Table of Content

25 June 2021, Volume 43 Issue 6 Previous Issue    Next Issue

Select

Study on Energy Management Strategy for Parallel Plug⁃in Hybrid Electric Vehicles Considering Battery Electric⁃Thermal⁃Depth⁃of⁃Discharge Shaobo Xie,Kangkang Zhang,Qiankun Zhang,Huiran Luo 2021, 43 (6):  791-798.  doi: 10.19562/j.chinasae.qcgc.2021.06.001 Abstract ( 465 )   HTML ( 20 )   PDF (3873KB) ( 491 )   Save Considering the influence of battery aging in the design of energy management strategy of plug?in hybrid electric vehicles （PHEVs） is of great significance for improving the vehicle economy. An energy management strategy is proposed for parallel PHEVs equipped with automatic mechanical transmission while considering triple coupled factors of electric?thermal?depth?of?discharge which influence battery aging， and model predictive control （MPC） is used to achieve real?time co?optimization of gear?shifting and torque distribution. Firstly， the optimal depth of battery discharge over the entire trip is identified based on historical speed profiles and battery aging dynamics. Secondly， with the optimal depth?of?discharge， the battery discharge planning and reference SOC in the preview horizon are established where the Pontryagin minimum principle （PMP） is leveraged to solve the co?optimization problem. Then， it is compared with the MPC without considering battery aging， one?dimensional MPC （1D?MPC） where a scheduled gear?shifting policy is used， and MPC with the dynamic programming based optimization method. The results show that the following ：（1） Optimizing the gear?shifting for parallel PHEVs can effectively reduce the battery aging cost；（2） Compared with the one?dimensional model predictive control strategy （1D?MPC） with regular shift strategy， the two?dimensional model predictive control strategy （2D?MPC） with simultaneous optimization of gear selection and torque distribution is conducive to reduce the battery core temperature and aging cost；（3） Compared with the rule?based CD?CS strategy， the proposed 2D?MPC method can reduce the total cost （the sum of energy consumption cost and battery aging cost） by 28.2 %. Figures and Tables | References | Related Articles | Metrics

Select

Research Progress in Mass Transfer Enhancement of Flow Field in Proton Exchange Membrane Fuel Cell Yingjie Liu,Ben Chen 2021, 43 (6):  799-807.  doi: 10.19562/j.chinasae.qcgc.2021.06.002 Abstract ( 293 )   HTML ( 9 )   PDF (3974KB) ( 403 )   Save As a new on?board power source， proton exchange membrane fuel cell （PEMFC） has come under the spotlight due to its wide application prospect. Flow field plate is one of the core components of PEMFC， which plays the role of distributing reaction gas， removing moisture and impurities and conducting electrons. At present， most of the research on the flow field of PEMFC has focused on the optimizations of the size of routine flow channel and the layout of flow field， while some studies add different forms of blocks in the flow channel to enhance gas mass transfer， or apply porous materials to flow field plate， or design a new 3D mesh flow field structure to optimize the thermal management of water in fuel cell and intensify the effects of mass transfer for enhancing the performance of PEMFC. These researches are summarized with a few conclusions drawn in this paper. Figures and Tables | References | Related Articles | Metrics

Select

Effect of Drivers’ Acceptance Level of Car⁃following Risk on the Takeover Performance Guangquan Lu,Facheng Chen,Penghui Li,Junda Zhai,Haitian Tan,Pengyun Zhao 2021, 43 (6):  808-814.  doi: 10.19562/j.chinasae.qcgc.2021.06.003 Abstract ( 219 )   HTML ( 13 )   PDF (1802KB) ( 278 )   Save Takeover in automated driving is a kind of manual driving operation with high safety requirements， which may be affected by the driver’s manual driving safety habits. Based on a driving simulator，the manual car?following and an automated driving takeover experiments are designed in this paper to study the influence of the driver’ acceptance level of car?following risk on the takeover performance. Meanwhile， the impact of takeover time budget and visual non?driving related task is also investigated. The results indicate that drivers with a low acceptable level of car?following risk in daily driving have shorter takeover reaction time， and show lower longitudinal collision risk after taking over from the condition of monitoring automated driving. In addition， when drivers are in the state of visual distraction， the 5 s takeover time budget leads to poor lateral and longitudinal stability and high longitudinal collision risk. The results of this study can provide a theoretical basis for the design of personalized automated driving system. Figures and Tables | References | Related Articles | Metrics

Select

Research on Transmission Strategy for Real⁃time Video Stream Under LTE⁃V2X Duanfeng Chu,Zhuoer Wang,Yujie Qiu,Liping Lu,Zhihao Xu,Xinyu Ji 2021, 43 (6):  815-824.  doi: 10.19562/j.chinasae.qcgc.2021.06.004 Abstract ( 333 )   HTML ( 9 )   PDF (1911KB) ( 268 )   Save In order to enhance the video transmission quality in cellular community， to strengthen the visual perception ability of the blind area in intelligent connected vehicles， and hence to improve driving safety and handiness， a reliable strategy for achieving real?time video stream transmission under the environment of long?term evolution?vehicle to everything （LTE?V2X） is proposed in this paper. First of all， for improving the system performance of cellular community and the image quality of real?time video stream transmission， a resource allocation algorithm is put forward for maximizing vehicle to vehicle communication throughput， while meeting the requirements of heterogeneous service quality for V2V user under the state of imperfect information channel. The algorithm adopts the first?order autoregressive model for the modeling of real information channel gain， and the power distribution and the multiple access and pairing of spectra are carried out step?by?step to complete the resource allocation. Then， scalable video coding technique is used to encode real?time video streams， the encoded video data frames having different video quality level are transmitted via network， and based on the perceived data from network， video stream buffer is set up. Finally， adaptive transmission algorithm is designed to calculate the current optimal quality level of transmitted video， and the quality of video images transmitted can be dynamically adjusted according to the communication network environment， so the chance of video playback freezing is effectively reduced. The results of experiment show that compared with video transmission algorithm RALSB having fundamental layer protection and video quality adaptive selection （VQAS） algorithm， the strategy proposed can effectively enhance the quality of real?time video stream and reduce the chance of video playback freezing. Figures and Tables | References | Related Articles | Metrics

Select

Research on Positioning Fusion and Verification Algorithm Based on UKF Xingjia Li,Jianfen Li,Min Zhu,Nengling Peng,Shuai Zuo 2021, 43 (6):  825-832.  doi: 10.19562/j.chinasae.qcgc.2021.06.005 Abstract ( 352 )   HTML ( 18 )   PDF (2996KB) ( 287 )   Save For the problem that the positioning accuracy of the sensor is greatly affected by the environment and it has great impact on the driving safety of autonomous vehicles， a positioning fusion and verification algorithm is proposed. The short?term short?range track calculation model is established for positioning prediction. And the UKF （Unscented Kalman Filter） is used to realize the nonlinear fusion between the predicted positioning result and the actual positioning result so as to improve the positioning accuracy. Whether the actual positioning has shifted can be judged by the positioning verification algorithm. And an active risk avoidance algorithm is studied to enable the vehicle to maintain the trajectory before positioning failure and continue to run until it stops， so as to improve the driving safety of the vehicle. Simulation and experimental results show that the algorithm can not only improve the positioning accuracy， but also effectively judge whether the positioning has offset， and ensure the vehicle to continue to run safely according to the saved reference trajectory after the positioning failure occurs at the same time. Figures and Tables | References | Related Articles | Metrics

Select

A Local Path Planning Algorithm for Intelligent Wheeled Vehicle Combining TangentBug and Dubins Path Xudong Zhang,Fukang Xu,Yuan Zou,Ningyuan Guo,Yu Zhang 2021, 43 (6):  833-841.  doi: 10.19562/j.chinasae.qcgc.2021.06.006 Abstract ( 281 )   HTML ( 15 )   PDF (2839KB) ( 542 )   Save Due to the limitation of environmental conditions， some intelligent wheeled vehicle with Ackermann steering can only obtain local map and location， which makes path planning difficult. To solve the problem， this paper proposes a local path planning algorithm combining TangentBug and Dubins path. Firstly， a set of reference points is established by sampling method. Then， Dubins path is generated as the planning path to satisfy the motion constraints of the minimum turning radius of the vehicle and the desired heading at the target point. Besides， collision checking along the planning path and mode switching rules considering the positioning error are added. Finally， real vehicle tests are conducted with results showing that the proposed algorithm can navigate the vehicle to the target with the required state while meeting the safety and real-time performance requirements of path planning， and eliminate the disturbance from the error of positioning. Compared with using arc curve， the path planned by this algorithm is more beneficial to path following control. Figures and Tables | References | Related Articles | Metrics

Select

Study on Discrepancy Between Drivers’ Favorable LKA Intervention Timing Characteristics and Their Naturalistic Driving Characteristics Jiachen Chen,Hui Chen,Xiaoming Lan,Nishimura Yosuke,Ishihara Atsushi 2021, 43 (6):  842-850.  doi: 10.19562/j.chinasae.qcgc.2021.06.007 Abstract ( 211 )   HTML ( 9 )   PDF (1499KB) ( 596 )   Save Some researches reveal that the prerequisite for LKA system can be adapted to different drivers is the consistency of driver’s favorable LKA intervention timing characteristic with his own naturalistic driving characteristic， but the real?world situation about this is still unknown at present. In view of this， a subjective / objective evaluation test is conducted in this paper to obtain the objective indicators of driver’s favorable LKA intervention timing， and the naturalistic driving data of drivers are collected and the steering return points of drivers are chosen. Then the characteristics of steering return points are quantified with the corresponding indicators extracted. Finally， the indicators of driver’s favorable LKA intervention timing characteristic and his steering return point indicators during naturalistic driving are compared and the results show that for 12.5% of drivers， the favorable LKA intervention timing characteristic is significantly different from his naturalistic driving characteristic. Figures and Tables | References | Related Articles | Metrics

Select

Research on Automatic Emergency Collision Avoidance of Intelligent Vehicle in Extreme Condition by Combined Braking and Steering Control Fei Lai,Chaoqun Huang,Hongliang Dong,Tao Chen,Shulian Zhao 2021, 43 (6):  851-860.  doi: 10.19562/j.chinasae.qcgc.2021.06.008 Abstract ( 307 )   HTML ( 25 )   PDF (3395KB) ( 405 )   Save In order to enhance the automatic emergency collision avoidance ability of intelligent vehicle under extreme working conditions， an integrated control method combined braking and steering is proposed. Firstly， an 18 degrees?of?freedom （DOF） unified dynamic model including the coupling characteristics of steering， braking and suspension subsystems is established， on which a braking on curve simulation is conducted on a level road. Then an overall framework of automatic emergency collision avoidance system combining braking and steering is put forward， in which a fifth?order polynomial programming algorithm is used for path planning， and a sliding mode tracking control is adopted for longitudinal movement while an optimal four?wheel steering tracking control based on two DOF reference model is used for lateral movement. Finally， taking the international standard on double lane change maneuver test under extreme condition as reference， a driving scene of automatic emergency collision avoidance is constructed， and a comparative simulation on above?mentioned 18 DOF model with automatic emergency steering collision avoidance and combined braking and steering collision avoidance at different speeds is carried out. The results show that when the vehicle speed is relatively high， the real trace has a certain lag relative to the ideal trace， and in extreme condition， the collision avoidance is hard to succeed only by steering maneuver， while the combined braking and steering collision avoidance system can further enhance the automatic emergency collision avoidance ability of vehicle in extreme condition， with the maximum passing speed increasing from 50 to 60 km/h. Figures and Tables | References | Related Articles | Metrics

Select

Multi⁃objective Optimization of Active Safety Airbag Based on the Injury Thresholds of 12⁃year⁃old Children Liang Hong,Gang Liu,Ruhai Ge 2021, 43 (6):  861-869.  doi: 10.19562/j.chinasae.qcgc.2021.06.009 Abstract ( 129 )   HTML ( 3 )   PDF (3439KB) ( 204 )   Save This paper aims to conduct a multi?objective optimization on the active safety airbag of a school bus for enhancing its protective effects on 12?year?old children. Firstly， the injury thresholds of 12?year?old children are determined and a coupling model of the school bus and the active airbag is built. Then， the original waveform measured in a sled test is simplified to simulate four collision conditions. Finally， based on response surface surrogate model and using modified NSGA?II algorithm， a multi?objective optimization on active airbag is conducted， with the installation position， the open pressure and the opening of air vent valve， as well as the middle strap length of the active airbag selected as the optimization parameters， and the weighted injury criterion for 12?year?old children （WICC） as the optimization objective. The results show that the optimization on the active airbag achieves significant effects under four collision conditions of school bus. For the condition with the highest collision speed， the head injury criterion HIC15， the neck injury criterion Nij， the thorax resultant acceleration T3ms， the chest deflection THPC and WICC reduce by 28.58%，14.79%，10.02%，10.26% and 18.08% respectively. Figures and Tables | References | Related Articles | Metrics

Select

Reliability Optimization Design of Occupant Restraint System Based on GWO⁃KRG Surrogate Model Xianguang Gu,Menglin Gao,Xiaole Wang,Yuezhu Huang 2021, 43 (6):  870-876.  doi: 10.19562/j.chinasae.qcgc.2021.06.010 Abstract ( 173 )   HTML ( 5 )   PDF (1700KB) ( 211 )   Save In order to enhance the safety performance of the occupant restraint system， the parameter optimization technology for surrogate model is applied to the reliability optimization design of the restraint system in this paper. Firstly， a simulation model for the driver?side restraint system of a vehicle is established and verified by real vehicle crash test. Then， the grey wolf optimization （GWO） algorithm is used to optimize the correlation parameters of Kriging （KRG） model， so a high?accuracy GWO?KRG surrogate model is obtained. Finally， based on GWO?KRG surrogate model， a reliability optimization is conducted on the restraint system. The results show that GWO?KRG surrogate model can provide more accurate predicted response， and after reliability optimization the safety performance of the restraint system is improved with its reliability also guaranteed. Figures and Tables | References | Related Articles | Metrics

Select

Human⁃vehicle Steering Collision Avoidance Path Planning Based on Pedestrian Location Prediction Caixia Li,Shaobo Lu,Bohan Zhang,Wenjuan Wu,Jiafeng Lu 2021, 43 (6):  877-884.  doi: 10.19562/j.chinasae.qcgc.2021.06.011 Abstract ( 239 )   HTML ( 13 )   PDF (2408KB) ( 365 )   Save Aiming at the traffic safety problem of pedestrian’s illegal street crossing， leading to frequent human?vehicle collisions， a dynamic programming for vehicle collision avoidance path is performed by adopting the improved artificial potential field method with consideration of the uncertainty of pedestrian behavior. In order to express the direction uncertainty of pedestrians crossing the street illegally， a probability model for the pedestrian’s direction of street crossing based on weighted utility function method is proposed， based on which the location of pedestrian is predicted. Aiming at the steering collision?avoidance trajectory planning of dynamic obstacles and based on the safety distance for avoiding collisions， a kind of variable long axis elliptic obstacle potential energy field with adaptive relative position is proposed， which can plan the collision avoidance path in real time according to the predicted location of dynamic pedestrians. The results of simulation on four working conditions show that the path planning method proposed can effectively plan the safe and smoother obstacle avoidance path according to the predicted location of pedestrians. Figures and Tables | References | Related Articles | Metrics

Select

Theoretical Study and Test Validation on Vehicle Rearview Mirror Whistling Based on γ Corrected Transition Model Haibo Wu,Peng Xing,Jiangbin Zhou 2021, 43 (6):  885-890.  doi: 10.19562/j.chinasae.qcgc.2021.06.012 Abstract ( 285 )   HTML ( 6 )   PDF (1948KB) ( 282 )   Save Aiming at the whistling of rearview mirror， the turbulence intermittence γ and momentum thickness Reynolds number are introduced to explain the changes in the laminar boundary layers on the surface of rearview mirror in detailed，based on the κ?ω SST transfer function. The transition model being corrected by γ is applied to a specific vehicle and verified by wind tunnel test and acoustic camera test. The results show that the corrected transition model can accurately predict the flow phenomenon such as the transition， separation and reattachment of laminar boundary layer though the nephograms of shearing force （or surface friction factor） and momentum thickness， and by combining with the kinetic energy κ， the causes of turbulence intensification can be clarified， and hence the tendency of rearview mirror whistling can be accurately predicted. This method provides certain guidance for estimating the tendency of rearview mirror whistling in the early phase of vehicle development， effectively shortening the lead time and reducing the cost of subsequent real vehicle whistling test. Figures and Tables | References | Related Articles | Metrics

Select

Adaptive Shifting Control for Data Driven Dual Clutch Transmission Yonggang Liu,Jingchen Zhang,Yougang Wan,Dongye Sun,Datong Qin 2021, 43 (6):  891-898.  doi: 10.19562/j.chinasae.qcgc.2021.06.013 Abstract ( 334 )   HTML ( 11 )   PDF (3333KB) ( 214 )   Save A data driven model?free adaptive control method is proposed to solve the problem that the shifting quality of dual clutch transmissions （DCT） deteriorates when internal and external interference exist or system state changes with the service time. First of all， the real vehicle tests are carried out to collect the data of DCT shifting process， and the clutch speed reference curve with relatively good shifting quality is selected by statistics after denoising. Then， a dynamic data model of DCT system is established by designing pseudo?partial derivative estimation method， and a data driven controller for shifting process is designed to follow the clutch reference speed based on model?free adaptive control method. Finally， the simulation model of DCT shifting process is established to verify the proposed controller under different throttle opening. The results show that the proposed method can ensure good shifting quality when internal and external interference exist or systematic state changes. Therefore， the proposed controller has great adaptive ability and robustness. Figures and Tables | References | Related Articles | Metrics

Select

Research on Active Returnability Control for Electro⁃hydraulic Coupling Steering System of Commercial Vehicles Guoqing Geng,Hao Li,Haobin Jiang,Jie Chen,Bin Tang 2021, 43 (6):  899-908.  doi: 10.19562/j.chinasae.qcgc.2021.06.014 Abstract ( 295 )   HTML ( 9 )   PDF (4446KB) ( 201 )   Save Aiming at the problem of under?return at low speed and over?return at high speed of commercial vehicle with traditional hydraulically assisted steering system， a nonlinear sliding mode controller with associated control of vehicle’s mass?center sideslip angle and yaw rate is designed based on electro?hydraulic coupling steering system， with concurrent consideration of the effects of load transfer， road condition and tire’s nonlinearity on returnability. In view of the difficulties in obtaining some state variables of control algorithm and directly measuring road adhesive coefficient， which affects the return stability of steering wheel， the unscented Kalman filter observer is utilized as sliding mode controller to dynamically estimate vehicle state information and road adhesive coefficient， and the differences between the observed value and the desired value of mass?center sideslip angle and yaw rate are taken as the input of the control system to find out the required angle correction for returnability control. Finally， simulations with Trucksim and Matlab/Simulink and steering bench tests are conducted on the retrunability performance of steering wheel under different conditions， and the results show that the active returnability control strategy proposed can effectively enhance the stability and control accuracy of steering wheel returnability. Figures and Tables | References | Related Articles | Metrics

Select

Simulation and Experimental Study on Vehicle Interior Sound Field During Door Closing Zhe Liu,Yunkai Gao,Xiang Xu,Jiaju Chen,Chao Ma 2021, 43 (6):  909-916.  doi: 10.19562/j.chinasae.qcgc.2021.06.015 Abstract ( 336 )   HTML ( 3 )   PDF (4728KB) ( 251 )   Save This paper aims to predict in advance the interior noise of vehicle at the moment of door closing before its completion of manufacturing. Firstly， the transient impact load on the vehicle door during door closing is discretized， the field and bench tests are conducted， and the discretized loads are calculated by transfer path analysis （TPA） method. Then， the finite element model of vehicle door is established， and the discretized loads obtained are used as the input of the finite element model of vehicle door. Finally， the time domain boundary element model of the interior radiated noise is established， and the transient finite element vibration response of vehicle door is mapped to the boundary element in time domain to predict the interior noise of the vehicle. The results show that the simulation data well agree with the test results， verifying the feasibility of predicting the noise inside the vehicle during door closing by discretizing transient load of door closing and acoustic?structure coupling analysis. Figures and Tables | References | Related Articles | Metrics

Select

A New Method for Solving the Statistical Characteristics of Vehicle Non⁃stationary Random Vibration Jie Li,Tao Chen,Wencui Guo,Qi Zhao 2021, 43 (6):  917-923.  doi: 10.19562/j.chinasae.qcgc.2021.06.016 Abstract ( 167 )   HTML ( 5 )   PDF (1479KB) ( 201 )   Save According to the characteristics that road roughness is a stationary random process in the spatial domain when a vehicle is running at variable speed， a new method is proposed to solve the statistical characteristics of the non?stationary random vibration response of the vehicle. A two?degree?of?freedom single wheel model of vehicle vibration in time domain is established. Based on the transformation of variables and their derivatives in time domain and space domain， the corresponding vehicle vibration model in space domain is deduced. With the Fourier transform of spatial frequency， the spatial frequency response characteristics of vehicle vibration and the spatial frequency response characteristics of vibration response variables in time domain are established. The road grading standard is used to represent the statistical characteristics of road roughness in space domain， and the formula for calculating the statistical characteristics of non?stationary vibration response variables is established based on the spatial frequency response characteristics. The parameters of a vehicle and typical driving conditions including uniform acceleration， uniform speed and uniform deceleration are selected to analyze the variation law of statistical characteristics of non?stationary vibration response variables. The results show that the statistical characteristics of vehicle non?stationary vibration response change with the change of vehicle speed and have time?varying characteristics. Figures and Tables | References | Related Articles | Metrics

Select

Study on In⁃plane Dynamic Performances of Re⁃entrant Hexagonal Cellular Structure with Incomplete Factors Ying Zhao,Jie Ma,Ye Sang,Kaifeng Wang,Fangwu Ma 2021, 43 (6):  924-933.  doi: 10.19562/j.chinasae.qcgc.2021.06.017 Abstract ( 159 )   HTML ( 6 )   PDF (6360KB) ( 145 )   Save In order to better clarify the effect of incomplete factors on dynamic performances of cellular material， a re?entrant hexagonal cellular structure with negative Poisson’s ratio is taken as research object. A joint simulation analysis approach with the adoption of Hyperworks and Ls?dyna software is employed to analyze the effect of incomplete factors and impact velocities on its in?plane dynamic performances with the lack of cell wall， and the corresponding deformation modes， stress?strain curves， platform stress curves and energy absorption curves are acquired. The results show that at the same impact velocity， with the increase of incomplete factors， the stable and orderly deformation mode tends to be disordered， and the platform stress and energy absorption capacity will decrease simultaneously. Under low velocity impact， the influence of incomplete factor on its dynamic performances is small， while under high velocity impact， the influence of incomplete factors on its dynamic performances is large， so the existence of incomplete factors should be avoided. The research results in this paper provide data support for the application of cellular structure material in vehicle energy absorbers. Figures and Tables | References | Related Articles | Metrics

Select

Modeling and Experimental Study on Hysteresis Characteristic of Composite Leaf Springs Helong Liu,Wenku Shi,Rui Gao,Zhiyong Chen,Huang Chen,Yunlong Sun 2021, 43 (6):  934-942.  doi: 10.19562/j.chinasae.qcgc.2021.06.018 Abstract ( 181 )   HTML ( 6 )   PDF (3285KB) ( 306 )   Save Bouc?Wen tribology theory is applied to the setting up of the hysteresis characteristic model of composite leaf springs， and according to the dynamic force?displacement curves obtained by bench test， the model parameters are identified by using the modified crow search algorithm （MCSA）. The simulation results well agree with the test ones， showing that the model can accurately predict the hysteresis characteristic of composite leaf spring. A seven DOF vehicle dynamic model is established with consideration of the hysteresis characteristic of composite leaf spring， and the influence of the hysteresis characteristic of composite leaf springs on the dynamic response of vehicle under random road excitation is analyzed. The results indicate that the hysteresis characteristic of composite leaf spring increases the mass?center acceleration of sprung mass and the restoring force of leaf spring， and reduces the dynamic deflection of suspension. This study provides a reference for the vehicle matching and the refined dynamic modeling of composite leaf springs. Figures and Tables | References | Related Articles | Metrics

Select

A Method for Fault Diagnosis of Fuel Injector of Diesel Engine Based on Res⁃CNN and Fuel Pressure Wave Ying Jin,Xinyong Qiao,Cheng Gu,Hao Guo,Chuming Ning 2021, 43 (6):  943-951.  doi: 10.19562/j.chinasae.qcgc.2021.06.019 Abstract ( 178 )   HTML ( 5 )   PDF (2243KB) ( 289 )   Save The working quality of the fuel injection system influences the working process and performance of diesel engine directly. Due to the difficulty in recognizing the characteristic points in fuel pressure wave automatically by using fuel pressure wave for fault diagnosis，the on?line fault diagnosis is influenced. This paper proposes a fuel injector fault diagnosis method by deep learning image recognition theory. Experiments are conducted on a fuel injection pump test bench to simulate typical faults，and the fuel pressure wave of high?pressure fuel pipe is measured . The fuel pressure wave characteristics and laws under different fault conditions are analyzed. A deep residual CNN network （Res?CNN） model is built to detect and verify the faults，with the one?dimension fuel pressure wave signal as the input，and the learning process of fault characteristics is analyzed visually. The results show that the model has higher diagnostic accuracy than the traditional method， which verifies the feasibility of direct application of fuel pressure wave image recognition method for on?line real?time monitoring. Figures and Tables | References | Related Articles | Metrics