Assessment of passenger car driveability with use of two axis accelerometer mounted on car body

Size: px

Start display at page:

Download "Assessment of passenger car driveability with use of two axis accelerometer mounted on car body"

Douglas Joseph
5 years ago
Views:

1 Assessment of ssenger cr drivebility with use of two xis ccelerometer mounted on cr body S. Brol, J. Mml Technicl University of Oole, Chir of Rod nd Agriculturl Vehicles Oole , ul Mikoùjczyk 5, Polnd, Phone: ,0 Abstrct In this er n ssessment of ssenger cr drivbility with use of two xis ccelerometer mounted on cr body is shown. The cr boy ccelertion mesurement ws nlyzed in four res: sensor ccurcy, ccelerometer nd working engine relted noises, sensor coordinte system reorienttion during instntneous ccelertion or brking nd disturbnces cused by the rod irregulrities. Simultions nd exerimentl results roved tht cr ccelertion mesurement is disturbed by the cr susension rection when it ccelerting or brking nd rod irregulrities. Introduction Till tody mny devices hve been constructed with bility to mesure or estimte cr drivbility rmeters like engine ower nd torque. There re mny solutions bsed on mny concets. The most oulr nd reltively che wy is to ccess the OBD system nd retrieve vlues of the ctul rmeters like engine lod, cr seed etc. Bsing on these vlues the ower nd torque re clculted [2, 4]. Flw of this solution is tht the cr must hve OBD system nd the device must be ble to communicte with it. It demnd from the device, if it should be fully ortble, to be equied with four interfces: ISO, PWM, VPW nd CAN which comlictes the hrdwre nd softwre. Another roblem is tht the OBD system is not designed for rel time dt trnsmission. More simle, ortble nd dedicted for ll crs or trucks solution is to utilize only multi xis ccelerometer with microrocessor system. There re mny lictions for so designed system: ssessment of ower trin nd brking system, brking efficiency, engine erformnce etc [6, 7]. If ny ssessment is to be mke with this kind of mesurement system then on-rod test must be erformed. Tyiclly it consist of ccelertion with fully deressed ccelertion edl hse nd free run to sto or full brking. If the device is ccelerometer bsed nd mounted in cr body then during the rod test mny fctors influences on mesured ccelertion. This is rimry rmeter which must be mesured roerly, if credible test results should be chieved [4, 5]. of the sensor mesurement direction to be erendiculr to the movement direction. The ccelertions mesured under the conditions re equl to the rel ccelertions cting on the cr. In rel cr moving on rel rod ccelertion mesured nd rel cting re not equl. There re mny significnt fctors influents on this sitution like rod irregulrities, cr susension construction nd stiffness, wheel bse, wheel trck. The ccurcy of ccelertion mesurement will be considered s the difference between rel nd mesured cr body ccelertion in three res: 1. Sensor ccurcy. 2. Accelerometer nd working engine relted noises. 3. Sensor coordinte system reorienttion during instntneous ccelertion or brking. 4. Disturbnces cused by the rod irregulrities. Sensor ccurcy Mesurements of cr ccelertion ws mde with use of dul xis ccelerometer. The low-cost nlogue nd digitl outut Anlog Devices ADXL202 sensor were used. (Fig. 1). Methods of nlysis Assessing cr drivbility with use of ortble ccelerometer bsed device demnd on-rod test. Mounting it in cr must be done without ny chnges in cockit construction. This ssumtion reduces the ossibilities of mounting oints to windshields, windows nd floor re. Considering cr s body with idel stiff susension moving on flt surfce tngentil oriented to the grvity direction the ccelertion mesurement is simle. Only one think is to do to set the orienttion one Fig. 1. Accelerometer used to mesure cr ccelertion The ADXL202 ccelertion rnge is 17.62m/s 2 nd outut signls re nlog, rnging from 0V to 5V, nd 31

2 digitl PWM. It cn mesure ccelertions or tilt becuse it is erth ccelertion sensible. Accelerometer nd working engine relted noises Accelerometer outut signls re not noise free. In generl in ll low-cost ccelerometers the outut signl contin gussin noise. b Fig. 3. Originl, moving verge with 10 smles nd digitl RC filtered ccelerometer outut signl when ccelertion =0 m/s 2 ws lied The most imortnt roblem is how to smooth the outut signl so, tht it still be dequte to dynmic chnges in ccelertion. Two digitl lowss filters were tested to sses which one better remove the noise: running verge nd RC. Filters rmeters ws chosen so, tht the signl mlitude reduction ws for both the sme nd ws equl to 50% of ek-to-ek vlue. The best erformnce ws chieved with use of RC filter (Fig. 3.). The outut of RC filter is more smooth then the originl signl nd gives more fst resonse then the running verge which genertes hse shift deendent on number of smles used to verge originl signl. Unfortuntely no of exmined filters ws ble to remove the Gussin roerty of outut signl nd reduce the ek-to-ek vlue to zero. Alterntive method to reduce the noise is to lower mesurement resolution to 0.05 m/s 2 witch is lower then ek-to-ek vlue when ccelertion =0 m/s 2 lied. The sensor noises re reltively low in comre to signl chnges when cr engine is running. The noise is engine tye (gsoline or diesel), rottion seed, number of cylinders, engine susension, nd unblnced msses deended [3]. Fig. 2. Accelerometer outut signl histogrms when ccelertion =0 m/s 2 ) nd 9.81 m/s 2 b) ws lied The outut noise is vrying with the mesurement bndwidth. Reduction of bndwidth will imrove the signl to noise rtio nd the resolution. The outut noise is roortionl to the squre root of the mesurement bndwidth. The noise RMS is lso deended on ccelertion cting on the sensor. As shown in Fig. 2, it is the lowest for ccelertion =0 m/s 2 nd become greter when ccelertion increses. For = 0m/s 2 the ek-to ek rtio is equl to 0.04 m/s 2 (Fig 2) nd for =9.81m/s m/s 2 (Fig 2b). The ek-to-ek vlue roximtely defines the worst cse resolution of the mesurement. Mesurement reduction ending to the ek-to-ek vlue in this kind of sensors is troublesome becuse of ek-to-ek vlue chnges in function of ccelertion. Fig. 4. Comrison of ccelerometer outut signl when engine stoed nd idling. The test ws mde on Citroen BX cr with four stroke, four cylinders 1.9 dcm 3 diesel engine. Engine ws idling t 900 RPM 32

3 As shown on Fig. 4. the engine indicted ek-to- ek vlue is bout 6 times greter then sensor when =0 m/s 2. It is cler to see tht if the engine is stbile idling it roduces sttionry sinusoidl-like wve. Generlly the gsoline engines roduces lower mlitudes then the diesels. In this cse, the worst in ll test results, the ccelertion mlitude (0.24 m/s 2 ) is close to ek-to-ek vlue of ccelerometer when only 9.81 m/s 2 ccelertion is lied (0.28 m/s 2 ). When this four stroke, four cylinders engine is idling, t 900 RPM (15 rev/s) the wve frequency is 30Hz. It mens tht this frequency is combustion deended becuse the combustion rocess hens two times er one crnkshft revolution. The ccelertion mesurement smle rte ws set to 100 Hz, so only three smles re mesured for one combustion. When the engine increse their rottionl seed the combustion frequency will lso increse. If the combustion frequency will be greter then Nyquist frequency (50 Hz t 1500 RPM) then it will led to lising henomen which cn deliver dditionl uncorrelted wve to the mesured signl. Sensor coordinte system reorienttion during instntneous ccelertion or brking It ws mde n ssumtion, tht the sensor will be in initil osition oriented so, tht one of its mesurement xis will be rllel to the cr movement direction nd the second to grvity vector. It is imortnt to notice, tht in ll considered sensors the mesurement xis should be oriented so, tht the mesured ccelertion will be in oosite to xis sense. As shown in Fig. 5 the sensor coordinte system is bound with cr body nd the globl with erth ccelertion nd movement direction. Fig. 5. Globl nd sensor ssocited coordinte systems As mentioned before, the device ws rojected so, tht it should be ble to mesure cr erformnce on on-rod test with use of ccelerometer mounted on cr body. The first rt of on-rod test is cr ccelertion hse. During it the cr body chnges the itch ngle reltive to rod surfce. When the cr ccelerte its body nd ssocited with it sensor coordinte system chnges osition nd itch ngle reltive to globl coordinte system. The ccelertion of Fig. 6. Sensor coordinte system reorienttion when cr is ccelerting cr is bounded with globl coordinte system but the mesured ccelertion with the sensor. If itch ngle is non zero between them, then the mesured ccelertion is equl to rojection of ll ccelertions on mesurement xis like shown on figure 6. The sense of erth ccelertion vector is ointing in oosite direction then tyiclly showed (like on Fig. 5). It is so becuse the inner sensor element decelerte while the housing is ccelerting. In fct it rects on decelertion reltive to housing so the erth ccelertion vector sense must be showed like on Fig. 6 if the ccelertions rojections should be roerly described. The mesured ccelertions on ccelerometer xis re s follows: g g cos( ) g sin( ) g cos( ) sin( ) where: rel cr ccelertion, cr ccelertion rojection on x s xis, cr ccelertion rojection on z s xis, cr body itch ngle between sensor nd globl coordinte system, g erth ccelertion, g=9.81 m/s 2, g erth ccelertion rojection on x s xis, g erth ccelertion rojection on z s xis. To ssess how the itch ngle influences on mesured ccelertions on x s nd z s simultion ws erformed. It ws mde n ssumtion, tht the cr ccelertion is rging form =-10 m/s 2 to =10 m/s 2 nd the itch ngle from = 10 to =10 degrees. The results s difference between mesured on x s nd z s xis ( nd ) nd forwrd cr ccelertion ccording to Eq. 1 re shown. As shown on Fig. 7 the difference between mesured on x s xis nd rel forwrd ccelertion for itch ngle = 10 degrees is equl to 1.52 m/s 2 nd for =10 degrees is 1.7 m/s 2. The difference is lso forwrd ccelertion deendent nd increses with itch ngle. The chnges of ccelertion difference vlue in z s mesurement direction re lso itch nd forwrd ccelertion deendent (Fig 8). They re little bit lower then in x s direction. The gretest vlue is 1.27 m/s 2 nd the lowest 1.27 m/s 2. In generlly the mesured ccelertion is greter when the chr is ccelerting nd lower when it brking. It is so becuse reorienttion of sensor bsed (1) 33

4 Fig. 7. Simulted ccelertion difference ( -) between mesured on x s xis nd rel cr body ccelertion s function of cr body ccelertion nd itch ngle Fig. 9. Simulted ccelertion mesured on x s nd z s xis nd rel cr body ccelertion s function of cr body ccelertion nd itch ngle The simultion results were verified on on-rod tests where the Fit Punto ws used. It ws chosen becuse its wheel bse equl to 2510 mm. This reltively short wheel bse could cuse significnt cr body itch chnge while it ccelerte or decelerte. Fig. 10 shows on-rod test results when ccelerometer outut from x s xis re comred to forwrd ccelertion comred with ccelertion clculted from seedometer signl. There is n offset between signls in ccelertion nd decelertion hse. The ccelertion mesured by ccelerometer is greter when cr is ccelerting nd lower when brking in comre to seed signl derivtive. Fig. 8. Simulted ccelertion difference ( -) between mesured on z s xis nd rel cr body ccelertion s function of cr body ccelertion nd itch ngle coordinte occurs nd the rojection of grvity ccelertion decides bout mesurement ccelertion vlue. As shown before the itch ngle chnge cuse significnt ccelertion error secilly mesured on cr longitudinl xis. A simultion ws erformed to check if it is ossibility to determine momentry itch ngle from ccelertions mesured on x s nd z s xis. For this urose grh ws mde where the mesured ccelertions distribution on x s nd z s xis re combined. Simultion results showing (Fig 9), tht for irs of ccelertions mesured in x s nd z s xis re more thn one combintion of cr body ccelertions nd itch ngles. For exmle if x s =0.6 m/s 2 nd z s =9.8 m/s 2 (see the shded circles on Fig. 9) the cr body ccelertion could be 0.1 m/s 2 or 0.75 m/s 2 n the itch ngle dequtely 0.1 nd 6 degrees. It leds to conclusion tht the determintion of cr body itch ngle cn not be done with use only mesured by ccelerometer vlues. Fig. 10. Accelerometer outut from x s xis nd ccelertion clculted from seedometer signl. Dt cquired on on-rod test with use of Fit Punto In generl the offset vlue is vrying nd is cr susension stiffness, wheel bse, cr mss nd driving force lied deended. The gretest registered difference on this rod test ws 0.9 m/s 2 when the cr is ccelerting with 1.5 m/s 2. Comred with the difference distribution m on fig. 7 the itch ngle in moment of mximum ccelertion ws 4 degrees. The ccelertion clculted from seedometer signl is smoother then from 34

5 ccelerometer. It is so, becuse the ccelerometer signl is disturbed by dditionl multi xil ccelertions cusing by the rod irregulrities nd engine noise. Disturbnces cused by the rod irregulrities When the cr s wheels re moving on the rod its irregulrities indicte ccelertions which cn be different in its vlue nd direction for every wheel. They cting through susension on cr body nd, cusing lterl nd longitudinl tilt rtio chnges [1]. Rod irregulrities deendent ccelertions nd both tilt rtio chnges cn disturb the ccelertion mesurement becuse they generte n dditionl ccelertions which vlue is rod surfce qulity nd sensor osition in cockit deended. Conclusions The ccelerometer outut noise floor is bndwidth nd ccelertion deended. If ny microrocessor bsed ccelertion system will be rojected the filtering method should be considered. The best erformnce ws chieved with use of RC filter. Its hrdwre reliztion will reduce rocessor lod, but will dd time constnt to filtered signl. The A/D mesurement resolution should be ekto-ek vlue of filtered signl relted. Lowering the mesurement resolution will reduce the noise but lso the system ccurcy. It should be otiml solution received which will get into ccount the ssumed ccelertions rnge, system ccurcy nd resonse to ccelertion vlue chnges. The engine noise is hrd to vnish becuse it is, beside others, engine rottionl seed relted. Reduction of engine noise could be chieved only when RPM signl will be delivered to the system. The cr ccelertion or brking is relted to cr body itch nd itch rtio chnges witch cusing difference between mesured nd forwrd cr ccelertion. It is imossible to correct this itch bsed only on mesured signls. If itch comenstion should be mde the chrcteristic of susension rection on ccelertion must be mde or itch ngle must be mesured. Acknowledgements The uthors wishes to thnk the Polish Stte Committee for Scientific Reserch which suorted this work in yers References 1. Duym S. nd Reybrouck K. Physicl Chrcteriztion of Nonliner Shock Absorber Dynmics. Euroen Journl Mech. Eng. Vol. 43. No. 4. P Jntos J. Control of the Trnsmission Rtio Derivtive in Pssenger Cr Powertrin with CVT, SAE Trnsctions. Journl of Pssenger Crs Mechnicl Systems, SAE P Mml J., Jntos J. Sterownie stoniem ilni silnik odc rzysieszni smochodu. Tek Komisji Nukowo-Problemowej Motoryzcji PAN 2003, Zeszyt P Serrrens A. F. A., Veldus F. E. Dynmic modelling for dimensioning nd control of flywheel ssisted driveline, Vehicle Systems Technology for the Next Century. Snish Society of Automotive Engineers STA, Brcelon P Siùk W. Teori ruchu smochodu. WNT Wrszw Constncis P., Leort F. Globl Powertrin Control. FISITA' No. F98P Wengung G. Wng, Red M. Bt. Trnsient Resonse in Dynmometer Power Absortion System. SAE P Received

6 36

Department of Technology, Yinjian Automobile Repair Co., Ltd., Beijng, China

Department of Technology, Yinjian Automobile Repair Co., Ltd., Beijng, China GLOBAL JOURNAL OF ADVANCED ENGINEERING TECHNOLOGIES AND SCIENCES THE INFLUENCES OF AIR INTAKE TEMPERATURE ON THE TURBOCHARGED GASOLINE ENGINE S OTTO CYCLE Feng Liu* * Dertment of Technology, Yinjin Automobile