning （RL） and traffic information was proposed. First， a powertrain model was built based on parameters of key components. Then， based on the characteristics of urban road conditions， a traffic model was built in VISSIM and the vehicle driving data and traffic data were extracted. The traffic data was then used as input to predict velocity by using the long short-term memory neural network. Finally， based on the RL algorithm and using the predicted velocity， acceleration， and the state of charge of battery as inputs， the fuel cell system power was used as the output to design the EMS. The simulation results show that the hydrogen consumption per hundred kilometers of the proposed strategy is only 1.27% different from that of the dynamic programming strategy， and fuel cell system average power fluctuation is reduced by 5.01%，effectively improving the fuel economy and durability of the vehicles.

Abstract:The flow rate and back-pressure of the air supply system in the high-power proton exchange membrane fuel cell （PEMFC） are regulated by adjusting the speed of the air compressor and the angle of the solenoid valve， respectively. Due to the nonlinearity and coupling of the multivariable system， the control method and control parameters of the fuel cell system are complicated. For this reason， a strategy combining feedforward and double loop PI is designed to control the flow rate and back-pressure. The speed of the compressor and the angle of the back-pressure valve are matched by calibrating the feedforward table and PI parameters. The results show that the flow rate and back-pressure follow the set value， the flow error is within 1.5 g/s， and the pressure error is within 0.25 kPa. The proposed strategy can coordinately control the airflow rate and back-pressure， and meets the requirements of PEMFC.

Abstract:In order to improve the fuel economy and environmental adaptability of fuel cell vehicles， the energy management and optimization algorithm of fuel cell vehicle is studied based on the equivalent fuel consumption minimization strategy. First， the output power or braking recovery power is obtained by solving the vehicle dynamics model， and the equivalent fuel consumption of the system is calculated and taken as the optimization objective to achieve the most economic power distribution. In order to improve the adaptability to different working conditions， considering the physical meaning of equivalent coefficient， a variable equivalent coefficient concerning the state of charge of battery is formulated for the first time， so that the fuel cell vehicle can better maintain the state of charge and make full use of the spare energy of the battery. The simulation results under WLTC （worldwide harmonized light vehicles test cycle） and CATC （China automobile test cycle） standard conditions show that the proposed equivalent consumption minimization strategy based on the variable equivalent factor can meet the function of reducing fuel consumption and maintaining battery state of charge， which realizes the energy management and optimization of fuel cell vehicle and verifies good adaptability to working conditions.

Abstract:A one-dimensional， non-isothermal， two-phase model was employed to investigate the effects of inlet gas temperature and relative humidity （RH） on the output performance of proton exchange membrane fuel cell （PEMFC）. A cathode catalyst layer （CL） agglomerate sub-model was also coupled into the model to consider the impact of the micro-structure in CL. It is found that the influence of operating temperature on the output performance of PEMFC is limited at the dry case （RHa50%/RHc50%）. The highest current density at various operating temperatures corresponds to different RH conditions. When the operating temperature is 90℃， the highest current density is attained at the RHa 90% / RHc 50% case， while a lower inlet temperature （70℃） has the highest current density at the RHa 90% /RHc 90% case. Additionally， when the RHa is constant at 90% and RHc is kept at a low level （0?50%）， the RHc corresponding to the maximum power density is lower with a higher temperature based on particle swarm optimization. Under whole operating conditions， the peak power density can be up to 0.88 W/cm2 when the operating temperature is 90℃ and RHc is 14.6%.

Abstract:In order to characterize the effects of operating temperature， relative humidity， and back pressure on the proton exchange membrane fuel cell （PEMFC） performance at the applied step current， a dynamic transfer （DT） model that accounts for the coupled variation between relative humidity and operating temperature is established， which illustrates the characteristic relationship between membrane electrode parameters and PEMFC performance parameters in terms of operating temperature and relative humidity， and analyses the differences between two operating parameters on the transient response of water transport inside PEM， output voltage， and power density distribution with time at the applied step current. A combination of theoretical calculations and experiments is proposed. First， the DT model including geometry and meshes is embedded into Fluent and the data in the solver of Fluent is called to simulate the electrochemical reactions. Secondly， experiments are conducted regarding the effect of temperature（50 ℃， 60 ℃ and 70 ℃）， back pressure（0 and 10 kPa）， and relative humidity（50%， 75% and 100%） on the transient response like voltage and power density. Finally， the transient response for voltage output， water transport inside the PEM， and power density distribution with time via polarization curves， contours， and I-P curves with time are compared and analyzed. The results indicate that the DT model is in good agreement with the experimental data observed， which elucidates that the anode relative humidity determines the capability of power density to operate steadily after the applied step current and the power density response time and the amplitude of undershoot are related to membrane. When the back pressure is 10 kPa，the anode relative humidity is 75%， and the cathode relative humidity is 100% at an operating temperature of 60 °C， PEMFC has the best dynamic performance.

Abstract:The dynamic properties of thin-film temperature sensors with different sizes are investigated in detail through numerical simulation and system identification modeling. A one-dimensional transient heat transfer model for the sensor is built based on its location in the proton exchange membrane fuel cell （PEMFC）. The dynamic mathematical model， dynamic performance indicators， and dynamic error are obtained by employing COMSOL simulation and the system identification method. Notably， several significant dynamic parameters including working frequency bands， delay time， rise time as well as dynamic error peak， are determined for insulation layers of 1 μm， 2 μm， 3 μm， 5 μm， and 10 μm thick， and a real thin-film sensor is fabricated and calibrated. The results demonstrate that the sensor dynamic performance reduces with the growth of the insulation layer thickness. This paper reports a novel method to identify whether a thermal probe can capture the internal dynamic temperature variety of PEMFC， thus benefiting the further development of thermal probe on the research for PEMFC dynamic temperature variation under transient conditions， which is likely to inspire the sensor design contained physical parameters selection and structural design.

Abstract:The purge of fuel cell during shutdown is very important for cold start. The purge of proton exchange membrane fuel cell was experimentally studied. The effect of current， flow rate， and single cell temperature on purge was studied by using the method of purge with load. The results show that the voltage of the single cell reduced effectively when purge with load. The voltage decreases continuously with an increase in the load current. With the increase of the purging time， the voltage of the single cell continues to decrease， which can effectively avoid the high potential of the single cell for a long time. In addition， increasing the purge gas flow rate and the cell temperature can increase the purge rate， both of which have obvious effects on the purge rate. Under the conditions of experiment， the anode inlet flow rate was 0.34 L/min1， the cathode inlet flow rate was 1.32 Lmin1， and the load current density was 0.04 A/cm2， which could meet the purging goal of a short purging time， less hydrogen consumption， and lower cell voltage during purging.

Abstract:As the core component of electric vehicles， the battery box plays a role in supporting and protecting the battery pack. The strength of the overall structure directly affects the safe driving of electric vehicles. In this paper， a static and dynamic analysis of the battery box of an electric vehicle is conducted. The static analysis results show that the stress of the battery box is relatively small， much smaller than the yield strength of the material， and the bottom thickness of the battery box is too conservative. The dynamic analysis is obtained under the extreme operating conditions of the vehicle. The distribution of stress and displacement of the battery box show that the maximum displacement is located at the bottom of the box. A modal analysis is also performed to analyze the vibration response characteristics of the battery box under different vibration sources. The results show that the first 6 modes of the battery box for different roads are mainly manifested as the local vibration of the upper cover of the battery box. Within the safe range， according to the above analysis results， the structure of the battery box is optimized. Under the condition of the strength and rigidity of the battery box， the weight of the optimized battery box is reduced by 25.54%， achieving the design goal of lightweight.

Abstract:Subzero environment results in the accelerated performance degradation， lifespan shortage， lithium planting， and even some severe problems such as internal short circuit of lithium-ion battery cells. Therefore， preheating has become a critical issue for electric vehicle application in the low-temperature area. The consistency between the battery cells has a significant impact on the module performance and aging mechanism. In this paper， the clustering approach was utilized for battery cell selection. In addition， a hybrid battery thermal management system （BTMS） coupled with positive temperature coefficient （PTC） heating film and cooling plate is proposed， the heating efficiency of the proposed system at the ambient temperature （－40℃）was investigated through computational fluid dynamics （CFD） modelling and numerical calculation. The results indicate that the minimum temperature of the battery module can be heated to a state higher than 0 ℃ after 695 seconds of preheating. Besides， the temperature standard deviation of the battery module gets decreased by 5.9 ℃ compared with the pure PTC heating approach. Therefore， this proposed method efficiently heat the battery module to an operational state within a short heating interval， without much energy cost. The preheating speed reaches 3.56 ℃/min. Moreover， the thermal uniformity gets enhanced.