American Control Conerence Fairmont Queen Elizabeth, Montréal, Canada June 7-June 9, Air-to-Fuel Ratio Regulation during SI to HCCI Combustion Mode ransition Using the LQ racking Control Xiaojian Yang, Guoming G. Zhu, and Xueei Chen Abstract he combustion mode transition between spark ignition (SI) and homogeneous charge compression ignition (HCCI) combustions o an internal combustion (IC) engine is challenging due to the distinct engine operating parameters over the two combustion modes and the cycle-to-cycle residue gas dynamics during the mode transition. he control problem becomes even more complicated or a multi-cylinder engine without camless valve actuation. his paper studies the combustion mode transition problem o a multi-cylinder IC engine with dual-stage valve lit and electrical variable valve timing (VV) systems. Hardware-in-the-loop (HIL) simulations were used to develop and validate the proposed control strategies. he HIL simulation results show that smooth combustion mode transition can be realized utilizing the hybrid combustion mode in a ew engine cycles and in-cylinder air-to-uel ratio during the mode transition needs to be regulated to the desired level. his paper presents a model based linear quadratic tracking strategy to track the desired air-to-uel ratio by controlling the engine throttle. he HIL simulations demonstrated the eectiveness o the developed control strategies. As a result, it is easible to have a smooth combustion mode transition with dual-stage valve lit and electrical VV systems. H I. INRODUCION OMOGENEOUS charge compression ignition (HCCI) combustion has the potential or internal combustion (IC) engines to meet the increasingly stringent emissions regulations with improved uel economy []. he lameless nature o the HCCI combustion and its high dilution operation capability lead to low combustion temperature. As a result, the ormation o NOx (nitrogen oxides) can be signiicantly reduced []. Furthermore, HCCI engine is capable o un-throttled operation that greatly reduces pumping loss and improves uel economy ([3], []). On the other hand, the HCCI combustion has its own limitations. It is limited at high engine load due to the mechanical limit; and at low load due to misire caused by the lack o suicient thermal energy to initiate the auto-ignition o the gas-uel mixture during the compression stroke [5]. In act, HCCI combustion can be regarded as a type o engine Manuscript received September 3,. his work is partly supported by the National Science Foundation under Grant CMMI-336 and partly supported by the U.S. Department o Energy under Grant DE-EE. Yang, Zhu, and Chen are with the Michigan State University, East Lansing, MI 88, USA (yangxia@egr.msu.edu, zhug@egr.msu.edu, chenxue@msu.edu, respectively). operating mode rather than a type o engine [6]. hat is, the engine has to be operated in dual-combustion mode to cover entire operational range. It is airly challenging to operate the engine in two distinct combustion modes, and it is even more diicult to have the smooth combustion mode transition between SI and HCCI combustions because the avorable thermo conditions or one combustion mode are oten adverse to the other [7]. For example, high intake charge temperature is required in HCCI mode to initiate the combustion, while in SI mode it leads to reduced volumetric eiciency and increased knock tendency. For this reason, engine control parameters, such as intake and exhaust valve timings and lits, throttle position and EGR (exhaust gas recirculation) valve opening, are controlled dierently between these two combustion modes. During the combustion mode transition, these engine parameters need to be adjusted rapidly. However, the physical actuator limitations on response time prevent them rom completing their transitions within the required duration, speciically, within one engine cycle. he multi-cylinder operation makes it challenging [8]. And this problem becomes more diicult when two-stage lit valve and electrical variable valve timing (VV) systems are adopted. Accordingly the combustion perormance during the transition cannot be maintained unless proper control strategy is applied. he control o HCCI combustion has been widely studied in past decades. Robust HCCI combustions can be achieved through model based control as described in [9]-[]. o make the HCCI combustion easible in a practical SI engine, the challenge o the combustion mode transition is inevitable. In recent years, more and more attention has been paid to the mode transition control between SI and HCCI combustion. In [] and [3], smooth mode transitions between SI and HCCI combustions are realized or a single cylinder engine equipped with the camless VVA (variable valve actuation) system. However, high cost prevents the implementation o the camless VVA system in production engines. In [] a VV system with dual-stage lits is used on a multi-cylinder engine or the study o the mode transition. Experimental results show the potential o achieving smooth mode transition by controlling the step throttle opening timing and the DI (direct injection) uel quantity. However, satisactory mode transition has not been accomplished due to the lack o the robust mode transition control strategy. he hardware-in-the-loop (HIL) simulation results 978--577-96-//$6. AACC 5
demonstrated that unstable combustions during the transition can be eliminated by using the multi-step strategy as discussed in [5] or a our-cylinder engine equipped with external cooled EGR, dual-stage valve lit and electrical VV system. his paper utilizes the LQ (linear quadratic) optimal MAP tracking control strategy to maintain the air-to-uel ratio in the desired range so that the hybrid (or spark assisted) combustion is easible. Under the optimal MAP control, smooth combustion mode transition can be achieved with the help o the iterative learning control (ILC) o the DI uel quantity o individual cylinder. Note that the ILC is mainly used to generate transient uel calibrations. he entire control strategy was validated in the HIL engine simulation environment [5], and satisactory engine perormance was achieved during the combustion mode transition or both steady state and transient operating conditions. he paper is organized as ollows. Section II introduces the multi-step SI to HCCI combustion mode transition control strategy. Section III discusses the air-to-uel ratio control problem. he engine throttle and maniold dynamics modeling is presented in Section IV and the air-to-uel ratio tracking control strategies in Section V. he control strategy is demonstrated in Section VI through simulations. Conclusions are inally drawn in Section VII. II. MULI-SEP COMBUSION MODE RANSIION he coniguration o the target HCCI capable SI engine and the engine speciications are listed in ABLE. he key eature o this engine is its valvetrain system. It has two-stage lit or both intake and exhaust valves. he high lit has 9 mm maximum lit or the SI combustion mode; and the low lit has 5 mm maximum lit or the HCCI combustion mode. he ranges o both intake and exhaust valve timing are extended to ± crank degrees to improve the controllability o the internal EGR raction, the eective compression ratio, and the engine volumetric eiciency during the combustion mode transition and HCCI operations. he externally cooled EGR is used to enable high dilution charge with a low charge mixture temperature. ABLE HCCI capable SI engine speciications Engine parameter Model value bore/stroke/con-rod length 86mm/86mm/3.6mm compression ratio 9.8: Intake / exhaust valve lits o high stage 9mm/9mm Intake / exhaust valve lits o low stage 5mm/5mm Intake / exhaust valve timing range ±deg/±deg Intake / exhaust valve lits lash.mm/.5mm Intake maniold volume 3. liter hrottle diameter mm For this paper, the combustion mode transition was studied or the engine operated at rpm with.5 bar IMEP. ABLE lists the engine parameters associated with the SI and HCCI combustion. hese parameters were optimized or steady state engine operation with the best uel economy that satisies the engine knock limit requirement. It can be seen in ABLE that the optimized engine control parameters are quite dierent between the SI and HCCI combustion modes. Some o these parameters can be adjusted within one engine cycle, such as spark timing θ S, electronic throttle control (EC) drive current I EC that is proportional to throttle motor torque, DI uel quantity F DI and valve lit Π lit ; the others cannot due to actuator dynamics. ABLE Engine control parameters or SI and HCCI modes Engine control parameter SI HCCI θ S (deg ACDC) -36 none φ EGR (%) 3 6 I EC (A) 5 F DI (ms/cycle).6.6 θ INM (deg AGDC) 7 95 θ EXM (deg BGDC) 3 Π lit (mm) 9 5 In-cylinder pressure (bar) 35 3 5 5 5 HCCI mode SI mode Crank angle (deg ACDC) Mass raction burned.6.. -5 5 Crank angle (deg ACDC) Fig.. Steady state combustion characteristics o SI and HCCI modes he combustion characteristics are also quite dierent between these two combustion modes as illustrated in Fig.. For example, the HCCI combustion has higher peak in-cylinder pressure compared with that o the SI combustion due to the aster burn rate. Most likely, it also has a recompression phase (see the second peak o the solid line in Fig. ) due to negative valve overlap (NVO) operation, while the SI combustion does not. he goal o the combustion mode transition is to switch the combustion mode without detectable engine torque luctuation by regulating the engine control parameters, or in other words to maintain the engine IMEP during the combustion mode transition. he earlier work in [6] demonstrated that the engine charge temperature ( IVC ) has a response delay during the combustion mode transition, mainly caused by the response delays o the engine intake/exhaust valve timings and maniold illing dynamics. As a result, i the engine were orced to switch to the HCCI combustion mode, the engine IMEP could not be maintained with cycle-by-cycle uel control F DI. Also the increased cooling eect caused by the increment o F DI reduces the charge temperature and leads to degraded HCCI combustions. However, the transitional thermodynamic conditions are suitable or the SI-HCCI hybrid combustion mode proposed in [5] and [5]. By maintaining the engine spark (SI spark location), combustions during the mode transition could start in SI combustion mode with a relatively low heat release rate; and 55
once the thermo and chemical conditions o the unburned gas satisy the start o HCCI (SOHCCI) combustion criteria, the combustion continues in HCCI combustion mode, which is illustrated by the solid curve o mass raction burned (M), shown in Fig., obtained through G-Power simulations. During an ideal SI to HCCI combustion transition process, the HCCI combustion percentage (the vertical distance rom SOHCCI to M = ) increases gradually along with the incremental increase o charge temperature ( IVC ). For the HCCI to SI combustion transition, the process is HCCI combustion percentage will be gradually reduced. More importantly, during the SI-HCCI hybrid combustion, engine IMEP can be controlled by regulating the DI uel quantity that will be discussed in the next section. his is the other motivation o utilizing the hybrid combustion mode during the combustion mode transition. used or engine throttle control. hey provide enough time or the engine MAP to increase to compensate the valve lit (Π lit ) switch. At the end o cycle, the intake/exhaust valve lit Π lit switches rom high lit to low lit, and the control reerences o EGR raction φ EGR, intake valve timing θ INM and exhaust valve timing θ EXM are set to those o the steady state HCCI combustion mode as listed in ABLE. Spark timing θ S o each cylinder was kept constant during the transitional cycles and was eliminated at the end o cycle 5. hroughout the transitional cycles, the engine control parameters, throttle current I EC and DI ueling F DI, are regulated at millisecond sampling period and cycle-based controls, respectively. he corresponding control algorithm will be described in the next two sections. Mass Fraction Burned.6.. SI-HCCI Hybrid mode HCCI mode SI mode S SOHCCI -8-6 - - 6 8 Crank angle (deg) Fig.. M trace o SI-HCCI hybrid combustion mode In [5], a crank based SI-HCCI hybrid combustion model was developed or real-time control strategy development. It models the SI combustion phase under two-zone assumptions and the HCCI combustion phase under one-zone assumptions. he SI and HCCI combustion modes are actually special cases o the SI-HCCI hybrid combustion mode in the model, since the SI combustion occurs when the HCCI combustion does not occur, and the HCCI combustion occurs when the percentage is one hundred percent. Accordingly this combustion model is applicable or all combustion modes during the mode transition. In [6], the one-step combustion mode transition was investigated. he control reerences o all engine parameters were directly switched rom the SI mode to the HCCI mode, as listed in ABLE, in one engine cycle. he simulation results showed that misires occur during the one-step mode transition, and signiicant torque luctuation was discovered. hereby, a multistep mode transition strategy was proposed in [6] by inserting a ew hybrid combustion cycles between the SI and HCCI combustions, see Fig. 3. he proposed control strategy is based on this multistep strategy. As illustrated in Fig. 3, ive engine cycles are used during the SI to HCCI mode transition. During the transitional cycles some engine parameters are adjusted in open loop according to the schedule shown in Fig. 3. Cycles and are Fig. 3. Multistep SI to HCCI combustion mode transition control schedule III. HE AIR-O-FUEL RAIO RACKING PROBLEM o study the easibility o using uel injection quantity F DI to regulating the engine IMEP, intensive simulations were conducted to map out the engine IMEP and air-to-uel ratio as unctions o engine uel injection quantity F DI and MAP (maniold air pressure). he simulation results are shown in Fig., indicating that the engine IMEP is highly correlated to F DI with the lean air-to-uel mixture. As a result, it is possible to control the individual cylinder IMEP by regulating the corresponding F DI. F DI (ms)..3...5.9.8.7.6.5.5....5.5.5.5.6.5 3.6 3 3 IMEP 3.5 3.5 3.5.6.5 5.5.5.55.6.65.7 MAP (bar) Fig.. IMEP sensitivity analysis o the SI-HCCI hybrid combustion mode o maintain the controllability o the DI ueling (F DI ), lean gas-uel mixture is required during the mode transition. However, the combustion could become unstable i the mixture becomes extremely lean since the engine spark might λ... 56
not be able to ignite the gas mixture. For this study the desired normalize air-to-uel ratio is set between λ min (.97) and λ max (.3). In [6], a step throttle pre-opening approach was proposed to prevent rich combustions at cycle 3, but it leads to very lean combustions at the ollowing engine cycles. In this paper, an LQ tracking control strategy is developed to regulate the air-to-uel ratio around the desired level. As discussed above, the normalized air-to-uel ratio needs to be maintained within the optimal range (λ min λ λ max ) during the SI to HCCI combustion mode transition. his control target is diicult to achieve through the air-to-uel ratio eedback control due to delay and a short mode transition period. It is proposed to use the LQ optimal tracking approach to regulate the air-to-uel mixture to the desired level. o implement this control strategy, the optimal operational range o λ is translated into the operational range o the engine MAP shown in Fig. 5, where the upper limit is corresponding to λ max and lower limit is corresponding to λ min. his provided an engine MAP tracking reerence, shown in Fig. 5, to maintain the engine MAP to stay within the desired range. he reerence signal is represented by ZSI i kb < k k k k z( k) = ZSI + ( Z ZSI ) i k < k k k k () k k Z + ( Z HCCI Z) i k k k k k < E where k is the time based sampling index; k B and k E represent the beginning and ending indices o the mode transition and they were set to 6 and 9, respectively, as shown in Fig. 5; k and k are switch indices and they equal 67 and 7, respectively; Z SI and Z HCCI are the desired MAP o SI and HCCI modes, respectively; Z is the desired MAP at k. MAP (bar)..9.7.6.5 SI SI to HCCI mode transition HCCI upper limit, λ max =.3 lower limit, λ min =.97 MAP reerence z(k) 6 67 7 78 8 9 96 8 k B k k k E ime (ms) Fig. 5. he target MAP operational range and MAP tracking reerence IV. ENGINE AIR CHARGE DYNAMIC MODEL o develop the proposed LQ tracking control strategy, a simpliied engine MAP model is required to represent the relationship between the control input (I EC ) and the system output (MAP). he simpliied dynamics are represented by second order dynamics due to the gas illing dynamics (irst order) o the engine intake maniold and the irst order response delay o the engine throttle. he governing equation o gas illing dynamics is represented by With illing dynamics time constant around 6ms. he dynamics o the throttle response is approximated by dφps kec cec = φps + I (3) EC dt bec bec where η, V, d V and m N are volumetric eiciency o intake e process, engine displacement, intake maniold volume, and engine speed, respectively; R, amb, P amb and C are gas constant, D ambient temperature, ambient pressure, and valve discharge constant, respectively; and φ PS, k EC, bec and EC c are engine throttle position, spring stiness o the throttle plate, damping coeicient o the throttle plate, and throttle motor torque constant, respectively. he throttle time constant is around 5ms. Equations () and (3) can be combined, discretized and represented by the ollowing discrete state space model where dmap Vd Ne Ramb CDπ r Pamb = η MAP + ϕ φ () PS dt V V R m m amb x( k + ) = Ax( k) + Bu( k) y( k) = Cx( k) + Du( k) x MAP u = IEC ; x = ; y MAP x = φ = (5) PS are the system input, state and output respectively. he system matrices are η( k) Vd Ne ϕ( k) Ra acdπ r P a t t V m Vm Ra a A=, B = k EC k EC t (6) t bec bec C = [ ], D = where t is the sample period. State space model () is linear time-variant since the volumetric eiciency η and multiplier φ in equations () and (6) are unctions o the engine operating condition. Moreover, the sampling time in (6) equals millisecond, and sample time index k is the same as that in equation (). V. LQ RACKING CONROL SYNHESIS Based on the control oriented engine MAP model, a inite horizon LQ optimal tracking controller was designed to ollow the reerence z(k). More speciically, the control objective is to minimize the tracking error e(k) deined in (7) with the easible control eort I EC. he tracking error e(k) is deined as and the constraint on I EC is -5A < I EC < 5A. he cost unction o the LQ optimal controller is deined as () e( k) = y( k) z( k) = Cx( k) z( k) (7) 57
[ ( ) ( )] J = Cx k z k F [ Cx ( k ) z ( k )] (8) k= k + {[ Cx( k) z( k)] Q[ Cx( k) z( k)] + u ( k) Ru( k) } k= ki where F and Q are positive semi-deinite and R is positive deinite. For this paper, F and Q are constant matrices deined in (9) and R is a unction o sample index and tuned to optimize the tracking error with easible throttle control eort, see Fig. 6. R.5.5 F Q R R k 6 66 7 78 8 9 Sampling index k, time (ms) Fig. 6. Adjustment o weighting matrix R Based on the cost unction the corresponding Hamiltonian is as ollows H = [ Cx( k) z( k)] Q[ Cx( k) z( k)] + u ( k) Ru( k) () + p ( k + ) [ Ax( k) + Bu( k) ] According to [7], the necessary conditions or the extremum in terms o the Hamiltonian are represented as H * * * * = x ( k + ) x ( k + ) = Ax ( k) + Bu ( k) () * p ( k + ) H * * * * = p ( k) p ( k) = A p ( k + ) + C QCx ( k) C Qz( k) () * x ( k) H * * = = B p ( k + ) + Ru ( k) (3) * u ( k) Note that the superscript * denotes the optimal trajectories o the corresponding vectors. he augmented system o () and () becomes * * x ( k + ) A BR B x ( k) ( ) * = * + z k p ( k) C QC A p ( k ) C Q + Based on (3) the optimal control is in the orm o * * u ( k) = R B P( k) x ( k) g( k) (5) Matrix P(k) can be computed by solving the dierence Riccati equation backwards with the terminal condition and vector g(k) can be computed by solving the vector dierence equation with the terminal condition 8 7 =, =, = ( ) (9) ( ) P k A P( k )[ I EP( k )] A C QC () = + + + + (6) P( k ) { } = C FC (7) g( k) = A I [ P ( k + ) + E] E g( k + ) + C Qz( k) (8) g( k ) = C Fz( k ) (9) he optimal control in (5) can be written into the ollowing orm u ( k) = L ( k) x ( k) + L ( k) g( k + ) () * * where the eedorward gain L FF is computed by and the eedback gain L is computed by Note that in equation () the state x * used in the eedback control is computed exactly rom the closed loop system model deined below However, when the control is implemented into the HIL simulation environment or the actual engine control system, the eedback states are replaced by the actual signals (MAP and φ PS ) measured by the on-board engine sensors. In these cases the LQ controller is represented by the online orm as where x represents the sampled states. Note that both o the states, MAP and φ PS can be measured in the HIL simulator or in the engine system. VI. SIMULAION RESULS AND DISCUSSION he developed LQ optimal MAP tracking control was implemented into the prototype engine controller and validated through the HIL engine simulations. he simulated control input I EC, the system states MAP and φ PS, and λ are plotted in Fig. 7. For comparison purpose, the simulated responses o these variables with a step I EC control approach are also shown in Fig. 7, in which I EC is set to the target level beore the adjustment o Π lit (happens at 7 th ms), as a result, the engine throttle is gradually opened to the wide open throttle (WO) position and the MAP is increased beore the valve lit switch. he increased MAP ensures enough resh charge to each cylinder when the valve lit switches to the low lit. However the step I EC control leads to a rapid increment o the engine MAP or excessive resh air charge, leading to extreme lean air-to-uel ratio λ in the ollowing engine cycles. Using the proposed LQ MAP tracking control strategy, throttle current I EC is regulated in a non-monotonic increasing pattern. Note that to maintain I EC in the easible range (-5A < I EC < 5A) the weighting matrix R in the cost unction (8) is adjusted as illustrated in Fig. 6. he similar pattern can also be ound or φ PS with a small phase lag. As a result, the engine MAP tracks the reerence z(k) ater the intake valve lit Π lit switches to the low lit, and λ o each cylinder is successully maintained within the desired range. hereore, with the help o the LQ optimal tracking control, the in-cylinder air-to-uel ratio is maintained within the FF LFF ( k) [ R B P( k ) B] B = + + () L k = R+ B P k + B B P k + A () ( ) [ ( ) ] ( ) x ( k + ) = [ A BL ( k)] x ( k) + BL ( k) g( k + ) (3) * * u( k) = L ( k) x( k) + L ( k) g( k + ) () FF FF 58
desired range, leading to stable combustions. I EC (A) φ PS (%) MAP (bar) λ 6-5.6 6 75 9 5 35 5 65 w/o LQ w/ LQ 6 75 9 5 35 5 65 λ max λ min z(k) 6 75 9 5 35 5cylinder 65#. cylinder # cylinder #3. cylinder # 6 75 9 5 35 5 65 ime (ms) Fig. 7. Engine perormances o the optimal MAP tracking control Slight oscillations in the MAP responses are ound with both control approaches, which are due to the low dynamics o the engine air-handling system and the engine MAP modeling error. It is almost impossible to eliminate them. Moreover, the MAP oscillation associated with the LQ optimal tracking control is within the desired MAP range. VII. CONCLUSIONS he combustion mode transition between the spark ignition (SI) and homogeneous charge compression ignition (HCCI) combustions is challenging but necessary to implement the promising HCCI combustion technology to production SI engines. o ensure the smooth combustion mode transition the engine air-to-uel ratio needs to be precisely controlled. his paper shows through the hardware-in-the-loop simulations that the LQ control is capable o tracking the engine MAP to the desired target with small oscillations. As a result, the engine air-to-uel ratio is maintained within the desired range. his makes it easible to regulate the individual cylinder IMEP through adjusting the corresponding direct injection uel quantity. [3] N. J. Killingsworth, S. M. Aceves, et al, HCCI Engine Combustion-iming Control: Optimizing Gains and Fuel Consumption Via Extremum Seeking, IEEE ransactions on Control Systems echnology, Vol. 7, No. 6, November 9, pp. 35-36. [] C. J. Chiang and A. G. Steanopoulou, Stability Analysis in Homogeneous Charge Compression Ignition (HCCI) Engines With High Dilution, IEEE ransactions on Control System echnology, Vol. 5, No., March 7. [5] X. Yang and G. Zhu, A wo-zone Control Oriented SI-HCCI Hybrid Combustion Model or the HIL Engine Simulation, Proceedings o American Control Conerence, San Francisco, Caliornia, USA,. [6] S. C. Kong and R. D. Reitz, Application o Detailed Chemistry and CFD or Predicting Direct Injection HCCI Engine Combustion and Emission, Proceedings o the Combustion Institute, Vol. 9,, pp. 663 669. [7] X. Yang, G. Zhu, and Z. Sun, A Control Oriented SI and HCCI Hybrid Combustion Model or Internal Combustion Engines, Proceedings o ASME Dynamic Systems and Control Conerence, Cambridge, MA,. [8] N Kalian, H Zhao, and J Qiao, Investigation o transition between spark ignition and controlled auto-ignition combustion in a V6 direct-injection engine with cam proile switching, Proc. IMechE Vol. Part D: J. Automobile Engineering, IMechE 8. [9] N. Ravi, M. J. Roelle, et al, Model-Based Control o HCCI Engines Using Exhaust Recompression, IEEE ransactions on Control Systems echnology, Vol. 8, No. 6, November, pp. 89-3. [] J. Kang, C. Chang, and. Kuo, Suicient Condition on Valve iming or Robust Load ransients in HCCI Engines, SAE International, --3. [] G. M. Shaver, Physics based modeling and control o residual-aected HCCI engines using Variable Valve Actuation, PhD thesis, Stanord University, September, 5. [] M. J. Roelle, G. M. Shaver, and J. C. Gerdes, ackling the ransition: A Multi-Mode Combustion Model o SI and HCCI or Mode ransition Control, Proceedings o IMECE International Mechanical Engineering Conerence and Exposition, Anaheim, Caliornia, USA, November 3-9,. [3] Y. Zhang, H. Xie, et al, Study o SI-HCCI-SI ransition on a Port Fuel Injection Engine Equipped with VVAS, SAE Paper 7--99. [] H. Wu, M. Krat, et al, Experimental Investigation o a Control Method or SI-HCCI-SI ransition in a Multi-Cylinder Gasoline Engine, SAE International, --5. [5] X. Yang and G. Zhu, A Mixed Mean-Value and Crank-based Model o a Dual-Stage urbocharged SI Engine or Hardware-In-the-Loop Simulation, Proceedings o American Control Conerence, Baltimore, MD,. [6] X. Yang and G. Zhu, SI and HCCI Combustion Mode ransition Control o A Multi-Cylinder HCCI Capable SI Engine Via Iterative Learning, Proceedings o the th Annual Dynamic Systems and Control Conerence, Westin Arlington Gateway, Arlington, VA, USA. Oct 3 - Nov,. [7] D. Naidu, Optimal Control Systems, CRC Press LLC, 3, pp. 3-39. REFERENCES [] F. Zhao,. Asmus, D. Assanis, J. E. Dec, J. A. Eng, and P. M. Najt, Homogeneous Charge Compression Ignition (HCCI) Engines Key Research and Development Issues, 3, Warrendale, Pennsylvania: Society o Automotive Engineers. [] R. M. Wagner, K. D. Edwards, et al, Hybrid SI-HCCI Combustion Modes or Low Emissions in Stationary Power Applications, 3rd Annual Advanced Stationary Reciprocating Engines Meeting Argonne National Laboratory, Argonne, IL June 8-3, 6. 59