EP0886055B1 - Method and apparatus for controlling a spark ignited internal combustion engine - Google Patents

Description

The invention relates to a method for controlling an engine. internal combustion, positive ignition and injection electronically controlled, with at least three actuators, acting on the ignition advance, the engine air flow control and richness of the air-fuel mixture, and several sensors to determine the operating point of the engine. This control process essentially consists by regulating the torque supplied by the motor. She also relates to a device for implementing this process.

Today's motor vehicles are increasingly equipped more electronic equipment and services news, in addition to electronic engine control, such as the electronic control of the gearbox speed, anti-lock braking system when braking or electronic regulation of air conditioning by example. All this equipment induces variations of the torque consumed, so that the correct operation of the engine and associated equipment driver comfort and comfort passengers, requires control at all times and precisely the torque delivered by the engine.

This torque delivered by the motor can be modulated by variation of command values applied to actuators, such as the amount of air admitted to the manifold, the amount of fuel mixed with air, or the instant of ignition of the air / fuel mixture. These actuators have different effects on the torque, in terms of speed of action due to pure delay or a rise time, and in terms of authority by the magnitude of the effects. They also have effects on other quantities whose regulation must be otherwise assured, such as wealth in fuel of the fuel mixture, for reasons of consumption and pollution. That's why it is desirable to limit the impact of the regulation of engine torque on these other quantities, while advantageously using several of these actuators.

Current solutions to the problem of regulating engine torque use regulation technology so-called proportional integral derivative, without engine modeling, which has the disadvantage of do not know the effects of commands applied by this regulation only by observing its effects on regulated quantities, causing a delay detrimental to the quality of regulation.

Some solutions have however been developed with an engine model from which regulation can determine the commands to be applied, in order to obtain a desired effect on the quantities to be regulated.

An example of current implementation of regulation of engine torque is described in the European patent EP 0 185 552, on behalf of NIPPONDENSO. The mode of regulation uses several linear models identified tangent to several points of engine operation, intended for regulation integral quadratic linear multivariable to regulate the engine torque while minimizing the fuel consumption, from air controls admitted to the manifold and injected gasoline.

The main disadvantage of this solution is due to approximations related to the use of models linearized tangents while the operation of the motor is not linear, causing degradation torque regulation performance.

In addition, these models do not exploit knowledge that we have physical equations describing the phenomena which intervene in the functioning real engine.

Another disadvantage of this example of regulation is its high cost due to the addition of a sensor chamber pressure to measure the torque supplied by the engine.

The invention aims to solve these drawbacks by proposing a method of controlling the engine by a motor torque regulation, multivariable type integral quadratic linear based on a non-model linear of the motor, using as variables of controls the ignition advance and the amount of air allowed in the collector, and the quantity of gasoline injected in a particular case, and receiving two torque setpoints calculated simultaneously according to the engine operating model.

• en phase de développement du moteur :
  • modélisation dynamique linéaire de la production du couple moteur au moyen d'une fonction logarithme et de sa fonction inverse ;
  • définition du gain optimal minimisant l'indice de performance quadratique de la régulation linéaire quadratique intégrale, à partir duquel les variables de commande U_k seront calculées en fonction des variables d'état X_k et de consigne Xs_k, et sa mémorisation dans le calculateur ;
• en cours de fonctionnement du moteur, à chaque point mort haut d'ordre k, calcul des variables d'état X_k et des consignes Xs_k, par le calculateur électronique, à partir :
  • de la mesure du régime N_k et d'une grandeur représentative de la masse d'air absorbée réellement par le moteur, comme la pression collecteur P_k ;
  • du calcul de la masse d'air M_AIRMOT sortant du collecteur en fonction du régime et de ladite grandeur ;
  • du calcul de la richesse et du rendement de richesse à partir du temps d'injection et de la masse d'air M_AIRMOT ;
  • du calcul de la masse d'air M_AIRPAP entrant dans le collecteur en fonction de la masse d'air de commande M_AIRCMD de l'actionneur d'air;
  • du calcul du couple réel CMI⁺ et du couple projeté CP fournis par le moteur;

puis calcul des variables des commandes U_k selon la régulation linéaire quadratique intégrale.For this, a first object of the invention is a method of controlling a combustion engine, with spark ignition, with injection controlled by an electronic computer, having at least three actuators to which an air control are respectively applied, an ignition control and a fuel mixture richness control, consisting, from a model of the production of the engine torque, in an integral quadratic linear type regulation of engine operating state variables by action on control variables as a function of setpoint variables and state variables, characterized in that it comprises the following steps:

• in engine development phase:
  • linear dynamic modeling of the production of the engine torque by means of a logarithmic function and its inverse function;
  • definition of the optimal gain minimizing the quadratic performance index of the integral quadratic linear regulation, from which the control variables U _k will be calculated as a function of the state variables X _k and setpoint Xs _k , and its storage in the computer ;
• during engine operation, at each top dead center of order k, calculation of state variables X _k and setpoints Xs _k , by the electronic computer, from:
  • measuring the speed N _k and a quantity representative of the mass of air actually absorbed by the engine, such as the manifold pressure P _k ;
  • calculating the mass of air M _AIRMOT leaving the manifold as a function of the speed and of said quantity;
  • calculation of the richness and the richness yield from the injection time and the air mass M _AIRMOT ;
  • calculating the mass of air M _AIRPAP entering the manifold as a function of the mass of control air M _AIRC® of the air actuator;
  • calculating the actual torque CMI ⁺ and the projected torque CP supplied by the engine;

then calculation of the variables of the commands U _k according to the integral quadratic linear regulation.

A second object of the invention is a device for implementation of the torque regulation process engine, consisting of the electronic computer of motor control including a linear regulator integral quadratic.

• figure 1 : le schéma fonctionnel de la modélisation d'un moteur thermique, recevant une commande de débit d'air et fournissant un couple;
• figure 2 : le schéma fonctionnel de la modélisation d'un moteur thermique, qui fonctionne à richesse unique;
• figures 3 et 5 : le schéma fonctionnel du dispositif mettant en oeuvre le procédé de régulation du couple moteur selon l'invention, respectivement avec commande de richesse et richesse constante, dans le cas d'un papillon avec vanne additionnelle d'air ;
• figure 4 et 6 : le schéma fonctionnel mettant en oeuvre le procédé de régulation du couple moteur selon l'invention, respectivement avec commande de richesse et richesse constante, dans le cas d'un papillon motorisé.

Other characteristics and advantages of the invention will appear on reading the description of the process and an embodiment of the device for implementing the process, illustrated by the following figures which are:

• Figure 1: the block diagram of the modeling of a heat engine, receiving an air flow command and providing a torque;
• FIG. 2: the functional diagram of the modeling of a heat engine, which operates at single richness;
• Figures 3 and 5: the block diagram of the device implementing the method of regulating the engine torque according to the invention, respectively with richness control and constant richness, in the case of a butterfly valve with additional air valve;
• FIGS. 4 and 6: the functional diagram implementing the method of regulating the engine torque according to the invention, respectively with richness control and constant richness, in the case of a motorized throttle valve.

The first step in the torque regulation process engine consists in performing dynamic modeling of the production of the engine torque reflecting the physical operation of the engine. this modeling is performed during the engine development phase, on bench test.

R(N_k,P_k)

en N.m/kg,

N

en tour/min,

P

en mbar.

First of all, the torque supplied by the engine, with optimal ignition advance and richness of the air-fuel mixture equal to 1, is defined from the mass of air M _{AIRMOT (k)} leaving the intake manifold and entering a cylinder of the engine, at each top dead center of order k, and of the combustion efficiency of the engine R (N _k , P _k ) which is a function of the value N _k of the speed and the value P _k of pressure in the manifold.
This engine torque C _{AIRMOT (k)} is expressed in Newton meters, according to the following equation (E ₁ ): VS AIRMOT (k) = R (N k P k ) * M AIRMOT (k) with

R (N _k , P _k )

in Nm / kg,

NOT

in rev / min,

P

in mbar.

This value must be corrected as a function of the ignition advance Av and of the richness Ri of the mixture, by feed efficiencies µ _{Av (k)} and richness µ _{Ri (k)} defined at each top dead center d 'order k, according to the following equation (E ₂ ), so as to obtain the real value CMI ⁺ of the torque supplied by the motor: CMI + (K) = C AIRMOT (k) * µ V (k) * µ Ri (k - 2) = R (N k P k ) * M AIRMOT (k) * µ V (k) * µ Ri (k - 2)

The richness yield µ _Ri is a mapped function of the function R _i for example, and the advance yield µ _AV is a mapped function of the advance, the speed, the manifold pressure and possibly the richness.

This expression of the couple being multiplicative, it is non-linear. According to the invention, the method calculates the logarithm thereof and performs a variable change. This allows, from multiplicative equations, to use a couple model which is additive without any approximation being made. The new variable is therefore the logarithm logCMI ⁺ of the couple, expressed by the following equation (E ₃ ): logCMI + (K) = logC AIRMOT (k) + logµ V (k) + logµ Ri (k-2)

According to the invention, another term is also defined corresponding to a projection of the previous torque CMI ⁺ , a few dead centers later, with the aim in particular of anticipating the commands of the actuators when possible. This projected torque CP is therefore defined as the torque supplied by the engine, according to an advance yield setpoint µ _Avcons and a wealth yield setpoint µ _Ricons , by the following equation (E ₄ ): CP (K) = R (N k P k ) * M AIRMOT (k) * µAv cons (k) * µ Ricons (k) = C AIRMOT (k) * µ Avcons (k) * µ Ricons (k) .

This expression of the couple being nonlinear, we calculate its logarithm according to the following equation (E ₅ ): logCP (K) = logC AIRMOT (k) + logµ Avcons (k) + logµ Ricons (k)

In stabilized operation, the feed rate and the richness are equal to their set values, so that the projected torque CP is equal to the torque CMI ⁺ .

Two torque setpoints are sent to the torque regulator and are calculated simultaneously, one CMI ⁺ _cons from the CMI ⁺ torque supplied by the engine and which the regulator must follow as quickly as possible, and the other CP _cons , which is in the longer term, from the projected pair CP.

The modeling of the engine according to the invention also consists in describing the engine air intake manifold by a first order model, from the mass of air M _AIRPAP entering the manifold by the throttle body, at each top dead center, whether this case is constituted by a single butterfly or that it includes other actuators. The term describing the manifold expresses the mass of air M _AIRMOT leaving the manifold to enter a cylinder, according to the following equation (E ₆ ): M AIRMOT (k) = M AIRMOT (k-1) + F collar * [M AIRPAP (k) - M AIRMOT (k-1) ] in which the term F _col is a filtering factor translating the dynamics of the collector and expressed according to the following equation (E ₇ ), from the total displacement V _word of the engine, the number of cylinders n _cyl , the volume V _col of the manifold and of the average filling rate Remp of the engine with air, related to the conditions of the manifold. (E 7 ): F collar = {V word * Remp} / {n cyl * V collar }

To ensure compatibility with the writing of the torque, the collector model is linked to the C _AIRMOT torque supplied by the engine and expressed by equation (E ₁ ). By analogy, it is necessary to define from the mass of air M _AIRPAP entering the collector, expressed in kilograms, a quantity homogeneous to a couple, according to the following equation (E ₈ ): vs AIRPAP (k) = R (N k P k ) * M AIRPAP (k) .

By neglecting the variations in the combustion efficiency R (N _k , P _k ) of the engine, between two top dead centers of indices k and k + 1, the above equation (E ₆ ) becomes: VS AIRMOT (k + 1) = C AIRMOT (k) + F Collar *[vs AIRPAP (k + 1) - VS AIRMOT (k) ].

We will admit that the difference between the values of the pairs C _AIRMOT and C _AIRPAP , ie C _AIRPAP - C _AIRMOT , is small _compared to C _AIRMOT and that we can use the approximation log (1 + x) ≈ x, so that the collector model is written according to the following equation (E ₉ ): logC AIRMOT (k + 1) = logC AIRMOT (k) + F Collar * [LogC AIRPAP (k + 1) -logC AIRMOT (k) ]

Different air actuators can be used in an engine, such as an RCO valve, a motorized throttle valve or a stepper. The dynamics of the air actuator used is described as a first-order model, as part of the theoretical modeling of the production of engine torque, from the air flow Q _AIRPAP entering the manifold and the flow d air Q _AIRCMD actuator control.

The following equation (E ₁₀ ) expressing the air flow Q _{AIRPAP (k + 1)} which enters top dead center of order k + 1: Q AIRPAP (k + 1) = Q AIRPAP (k) + F act * [Q AIRCMD (k) - Q AIRPAP (k) ] with the filtering factor F _act which translates the dynamics of the air actuator, must be brought back in the form of torque, whose logarithm must then be calculated before a variable change.

However, the air flow rate Q _{AIRPAP (k)} is connected to the mass of air M _{AIRPAP (k)} entering the manifold by the following expression (E ₁₁ ), as a function of the value N _k of the engine speed and the number of cylinders n _cyl : (E 11 ): M AIRPAP (k) = {120 * Q AIRPAP (k) } / {not cyl * NOT k } so that the above equation (E ₈ ) becomes: vs AIRPAP (k) = R (N k P k ) * M AIRPAP (k) = R (N k P k ) * {120 * Q AIRPAP (k) } / {not cyl * NOT k }.

By analogy, the air flow Q _AIRCMD for actuator control gives rise to a torque value C _AIRCMD defined according to the following equation (E ₁₂ ): VS AIRCMD (k) = R (N k P k ) * {120 * Q AIRCMP (k) } / {not cyl * NOT k }.

Thus, the equation (E ₁₀ ) of the air flow entering the manifold, modeling the behavior of the air actuator, is brought back in the form of a torque which is written, neglecting the variations of the regime N and the yield R (N _k , P _k ): VS AIRPAP (k + 1) = C AIRPAP (k) + F act * [VS AIRCMD (k) - VS AIRPAP (k) ].

We will admit that the difference between C _AIRPAP and C _AIRCMD remains small _compared to C _AIRPAP , so that the calculation of the logarithm of C _AIRPAP results in the following equation (E13): logC AIRPAP (k + 1) = logC AIRPAP (k) + F act * [LogC AIRCMD (k) -logC AIRPAP (k) ].

(E₃) : logCMI+ (k) = logCAIRMOT(k) + logµAv(k) + logµRi(k-2)

(E₅) : logCP(k) = logCAIRMOT(k) + logµAvcons(k) + logµRicons(k)

(E₉) : logCAIRMOT(k+1)=logCAIRMOT(k) + Fcol*[logCAIRPAP(k+1)-logCAIRMOT(k)]

(E₁₃) : logCAIRPAP(k+1)=logCAIRPAP(k)+Fact*[logCAIRCMD(k)-logCAIRPAP(k)].

The production of the engine torque thus modeled according to the invention is represented by the following four equations, representing respectively, (E ₃ ) the engine torque, (E ₅ ) the projected torque, (E ₉ ) the operation of the collector and (E ₁₃ ) the operation of the air actuator:

(E ₃ ): logCMI + (K) = logC AIRMOT (k) + logµ V (k) + logµ Ri (k-2)

(E ₅ ): logCP (K) = logC AIRMOT (k) + logµ Avcons (k) + logµ Ricons (k)

(E ₉ ): logC AIRMOT (k + 1) = logC AIRMOT (k) + F collar * [LogC AIRPAP (k + 1) -logC AIRMOT (k) ]

(E ₁₃ ): logC AIRPAP (k + 1) = logC AIRPAP (k) + F act * [LogC AIRCMD (k) -logC AIRPAP (k) ].

Figure 1 is the _block diagram of the modeling of the production of the engine torque receiving an air flow control C _AIRCMD applied to the throttle body 1 which delivers a mass of air entering the manifold 2 and providing a torque, in the form logCMI ⁺ logarithm and logCP logarithm, to torque regulator 3.

In order to ensure the return of the logarithms of the motor torque logCMI ⁺ , of the projected torque logCP and of the richness yield logµ _Ri towards their respective reference values, the invention defines three new variables which are the discrete time integrals of these three quantities : ΣlogCMI + (K + 1) = ΣlogCMI + (K) + elogCMI + (K) ΣlogCP (k + 1) = ΣlogCP (k) + elogCP (k) Σlogμ Ri (k + 1) = Σlogµ Ri (k) + elogµ Ri (k) with the following error terms: elogCMI + = logCMI + - CMI log + cons elogCP = logCP - logCP cons elogμ Ri = logµ Ri - logµ Ricons these errors being equal to the difference between the logarithm of the quantities measured and the logarithm of their respective set values.

An alternative embodiment of the method according to the invention relates to an operation of the engine unique wealth. Figure 2 is the block diagram modeling the torque production of a single-wealth engine.

According to the invention, after the development phase, the control process is carried out during engine operation and consists of regulation of the engine torque developed from the linear model of the engine which has just been described, and of linear type integral quadratic.

The linear model of the motor can be described by its state representation: X k + 1 = A * X k + B * U k + F * Xs k xs k + 1 = As * Xs k , in which X _k represents the state vector of the model, constituted by variables describing the state in which the engine is located;
U _k represents the commands on which the torque regulator can act to ensure the operation of the engine, and
Xs _k represents the variable quantities which have an effect on the operating state of the engine, according to the equations of the engine model, but which are not controlled by the regulator, that is to say the disturbances.

Thus, as shown in table 1 in the appendix, the matrix of state variables X _k contains on the one hand the logarithm logC _AIRPAP of the quantity C _AIRPAP proportional to the mass of air entering the manifold through the throttle body, and that log _CAIRMOT of the quantity C _AIRMOT proportional to the mass of air leaving the manifold, on the other hand the error of the logarithm of the torque supplied by the engine elogCMI ⁺ _k-1 , the error of the logarithm of the projected torque elogCP _k-1 , the error of the logarithm of the wealth yield for two consecutive high dead points elogµ _{Ri (k-1} ) and elogµ _{Ri (k-2} ), and finally the three previously defined integral variables ΣlogCMI ⁺ _(k) , ΣlogCP _(k) and Σlogµ _{Ri (k)} .

The disturbance matrix Xs _k contains the logarithm of the torque setpoint supplied by the logCMI ⁺ _cons engine and of the projected torque setpoint logCP _cons , and the logarithm of the advance setpoint _returns logµ _Avcons and the wealth setpoint logµ _Ricons .

The matrix U _k of the commands comprises the logarithm logC _AIRCMD of the quantity proportional to the actuator control air flow, the logarithms logµ _Av and logµ _Ri of the ignition advance and of the richness.

Integral quadratic linear regulation is state feedback regulation, so that at each iteration, the command U _k is obtained from a gain matrix K calculated once and for all during the development phase of the regulator: K = [KK S ]

où les matrices Q et R sont des matrices de pondération définissant les caractéristiques de la régulation. La résolution de ce problème conduit au calcul suivant de la matrice de gain K, K = [K KS] : P = Q + AT.P.A - AT.P.B. (R + BT.P.B)-1.BT.P.A K = (R + BT.P.B)-1.BT.P.A Ps = (A - B.K)T.(P.F + Ps.As) KS = (R + BT.P.B)-1.BT.(P.F + Ps.As) P étant l'inconnue de l'équation matricielle.The theory of integral quadratic linear regulation gives the matrix K as the optimal gain minimizing the quadratic performance index:

where the matrices Q and R are weighting matrices defining the characteristics of the regulation. The resolution of this problem leads to the following calculation of the gain matrix K , K = [KK S ]: P = Q + A T .PA - A T .PB (R + B T .PB) -1 .B T .PA K = (R + B T .PB) -1 .B T .PA P s = (A - BK) T . (PF + Ps.As) K S = (R + B T .PB) -1 .B T . (PF + Ps.As) P being the unknown of the matrix equation.

It is possible to use several gain matrices depending on the operating mode used for the engine, favoring either the ignition advance, the wealth or anti-pollution for example.

As for the matrices A, B, F and As coming from the modeling of the engine to describe the evolution of the state variables X _k as a function of time, they are obtained from the four equations E ₃ , E ₅ , E ₉ and E _{13 above} and appear in table 2 in the appendix. The state return matrix K is calculated from these matrices.

In the particular case of a simplified control of the engine by the ignition advance and the air flow, with single richness, the representation matrices of the state variables X _k , U _k and Xs _k do not have of variable relating to wealth. They are shown in table 3 in the appendix, while the matrices A, B, F and As which are also simplified are shown in table 4.

Figure 3 is the block diagram of the device implementing the method of regulating the engine torque according to the invention, in the case of an engine 4 whose throttle body 5 comprises a mechanical throttle 6, including the opening control S _p is determined by depressing the accelerator pedal, associated with an additional air valve 7, the opening of which is electronically controlled, in other words section S _v .

FIG. 4 is the same diagram in the case of a motorized throttle valve 8 whose opening S _pm is completely controlled.

In FIG. 3, the engine 4 receives a command to open the throttle valve S _p as well as an electronic command for ignition advance Av, for opening the valve S _v and possibly for injection time Ti, and in FIG. 4, the engine receives an opening command S _pm from its motorized throttle valve 8. At each top dead center of order k, the speed value N _k and that of the manifold pressure P _k or of any other quantity representative of the mass of air actually absorbed by the engine, such as the air flow rate, entering the manifold for example. The air flow is measured by a flow meter and the manifold pressure is measured by a pressure sensor. The electronic engine control computer 9 calculates the mass of air Q _AIRMOT leaving the manifold, in means 10, then the richness Ri and the richness yield µ _Ri of the mixture from this air mass and the time d injection Ti, in means 11 and 12. From the sections of the butterfly S _p and of the valve S _v , in the case of FIG. 3, and from the section S _pm of the motorized butterfly in the case of the figure 4, of the advance Av and of the injection time Ti, the modules 13 of the computer determine the torque values defined above, ie the supplied torque CMI ⁺ , the projected torque CP, the torque defined from the mass of air entering the _AIRPAP collector C and that leaving the _AIRMOT C _collector , the instructions for the couples supplied CMI ⁺ _cons and projected CP _cons , the instructions for the wealth returns µ _Ricons and in advance µ _Avcons and the wealth returns µ _Ri . Means 14 compute the logarithm of these values which enter into an integral quadratic linear regulator 15, the role of which is to reconstruct the state vector of the torque model and the matrix product to deliver the logarithm of the feed efficiency commands logµ _Av , the air flow of the logC actuator _AIRCMD and possibly the logµ _Ri richness efficiency. Means 16 calculate the exponential of these values, which is then clipped in means 17, to take account of the limitations linked to the operation of the actuators and of the motor, which deliver the advance efficiency μ _Av , the air control C _AIRCMD and the wealth yield µ _Ri .
Means 18 calculate the value of the ignition advance Av from the yield µ _Av , the speed N, the manifold pressure P and possibly the richness Ri.
Means 19 calculate the final control of the air actuator 5, either the section of the valve S _v or the opening S _pm of the motorized throttle valve.
Means 20 calculate the richness Ri from the richness yield control μ _{Ri as} well as from a richness yield table as a function of the richness. Means 21 then calculate the injection time Ti on the basis of this richness objective and the operating conditions of the engine, such as the pressure in the manifold, the speed, etc. From this injection time, means 22 estimate the richness in each cylinder and the associated richness yield.

FIGS. 5 and 6 are the same functional diagrams of the regulation device as FIGS. 3 and 4, respectively with a throttle associated with an air valve and with a motorized throttle, but in the case of an engine operating with single richness . The wealth yield µ _Ri and the wealth yield setpoint µ _Ricons are equal to 1.

A first advantage of the engine control method according to the invention comes from the use of torque control from a non-linear model of the engine, which allows the regulator of the electronic computer to know precisely the effects of controls the behavior of the engine and determine the values of the controls to cause the expected effect.
In addition, the interest of formulating the model of the motor torque, described by multiplicative equations, by a logarithmic function is to obtain a single linear model, which describes in all points the non-linear operation of the motor. This results in a gain in performance for the regulation proposed by the invention compared to the regulations currently existing.
On the other hand, the regulation of the motor torque is multivariable, that is to say that it acts simultaneously on several control variables to simultaneously regulate several state operating variables. This is particularly advantageous since each of the state variables is thus regulated taking account of the couplings between the different state and control variables, and it is then possible to exploit the complementarity of the different commands in terms of authority and speed of action.
Another advantage comes from the definition of the projected engine torque, which makes it possible, in certain operating modes, to position the different commands in advance in a coordinated manner to better respect the engine torque setpoint, by anticipating the dynamics or the delays in the operation of controls.
Finally, this regulation of the engine torque makes it possible, by the addition of simple methods, to take charge of the regulation of other quantities, such as the idling speed, without the development of a complete regulation method acting directly on the actuators. This advantage results in a reduction in the complexity and in the size of the processes to be implemented in the electronic engine computer, as well as in a simplification of the procedures for calibration and development of the processes.

Claims (8)

1. Method for control of the torque of an internal combustion engine with spark ignition, with injection controlled by an electronic computer, having at least three actuators to which air control, ignition control and fuel mixture richness control are respectively applied, consisting, from a model of the engine torque, of producing regulation, of the integral quadratic linear type, of variables of the state of operation of the engine by acting upon control variables according to reference variables and state variables, characterised in that it comprises the following steps:

   in the engine development phase:

   linear dynamic modelling of the engine torque production by means of a logarithmic function and its inverse function;

   definition of the optimum gain minimising the quadratic performance index of integral quadratic linear regulation, from which the control variables (U_K) will be calculated according to state variables (X_K) and reference variables (Xs_K), and storage thereof in the computer;

   during operation of the engine, at each top dead centre of the order k, calculation of state variables (X_K) and reference variables (Xs_K) by the electronic computer, from:

   measurement of the engine speed (N_K) and a representative value of the air mass actually absorbed by the engine;

   calculation of the air mass (M_AIRMOT) exiting the manifold according to the engine speed and to said value;

   calculation of the richness (Ri) and richness efficiency (µ_Ri) from the injection timing and the air mass (M_AIRMOT);

   calculation of the air mass (M_AIRPAP) entering into the manifold according to the controlling air mass (M_AIRCMD) for the actuator;

   calculation of the actual torque (CMI⁺) and the projected torque (CP) supplied by the engine;

   then calculation of the variables of the controls (U_K) according to integral quadratic linear regulation.
2. Method for control of torque according to claim 1, characterised in that the value representative of the air mass actually absorbed by the engine is the pressure (P_K) in the manifold, or the rate of air entering into the manifold.
3. Method for control of torque according to claim 2, characterised in that the modelling of the torque production is defined by four linear equations obtained from the effects produced during the operation of the engine, by:

   the rate of controlling air (Q_AIRCMD) for the air actuator over the rate of air (Q_AIRPAP) entering into manifold according to the following first order equation: QAIRPAP(k+1) = QAIRPAP(k) + Fact*[QAIRCMD(k) - QAIRPAP(k)] where F_act : a filtering factor translating the dynamics of the air actuator;

   the air mass (M_AIRPAP) entering into the manifold via the throttle valve over the air mass (M_AIRMOT) exiting the manifold and entering into a cylinder according to the first order equation: MAIRMOT(k) = MAIRMOT(k-1) + Fcol * [MAIRPAP(k) - MAIRMOT(k-1)] where F_col : a filtering factor translating the dynamics of the manifold;
   the air mass (M_AIRMOT) exiting the manifold on the one hand over the torque (CMI⁺) supplied by the engine according to the advance efficiencies (µ_AV) and richness (µ_Ri) and the engine combustion efficiency R(N_K, P_K): CMI+ (k) = R(NK, PK) * MAIRMOT(K) * µAv(K) * µRi(K-2) and on the other hand, on the projected torque (CP) supplied by the engine according to the reference efficiency of the advance (µ_Avcons) and the richness (µ_Ricons) and the engine combustion efficiency R(N_K, P_K): CP(K) = R(NK, PK) * MAIRMOT(K) * µAvcons(K) * µRicons(K),

   these equations being subsequently linearised by calculation of their logarithm after a change in the variable.
4. Method for control of torque according to claim 3, characterised in that:

   a) the state variables (X_K) of engine operation are:

   the logarithms of two values homogeneous to a torque, one (C_AIRPAP) proportional to the air mass (M_AIRPAP) entering into the manifold via the throttle valve and the other (C_AIRMOT) proportional to the air mass (M_AIRMOT) exiting the manifold;

   the errors in the torque logarithms supplied by the engine (elogCMI⁺), in the projected torque (elongCP) and in the richness efficiency (elogµ_Ri) for two consecutive top dead centres, which errors are equal to the difference between the logarithm of the values measured and the logarithm of their respective reference values: elogCMI+ = logCMI+ - log CMI+ cons elogCP = logCP - logCPcons elogµRi = logµRi - logµRicons ;

   three variables defined as the integrals in discrete time of the torque supplied (ΣlogCMI⁺), of the projected torque (ΣlogCP) and the richness efficiency (Σlogµ_Ri): ΣlogCMI+ (k+1) = ΣlogCMI+ (k) + elogCMI+ (k) ΣlogCP(k+1) = ΣlogCP(k) + elogCP(k) ΣlogµRi(k-1) = ΣlogµRi(k) + elogµRi(k)

   b) the reference variables are the logarithms of the reference of torque supplied (logCMI⁺ _cons), of the projected torque (logCP_cons), and the efficiencies of ignition advance (logµ_Ricons); and

   c) the control variables (U_K) for the engine actuators are the logarithms (logC_AIRCMD) of the value homogenous to a torque and proportional to the rate of air for controlling the actuator, and the efficiencies of ignition advance (logµ_Av) and richness (logµ_Ri).
5. Method for control of the torque according to one of claims I to 4, characterised in that the engine operates at a single richness, with a richness efficiency equalling 1.
6. Device for controlling the torque of an internal combustion engine, with spark ignition, with injection controlled by an electronic computer, having at least three actuators to which air control, ignition control and mixture richness control are respectively applied, implementing the method according to one of claims 1 to 5, characterised in that the electronic computer (9) comprises:

   means (10) for calculating the air mass (M_AIRMOT) exiting the manifold;

   means (11) for calculating the richness (Ri) and means (12) for calculating the efficiency of the richness (µ_Ri) of the mixture from said air mass (M_AIRMOT) and the injection timing (Ti);

   modules (13) determining from the engine speed, the manifold pressure, the opening of the air actuator, the advance (Av) and the injection timing (Ti), values for the torque supplied (CMI⁺), the projected torque (C_AIRPAP) defined from the air mass entering into the manifold, and the torque (C_AIRMOT) defined from the air mass exiting the manifold, the references for the torque supplied (CMI⁺ _cons) and the projected torque (CP_cons), the references for richness efficiency (µ_Ricons) and the advance efficiency (µ_Avcons);

   means (14) for calculating the logarithm of the values calculated by the means (13);

   an integral quadratic linear regulator (15) receiving at its input the logarithms of the torques calculated by the means (14) and delivering the logarithm of the controls for the advance efficiency (logµ_Av), the air rate of the actuator (logC_AIRCMD) and the richness efficiency (logµ_Ri);

   means (16) for calculating the exponential of the values delivered by the regulator (15);

   means (17) for limiting the values delivered by the means (15) in order to take into account the limitations connected with the operation of the actuators and the engine, delivering the advance efficiency (µ_Av), the air control (C_AIRCMD) and the richness efficiency (µ_Ri);

   means (18) for calculating the value of the ignition advance (Av) from the advance efficiency (µ_Av), the engine speed (N), the manifold pressure (P) and possibly the richness (Ri);

   means (19) for calculating the final control (C_AIRCMD) of the air actuator (5);

   means (20) for calculating the richness (Ri) from the control of the richness efficiency (µ_Ri) and from a table of the torque efficiency according to the richness;

   means (21) for calculating the injection timing (Ti) from said target richness and engine operating conditions such as the pressure in the manifold or the engine speed;

   means (22) for estimation from the injection timing (Ti), the richness (Ri) in each cylinder, and the associated richness efficiency.
7. Device for controlling torque according to claim 6, characterised in that the air actuator is a throttle valve comprising a mechanical throttle (6), the opening (S_p) of which is directly controlled by the accelerator pedal of the vehicle and an additional air valve (7), the opening of which (S_v) is controlled by the electronic computer (9).
8. Device for controlling torque according to claim 6, characterised in that the air actuator is a throttle valve comprising a motorised throttle (8) the opening of which (S_pm) is controlled by the electronic computer (9).

Images