Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

controller

VIEWS: 12 PAGES: 11

									                                          Engine benchmark


                                    I. Controller design
The aim of this section is to present on the one hand an exhaustive list of the different rules and
constraints inherent to the GDI engine to respect. On the other hand, this section presents different
criteria (quantitative and qualitative) in order to evaluate and compare several controllers developed
within the framework of FAMIMO.

   Controller inputs and outputs

   Reduced benchmark controller (1st step)
Considering the reduced benchmark (cf. Error! Reference source not found.Figure 3), the EMS
inputs correspond to the desired torque pattern and the desired air/fuel ratio set point. According to
the implemented control strategy, other variables can be used as feedback inputs.
All the variables that can be used as feedback inputs of the Engine Management System are listed
below:
   Manifold pressure (Pmanifold),
   Engine speed (N),
   Air/fuel ratio (RATIOair/fuel),
   Fresh air throttle position (MTC POS),
   Fuel consumption (Instantaneous and specific consumption),
Other variables can be used as inputs of the engine management system however these variables
cannot be directly measurable on the engine and must be reconstructed via an observer.
Nevertheless, in a first approach, we will consider that observers are available and these variables
can be directly extracted from the model.
   Fresh air flow (Qair),
   Fuel flow (Qfuel),
   Effective engine torque (TQE),
   Indicated engine torque (TQI).

The EMS outputs are listed below:
  Fresh air throttle control (MTC),
  Ignition advance control (IGA),
  Fuel injection time control (Tinj),
  Fuel injection start angle control (SOI).




                                   SIEMENS CONFIDENTIAL                                             1
                                                         Engine benchmark

            Reference set point inputs
                       Desired torque (TQEd)
         Desired air/fuel ratio (RATIOair/fueld)
Feedback inputs from the powertrain                                                Fresh Air throttle control
                 Manifold pressure (Pmanifold)                                     (MTC)
                              Engine speed (N)                  Engine             Ignition advance control
                                                              Management           (IGA)
                             Fuel consumption
                           Fresh air flow (Qair)
                                                                System             Fuel injection time control
                                                                                   (Tinj)
                               Fuel flow (Qfuel)
                                                                                   Fuel injection start angle control
                   Air/fuel ratio (RATIOair/fuel)                                  (SOI)
                Effective engine torque (TQE)
                 Indicated engine torque (TQI)
          Fresh air throttle position (MTC POS)

                                            Figure 1 Engine Management System

         Intermediate benchmark controller (2nd step)
      Considering the intermediate benchmark controller (cf. Error! Reference source not found.Figure
      5), the EMS inputs correspond to the accelerator pedal position and the desired air/fuel ratio set
      point.
      In addition to the feedback inputs defined for the reduced benchmark controller (cf. section 0), the
     Torque demand from the driver
Accelerator pedal position (AcceleratorPedal)
            Reference set point input
           Desired air/fuel ratio (RATIOair/fueld)
Feedback inputs from the powertrain
                   Manifold pressure (Pmanifold)
                               Engine speed (N)                  Engine                Fresh Air throttle control
                              Fuel consumption                 Management              (MTC)
                            Fresh air flow (Qair)
                                                                 System
                                                                                       Ignition advance control
                                Fuel flow (Qfuel)                                      (IGA)
                    Air/fuel ratio (RATIOair/fuel)                                     Fuel injection time control
                 Effective engine torque (TQE)                                         (Tinj)
                                                                                       Fuel injection start angle control
                  Indicated engine torque (TQI)
                                                                                       (SOI)
          Fresh air throttle position (MTC POS)
       Feedback inputs from the driver
          Clutch pedal position (ClutchPedal)
            Brake pedal position (BrakePedal)
                                            Gear
      brake and clutch pedals as well as the current gear can be used as EMS inputs.
                                            Figure 2 Engine Management System

           Controller objectives
      Four main control objectives can be enhanced. The first three objectives concern the reduced and
      intermediate benchmarks with some specifities and the fourth objective only concerns the
      intermediate benchmark.

      2                                              SIEMENS CONFIDENTIAL
                                           Engine benchmark

In the case of the reduced engine benchmark, the first objective of the EMS is to control the engine
     in order to obtain the effective torque (TQE) following the torque reference set point (TQEd). In
     the case of the intermediate engine benchmark, this objective is to control the engine in order to
     follow an imprecise driver torque demand expressed through the accelerator pedal (The
     accelerator pedal value is proportional to the percentage of the maximal engine power desired
     by the driver).

The second objective of the EMS is to ensure a regulation of the air/fuel ratio to a constant value
    when the engine operates in the homogeneous mode.
                                       RATIO air / fueld  1  2 %
    Nevertheless, during the transient switching phases the air/fuel ratio in homogeneous mode
    could be different from 1.
    When the engine operates in the stratified mode, not any more regulation of the air/fuel ratio is
    to be taken into account as for homogeneous mode. Nevertheless, the engine must run in lean
    burn conditions. So an air/fuel ratio constraint must be respected.
                                              1  RATIO air / fuel


The third objective of the EMS is to minimise the fuel consumption.
    Due to high potentialities in terms of low pollutant emissions and low fuel consumption of the
    stratified mode, the controller objective will be to maintain as longer as possible the stratified
    conditions. Unfortunately, this combustion mode is possible only for a restricted engine
    operating domain (cf. section 0).
    The control strategy must be able to switch between these two combustion modes according to
    the working engine conditions. The commutation between homogeneous and stratified mode
    must respect some constraints (cf. section 1.1.1.2) to avoid any disturbances to the driver.

From the intermediate benchmark, the fourth objective of the EMS concerns the idle speed phases
    represented by an accelerator pedal position equal to zero. The EMS must manage the inputs
    into the idle speed phases and must ensure a good regulation of the engine speed (N) to a
    constant value (Nidle=1000 rpm) during the idle speed phases.

   Constraint definition

   Engine operating constraints
In order to design a realistic controller, the control strategy must respect physical constraints for
both the engine and its actuators. These constraints are described below.

           Critical engine speed
To avoid the engine from stalling, the engine speed must be always greater than a critical value.
                                            N  N Critical                                          (1)
Where
        N Critical is the minimal engine speed before the engine stalls (900 rpm).

           Conditions to be in Homogeneous/Stratified mode

                Homogeneous mode constraints
The engine can run in homogeneous mode if the following conditions are respected:
   IGA m in  IGA  IGA base hom ogeneous ,

                                    SIEMENS CONFIDENTIAL                                             3
                                                               Engine benchmark

        180  EOI   
                     (Fuel injection only during intake stroke).
  EOI  SOI  360
Where
      IGA m in  f NL28 M air , N  ,
          IGA base hom ogeneous  f NL29 M air , N  .

                  Stratified mode constraints
The engine can run in stratified mode if the following conditions are respected:
  f NL31 TQI S reference, N   1 (Defines a working engine domain),
    M air  M air m in ,
    IGA  IGA referencestratified ,
    EOI  EOI referencestratified ,
  EOI  SOI  180 (Fuel injection only during compression stroke),
  1  RATIO air / fuel .
Where
       M air m in  f NL30 TQI S reference, N  ,
          IGA referencestratified  f NL12 TQI S reference, N  ,
          EOI referencestratified  f NL13 TQI S reference, N .

                Switching mode constraint
         The commutation between stratified and homogeneous modes must occur from one cylinder
         to the next cylinder fuel injection (No possibility to go from one mode to the other in a
         smooth way).

   EMS implementation constraint
Due to physical engine constraints and despite the performance growth of new automotive ECU,
real time constraint still subsists and must be taking into account for the design of the EMS.
The input/output sampling time of the controller cannot be lower than 5 ms.
Concerning the computational delays inherent to digit controllers, due to the approximation of the
engine model, these delays can be neglected.

    Controller performances
This section defines several quantitative and qualitative criteria to evaluate the different control
strategies.

    Performance criteria

            1.1.1.1. Air/fuel ratio criterion
In order to evaluate the regulation of the air/fuel ratio in homogeneous mode, the following criterion
has been defined.

                                                1                   Ratio air/fuel  Ratio air/fueld               (2)
                      C Ratio air/fuel 
                                           Thom ogeneous      
                                                           Scenario
                                                                         Ratio air/fuelm ax
                                                                                                      mode  dt

where
         Thomogeneous is the period where the engine operates in homogeneous mode,
         Mode is a boolean flag (0: Stratified mode, 1: Homogeneous mode),

4                                                   SIEMENS CONFIDENTIAL
                                                        Engine benchmark

        Ratio air / fuelmax   is the maximal variation accepted ( Ratioair / fuelmax  0.02 ).

            1.1.1.2. Torque criteria
Several criteria have to be considered to evaluate the torque performances.

In the case of the reduced engine benchmark, one of the torque criteria evaluate the performances of
the EMS to follow the desired torque (TQEd).
                                                        1              TQE  TQE d                                   (3)
                                          C Torque1                               dt
                                                        T   Scenario    TQE m ax

where
        T is the scenario period considered,
        TQE max is the maximal torque variation accepted (2 Nm).
In the case of the intermediate benchmark, this criterion is not considered.

During acceleration and deceleration phases, in order to preserve car driveability, it is necessary to
limit the produced torque gradient in relation to time and the gear engaged. These limitations can be
expressed as follow.


                                                Yes                                 No
                                                                   Gear=1




            Yes                               No                                          Yes                No
                      TQE  5 Nm                                                                TQE  5 Nm




  dTQE                                   dTQE                                 dTQE                       dTQE
        35 Nm/s                               60 Nm/s                             50 Nm/s                   120 Nm/s
    dt                                     dt                                   dt                         dt

                                        Figure 3 Torque gradient limitations
The corresponding criterion is defined as follow.
                                                              dTQE dTQE                                            (4)
                                                              dt  dt max ,0 
                                                          max                 
                                               1                              
                                   C Torque2                                    dt
                                               T Scenario       dTQE
                                                                          0.2
                                                                  dt max

where
        T is the scenario period considered,
        dTQE
                 is the maximal torque gradient authorised (cf. Figure 3).
          dt max



Another torque criterion has been defined to evaluate the performances during the commutation
phases between the two operating mode of the GDI engine.

                                              SIEMENS CONFIDENTIAL                                                     5
                                                           Engine benchmark

The main problem during the switching phase is to ensure that the driver does not feel any
hiccough. This performance can be defined with the following condition.
                                      TQEbeforecommutatio  TQEaftercommutatio  10 Nm
                                                        n                    n
                                                                                                                     (5)

                             TQE
                                                                                  Commutation

                                TQEbefore commutation



                                 TQEafter commutation

                                                                                                t
                                                                           tk   tk+Te

                                      Figure 4 Commutation mode performance
The respective criterion is defined as follow.
                                                                                               dTQE                (6)
                                               max TQE beforecommutatio  TQEaftercommutatio 
                                                                       n                    n             Te ,0 
                           1                                                                     dt max         
        C Torque3                      
                      Nb commutatio
                                  n   Scenario                       TQE commutatio max
                                                                                     n

where
         Nb commutatio is the number of commutation between the two modes,
                     n
        Te is the sample time of the controller,
         dTQE
                  is the maximal toque gradient authorised (cf. Figure 3),
           dt max
         TQE commutatio max is the maximal torque variation authorised during commutation (10 Nm).
                       n


           1.1.1.3. Car speed criterion
In the case of the intermediate benchmark, in order to evaluate the car speed regulation, the
following criteria has been defined.
                                                          1          V  Vd                                          (7)
                                                CV             Vmax  dt
                                                          T Scenario
where
        T is the scenario period considered,
        Vmax is the maximal variation accepted (2 Km/h).

           1.1.1.4. Idle speed criterion
In the case of the intermediate benchmark, in order to evaluate the idle speed regulation, the
following criterion has been defined.

                                                       1             N  N Idle                                      (8)
                                         C N Idle              N Idle m ax  idle  dt
                                                      TIdle Scenario

where
        Tidle is the sum of the idle phase period for the scenario considered,
        idle is a boolean flag (0: non-idle phase, 1: idle phase),
         N Idlemax is the maximal variation accepted ( N Idlemax  50 rpm ).



6                                               SIEMENS CONFIDENTIAL
                                                    Engine benchmark

The idle phases start when the accelerator pedal position is null (AcceleratorPedal=0) and the
engine speed is lower than N Idle  N Idlemax and finish when the accelerator pedal position is non null.

                       AcceleratorPedal




                                                                                              t
                                  N




                           NIdle                                                              2.NIdle max
                                                                                              t
                                 idle


                             1

                             0
                                                                                              t

                                   Figure 5 Idle speed phase definition

           1.1.1.5. Throttle control criterion
In order to evaluate the solicitations of the throttle actuator, the following criterion has been
defined.

                                                  MTC        k    MTCk 1                                   (9)
                                 C Throttle    Scenario
                                                                                     Temin
                                                           n  Te
where
        MTCk and MTCk -1 are the control actions at time k and k-1 respectively,
        Te min is the minimal sample time of the controller (0.005 s),
        n is the number of control actions made within the scenario period.

         1.1.1.6. Consumption criterion
The consumption criterion is the mean consumption on the scenario.
                                                               Q fuel.dt                                    (10)
                                  C Consumption           scenario
                                                                                    10   2
                                                     fuel density        V  dt
                                                                      scenario




                                         SIEMENS CONFIDENTIAL                                                   7
                                                                   Engine benchmark


           1.1.1.7. Criteria aggregation
In order to reduce the number of criteria, several criteria defined in the above sections have been
aggregated with the same weight factor.
Therefore, the controller performances are evaluated with three different criteria depending on the
engine benchmark considered.
    Reduced engine benchmark

                                                     i 3                                                                      (11)
                                                      CThrottle i
                                                     i 1
                                      C Throttle 
                                                             3
                                                            i 3
                                                             CConsum ption i
                                                            i 1
                                      C Consum ption 
                                                            3
                                                   i 3
                                                      1
                                                                                              
                                                                                               
                                                   C Ratio air/fueli  CTorque1i  CTorque3i 
                                      C Global  i 1                                         
                                                        3
                                                                         3
where
    CThrottle i , CConsumption i , Ratioair/fueli , CTorque1i , CTorque3i   are the criteria obtained with the ith scenario,
    Intermediate engine benchmark
                              C Throttle ,                                                                                     (12)
                              C Consumption ,

                              C Global 
                                             1
                                             5
                                                                                                      
                                               C Ratio air/fuel  C Torque2  C Torque3  C V  C Nidle .


For a criterion point of view, the objective of the controller is to minimise as much as possible the
three criteria.

   Implementation performances
The implementation capability of the different control strategies on a real time ECU must be taken
into account for the performance analysis.
Indeed, very good controllers can be unrealistic, criteria like real time consumption, but also
easiness of the implementation and tuning have to be considered.

            1.1.1.8. Evaluation of time consumption for each controller
In fact it is really difficult or even impossible to value the real time execution of a controller
designed in a MATLAB/SIMULINK environment.
Therefore, a relative comparison of the MATLAB/SIMULINK execution times, for the same
scenario in the same environment will give a ‘controller execution performance’ quotation.
This evaluation will be made within SIEMENS with the assistance of FAMIMO partners.
The controllers, which fulfil the ‘controller performances’ for the given constraints, will be tested
on the same computer.

            1.1.1.9. Implementation and tuning of the controllers
It is always difficult to value the easiness to implement and tune a strategy. Nevertheless these
parameters must be taken into account in a qualitative way for the evaluation performance of each
controller.
Some indicators have to be set up:
    Number of parameters to be tuned,

8                                                    SIEMENS CONFIDENTIAL
                                                          Engine benchmark

  Legibility of the controller, physical meaning of the parameters,
  Complexity and legibility of the tuning procedure,
  Pre-requirement for the controller design (model…).

In addition it is necessary to consider the ability of the controllers to be used for equivalent
processes without major or structural modifications (Portability).

   Expert knowledge required
The approach chosen in the FAMIMO project to develop the engine controller is to consider the
engine benchmark as a ‘grey box’. So, the structure of the engine model (mathematical equations,
look-up tables) must not be directly used for the design of the controller. FAMIMO partner can only
use the engine model in order to learn some basic expert knowledge for the engine control (For
example, considering the homogeneous engine torque structure, the following statement can be
expressed: ‘In the homogeneous mode, the engine torque mainly depend on the air mass introduced
into the cylinder and can be degraded with the ignition advance and the air/fuel ratio’).
Nevertheless, for control purpose some look-up tables can directly be used in the controller in order
to express the various engine constraints. They are listed below:
    IGA m in  f NL28 M air , N  ,
   IGA base hom ogeneous  f NL29 M air , N  ,
  Definition of the working engine domain in stratified mode: f NL31 TQI S reference, N ,
   M air m in  f NL30 TQI S reference, N  ,
   IGA referencestratified  f NL12 TQI S reference, N  ,
   EOI referencestratified  f NL13 TQI S reference, N .

Note:
To express the stratified constraints (and only for that), the reference torque ( TQI S reference) is
needed. It can be evaluated from the look-up table:
                                                   TQI S reference  f NL10 M fuel , N 




                                                 SIEMENS CONFIDENTIAL                              9
    Engine benchmark


          II.




SIEMENS CONFIDENTIAL   11

								
To top