EP1429012A1

EP1429012A1 - Method and system for estimation of air charge of an engine

Info

Publication number: EP1429012A1
Application number: EP02445173A
Authority: EP
Inventors: Sören ERIKSSON; Alexander Stotsky
Original assignee: Ford Global Technologies LLC
Current assignee: Volvo Car Corp
Priority date: 2002-12-09
Filing date: 2002-12-09
Publication date: 2004-06-16

Abstract

The invention relates to a method for estimation of air charge of an engine comprising at least one cylinder (1), an air inlet (3) and an intake manifold (8), said method comprising: measuring the air flow via said air inlet (3), and measuring the pressure in said intake manifold (8). The method further comprises: determining whether the engine is operated at a steady state condition or a transient condition by measuring the pressure in the intake manifold (8), determining a measure of the air charge of said cylinder (1) by measuring the air flow via said air inlet (3) during said steady state condition and by measuring said pressure during said transient condition, correcting said measure of said air charge by means of a correction value () during said transient condition, and adapting said correction value () during said steady state condition. The invention also relates to a system for estimation of air charge of an engine. By means of the invention, an improved system for determining the engine load is obtained, which in particular can be used during sharp transients in the engine load.

Description

TECHNICAL FIELD:

The present invention relates to a method for estimation of air charge of an engine comprising at least one cylinder, an air inlet and an intake manifold, said method comprising measuring the air flow via said air inlet and measuring the pressure in said intake manifold.
The present invention also relates to a system for estimation of air charge of an engine comprising at least one cylinder, an air inlet, an intake manifold, a first sensor for measuring the air flow via said air inlet and a second sensor for measuring the pressure in said intake manifold.

BACKGROUND ART:

In the field of vehicles which are operated by combustion engines, there is a general demand for low emissions of harmful substances in the exhaust gases from the engine. Said substances are primarily constituted by pollutants in the form of nitric oxide compounds (NO_x), hydrocarbon compounds (HC), and carbon monoxide (CO). In regards today's petrol engines, the exhaust gases are normally purified by means of an exhaust catalyst which forms part of the exhaust system. In a so-called three-way catalyst, which is previously known, the major part of the above-mentioned harmful compounds is eliminated by means of known catalytic reactions. In order to optimize the function of the catalyst so that it provides an optimal degree of purification for NO_x, HC, and CO, the engine is in most operating cases operated by a stoichiometric air/fuel mixture, i.e. a mixture where λ=1.
Furthermore, in the field of vehicles, there is a general demand for reducing the fuel consumption of the engine. To this end, during the last few years, engines have been developed having new types of combustion chambers in the engine's cylinders, so as to allow the engine to be operated by increasingly lean fuel mixtures, i.e. where λ>1.
In order to meet the increasing demands for low emissions of harmful substances and a low fuel consumption, the load of the engine must be estimated in a very accurate manner during operation of the engine. For this reason, it is previously known to use a measured value representing the air mass flowing into the intake pipe of the engine when determining the a measure of the engine load.
A problem with previously known engine management systems relates to the fact that it can be difficult to precisely detect and measure extremely quick transients in engine load by means of the conventional method of using a value representing the air mass flow into the engine.
This problem is particularly relevant due to the fact that, during recent years, certain complicated features such as variable valve timing, cam profile switching and variable geometry intake manifold in new engines have been introduced. This leads to requirements as regards more advanced engine load estimation algorithms due to large variations in volumetric efficiency and intake manifold temperature, especially during transients.
The patent document US 5635634 teaches a system in which a measure of the air charge of a combustion engine is dynamically corrected. This is achieved by considering the influence of a variable charge-cycle control such as with an adjustable inlet and/or outlet camshaft, especially during transient operation.
Furthermore, the patent document US 2002/0133286 teaches a method for determining the air charge of a combustion engine. This is achieved by determining the the air flow in the engine air intake, the pressure in the engine intake manifold and also the volumetric efficiency.
Although the above-mentioned known systems are adapted so as to take dynamic variations into account when determining the air charge of an engine, there is still a demand for further improvements in the field of engine management systems for reliable determination of the load of an engine, especially during quick transients in the operation of the engine.

DISCLOSURE OF INVENTION:

An object of the present invention is to provide an improved method and system for estimating the load of an engine by means of an estimation of the air charge into the engine, wherein the above-mentioned problem is solved.
In accordance with the invention, this object is accomplished by means of a method as mentioned initially, which method is characterized in that it comprises: determining whether the engine is operated at a steady state condition or a transient condition by measuring the pressure in the intake manifold, determining a measure of the air charge of said cylinder by measuring the air flow via said air inlet during said steady state condition and by measuring said pressure during said transient condition, correcting said measure of said air charge by means of a correction value during said transient condition, and adapting said correction value during said steady state condition.
Said object is also accomplished by means of a system as mentioned initially, which system is characterized in that it comprises a control unit to which said first sensor and said second sensor are connected, said control unit being arranged for determining whether the engine is operated at a steady state condition or a transient condition via a signal from said second sensor and for determining a measure of the air charge of said cylinder, said control unit also being arranged for correcting said measure of said air charge by means of a correction value during said transient condition, and for adapting said correction value during said steady state condition.

BRIEF DESCRIPTION OF DRAWINGS:

In the following text, the invention will be described in detail with reference to the attached drawings. These drawings are used for illustration only and do not in any way limit the scope of the invention. In the drawings:

Fig. 1: indicates schematically an internal combustion engine in which the present invention can be implemented;
Fig. 2: indicates in a schematical manner the variations in vehicle speed and load of an engine during a normal driving cycle;
Fig. 3: indicates in a schematical manner the principle of the invention as regards switching between a steady state condition and a transient condition; and
Fig. 4: indicates the manner in which a plurality of values representing a correction value is plotted in a diagram as a function of the engine load.

PREFERRED EMBODIMENTS:

In the following, an embodiment of the present invention will be described in detail. With reference to Fig. 1, there is shown a schematic and simplified view of an engine cylinder 1 and a piston 2 operating in a conventional manner. The cylinder 1 forms part of a combustion engine having a number of similar cylinders. An air inlet 3 is arranged so as to guide an incoming flow of air in a direction as indicated by means of an arrow in Fig. 1. The incoming air flow is guided through a air flow sensor 4, a so-called MAF (mass air flow) sensor, which is arranged so as to detect the amount of incoming air flowing through the air inlet 3. The MAF sensor 4 is electrically connected to a computer-based control unit 5, which is arranged so as to control the combustion operation of the engine, as will be described in greater detail below.
Furthermore, the air inlet 3 comprises a throttle valve 6, which is of conventional type and which is provided with a throttle angle sensor 7, which is also electrically connected to the control unit 5 in order to detect a value representing the angle of the throttle valve 6. The incoming air flow is guided via the throttle valve 6 and further to an intake manifold 8, in which a pressure sensor 9, a so-called MAP (manifold air pressure) sensor, is arranged. The MAP sensor 9 is also electrically connected to the control unit 5 in order to detect a value representing the existing pressure in the intake manifold 8.
The intake manifold 8 is connected to the above-mentioned cylinder 1 via a cylinder inlet 10. The intake manifold 8 is also connected to a number of additional cylinders (not shown) in the engine, via a corresponding number of further cylinder inlets which are shown schematically by means of reference numeral 11 in Fig. 1. The cylinder inlet 10 is provided with a fuel injector 12 which is adapted to inject a calculated amount of fuel towards the cylinder at a given moment. In order to control the operation of the fuel injector 12, it is also connected to the control unit 5. Furthermore, a spark plug 13 is arranged in the cylinder 1 so as to ignite a mixture of air and fuel being injected into the cylinder 1. To this end, the spark plug 13 is electrically connected to the control unit 5, which in turn is arranged so as to activate the spark plug 13 at a particular moment as determined by means of the control unit 5. The cylinder 1 is also connected to an exhaust outlet 14, which is arranged so as to guide the exhaust gases resulting from combustion in the cylinder 1 to the outside atmosphere, suitably via an exhaust catalyst (not shown), in a direction as shown by means of an arrow in Fig. 1. It should be noted that Fig. 1 is a simplified figure of an engine system and several components, for example the engine's intake and exhaust valves, are not shown in this figure.
The general operation of a combustion engine, including the operation of a fuel injector and a spark plug in order to ignite an air/fuel mixture, is previously known per se and for this reason it is not described in detail here.
In order to optimize the fuel consumption and to minimize the emissions of harmful compounds from the engine, the control unit 5 is provided with engine management routines which are adapted to provide a precise control of the combustion process in the cylinder 1. In particular, the signals from the three sensors, i.e. the MAF sensor 4, the throttle angle sensor 7 and the MAP sensor 9, are fed to the control unit 5. In this manner, a suitable control as regards the amount of fuel (m_fuel) to be injected by the fuel injector 12 and the timing (α_spark) for activation of the spark plug 13 for each combustion of an air/fuel mixture can be carried out by means of the control unit 5.
Other parameters and measured values can also be used for the control of the combustion process, for example the intake air temperature, the ambient air temperature, the feedback lambda value, a knock signal and the water temperature. For this reason, corresponding sensors and measurement devices can be added to the the system as shown in Fig. 1 and connected to the control unit 5. The use of various parameters for control of the fuel injection and the ignition timing, and other aspects of the control of a combustion engine, is previously known per se. For this reason, such devices are not described in detail here.
In order to control the combustion process, the present invention is based on the insight that the engine load, i.e. the mass of air inducted to each individual cylinder of the engine, must be determined in a precise manner. To this end, the control unit 5 is provided with routines and algorithms adapted for calculating a value of the air flow into the engine based on measured values from the MAF sensor 4 and the MAP sensor 9. In this regard, it should be noted that the signal from the MAF sensor 4 represents a value of the incoming air flow via the throttle 6 which is generally equal to the actual incoming air flow to the cylinder 1 (and to the remaining cylinders of the engine) during steady state operation of the engine, i.e. during situations in which there are essentially no variations of the load of the engine. It should also be noted that a value of the pressure p in the intake manifold 8, as provided by the MAP sensor 9, can be used to detect whether such a steady state condition exists. In particular, this is accomplished by calculating the derivative of the pressure p in the intake manifold, i.e. dp/dt. In the event that the absolute value of this derivative is approximately zero, it can be assumed that the engine is operated at a steady state condition. On the other hand, in the event that the absolute value of the derivative is more than approximately zero, or rather, more than a particular limit value which is close to zero, it can be assumed that the engine is operated at a transient condition.
In summary, the MAP sensor 9 can be used to detect whether the engine is operated in a steady state condition or if it is operated in a transient condition.
The present invention relies on the principle that a precise estimation of the incoming air flow to the cylinder 1 can be carried out not only during steady state conditions but also during extremely quick transients. As a result, a correct engine load estimation and precise control of the combustion engine is allowed, which in turn results in optimized fuel consumption, emission control and driveability. For this reason, the invention is based on the fact that the MAP sensor 9 can be used to provide a particular correction value . As will be described in detail below, the correction value  is adapted in a particular manner during steady state conditions. Also, the correction value  is used when a transient condition is detected in order to correct or adjust a measured value from the MAF sensor 4 representing the incoming air flow.
It has been discovered that the MAF sensor 4 can be used as such during steady state conditions for providing a generally true value of the air charge of the cylinder 1. However, during transient conditions, the MAF sensor 4 alone will not provide a true representation of the air charge, due to a relatively slow response and the difficulty in performing a precise filtering correcting for a low pass filter effect of the intake manifold. For this reason, the above-mentioned correction value  is used according to the invention during said transient conditions. Since the correction value  is not required for corrections during steady state conditions, said correction value  is adapted according to a particular algorithm during said steady state conditions. This will be described in detail below.
With reference to Fig. 2, a normal driving cycle of a vehicle in which the present invention can be used will now be described in greater detail. The upper graph in Fig. 2 indicates how the the speed of the vehicle varies with time during such a normal driving cycle. The lower graph in Fig. 2 corresponds to the same driving cycle but indicates how the load of the vehicle's engine varies with time. Vertically extending broken lines are used to indicate the manner in which the driving cycle progresses through several different phases or stages during the driving cycle. In particular, an initial phase of the driving cycle is indicated by means of reference numeral 15 and corresponds to a phase in which the vehicle speed is relatively low and is also generally constant. This means that the load of the engine is generally also low and constant during the start phase 15. Consequently, the vehicle is in a "steady state" condition during the start phase 15, in which there are generally no changes of the engine load. Furthermore, the driving cycle proceeds into a second phase 16 in which the vehicle speed increases in a generally linear manner. This means that the load of the engine, as shown in the lower graph of Fig. 2, will change quickly, i.e. the vehicle assumes a "transient" condition in which the engine load is rapidly increasing from the low value during the start phase 15 to a relatively high value which remains constant (i.e. again assuming a steady state condition) as long as the vehicle speed increases linearly during the second phase 16. Finally, the increase of the vehicle speed will cease, resulting in a constant, relatively high vehicle speed (i.e. higher than the initial vehicle speed during the start phase 15) during a third phase 17. When the increase of the vehicle speed ends and instead the vehicle speed becomes generally constant during the third phase 17, there will be a sharp decrease in engine load, i.e. a further transient condition, as shown in the lower graph of Fig. 2. This decrease in engine load will proceed into a relatively low and generally constant value of the engine load.
It should be noted that Fig. 2 describes a typical driving cycle of a conventional vehicle. It should be obvious that many different types of driving cycles may occur during operation of a normal vehicle.The type of driving cycle shown in Fig. 2 is only used as one possible example in order to describe the manner in which the present invention can be used, and the invention is obviously not limited to use in any particular driving cycle or with any particular vehicle or engine.
With reference to the lower graph of Fig. 2 it can furthermore be noted that the steady state conditions of the engine load during the driving cycle are indicated by means of broken lines, whereas the transient conditions are indicated by means of full lines. The transient conditions occur when the vehicle speed starts to increase, at the beginning of the second phase 16, and also when the vehicle speed changes from a linearly increasing value (during the second phase 16) to a generally constant value, i.e. at the beginning of the third phase 17.
As previously mentioned, an important principle on which the invention is based relates to the fact that the transient and steady state conditions of the engine load are detected during the driving cycle. More precisely, since changes in the engine load correspond to changes in the pressure of the intake manifold, the MAP sensor 9 (see Fig. 1) is used for detecting the pressure in the intake manifold 8. If the absolute value of the derivative of the pressure is higher than a particular limit value L, i.e. if | dp/dt | > L
it can be assumed that the engine load has entered a transient condition. Preferably, the limit value L of the pressure in the intake manifold should theoretically be zero. However, due to the fact that a certain amount of noise and small natural variations of the intake manifold pressure p occur, it is suitable to choose a limit value L which is slightly higher than zero, so that the true passages from a steady state to a transient condition, and vice versa, can be detected in a reliable manner. The actual value of the limit value L is determined depending on the quality of the signal from the MAP sensor 9 and the desired speed for detecting transient conditions.
The manner in which the engine in question switches between a steady state condition and a transient condition is indicated in Fig. 3, which is a state diagram indicating the two conditions, i.e. a steady state condition 18 and a transient condition 19. The steady state condition 18 indicated in Fig. 3 corresponds to the sections of the lower curve in Fig. 2 which are indicated by means of a broken line. The transient condition 19 according to Fig. 3 corresponds to the sections of the lower curve in Fig. 2 which are indicated by means of a a full line (i.e. at the beginning of the second phase 16 and the third phase 17, respectively).
As shown in Fig. 3, the transition from a steady state condition 18 to a transient condition 19 occurs when the absolute value of the derivative of the pressure p in the intake manifold is higher than a limit value L. Also, the transition from a transition condition 19 to a steady state condition 18 occurs when the absolute value of the derivative of the pressure p is lower than said limit value L.
According to the invention, a correction value  is derived from the measurements from the MAP sensor 9 and is used during the transient condition 19 for determining a corrected value of the engine load, i.e. a value representing a measure of the air charge during operation of the engine.
Furthermore, the invention is based on the fact that said said correction value  is not used during the steady state condition 18. Instead, the correction value  is adapted during said steady state condition 18. By adapting the correction value  continuously, i.e. each time the steady state condition 18 occurs during a driving cycle of a vehicle, it can be used for a very precise correction of the measurements each time the transient condition 19 occurs during a driving cycle.
Fig. 4 is a diagram showing the correction value  as a function of the engine load. More precisely, the diagram shows a plurality of plotted points 20, each representing an adapted value of the correction value  as determined at a particular engine load. This means that for each time the steady state condition 18 occurs, a new correction value  will be generated and plotted in the diagram in Fig. 4. Obviously, the diagram in Fig. 4 is used as a means for explaining the principles of the invention, and the invention is suitably implemented by storing the adapted correction values  in a lookup table in the control unit 5 (Fig. 1).
During a driving cycle involving a large number of occasions in which the steady state condition occurs, a corresponding large number of correction values  will be stored, one for each occasion in which the steady state condition is at hand. For example, assuming that the engine is operated in a steady state condition at a particular load which is indicated by means of the line 21 in Fig. 4, a particular value of the correction value  will be calculated. This particular value is indicated by means of reference numeral 22 in Fig. 4. Furthermore, the value of this particular correction value  is stored in a table in the control unit 5.
As explained above, a very large number of points will be gathered over a particular time period, or even during the entire life cycle of the engine in question, each time the steady state occurs. All these points which are gathered can be said to define an approximated relationship between the correction value  and the load. This relationship is indicated by means of a curve 23 in Fig. 4. As will be described in detail below, the curve 23 can be approximated by means of a suitable algorithm, for example the so-called method of least squares. In this manner, the relationship between the correction value  and the engine load is approximated as a linear curve or a polynomial. This relationship can be stored as a mathematical relationship in the control unit 5. As new values of the correction value  are continuously stored during operation of the engine, the relationship between the correction value  and the load (i.e. the curve 23) will be gradually and continuously adapted, i.e. updated.
As mentioned above, the continuous gathering of new values of the correction value  at given engine loads is carried out only during the steady state condition of the engine. Also, the result of this gathering of the correction value will be used for correcting the value of the load based on pressure measurement only when the transient condition occurs. In practice, this means that during control of the engine by means of the control unit 5, any occurring transient condition will be detected by means of the MAP sensor 9 as explained above. When the transient condition occurs at a given engine load, the correction value  will be used when determining the engine load, i.e. a value representing the signal from the MAP sensor 9 (and various other parameters, as will be described below) is corrected by means of the correction value , wherein the relationship between the load and the correction value  is given by means of the value of the curve 23. In this manner, the invention can be used for providing a highly accurate estimation of the engine load also during transient conditions.
In the following, a summary will be made of formulas used in an engine load estimation. The main load calculation is made with the following speed density equation: me = η(p,ne ) ne 2 Vd p RT(t) where n_e is the engine speed, η is the volumetric efficiency, V_d is the total displaced cylinder volume, T is the intake air temperature and  is the adjustable parameter, i.e. the correction value. The value m_e is the mean value of the flow into the engine cylinders, i.e. the engine load.
The pressure in the intake manifold is modelled as follows:
where V_im is the throttled volume and m_th is the flow via the throttle 6 (see Fig. 1).
The adaptation is performed with the following adjustment law:
where e = p - p and is a tracking error, σ is a non-negative calibrateable function of the derivative of the intake manifold pressure, g is a positive design parameter,  is a so-called feedforward part and ε and  are secondary variables defined by ε = -η ne 2 Vd Vim ε + α0(p - p - ε)- ϕ and ϕ = -(η ne 2 Vd Vim + α0)ϕ - RT Vim mth
For details, reference is made to the paper "Composite Adaptive Engine Load Estimation", Proc. of the 10th Mediterranean Conference on Control and Automation - MED 2002, Lisbon, Portugal, July 9-12, 2002, Stotsky A. and Eriksson S.
In the following, the invention will be described in greater detail with reference to specific algorithms which suitably can be used in order to implement the invention.
In order to control the operation of the engine, its load must be estimated. According to the preferred embodiment, the invention relates to an adaptive learning observation technique which is used in order to estimate the air flow into the engine. As will be described, the invention relies on the use of a speed-density calculation during the transient conditions (for determining the air charge of the engine), whereas measurements representing the air flow via the throttle 6 (Fig. 1), as detected by the MAF sensor 4, are used without corrections during steady state conditions.
The pressure in the intake manifold is described by the ideal gas law: p = RT* M Vim where R is the gas constant, T is the temperature of air in the intake manifold, V_im is the intake manifold volume, * is an unknown constant parameter which has to be estimated, and M is the mass in the intake manifold. The term p can be determined by means of the MAP sensor 9 (see Fig. 1).
Assuming that the intake manifold temperature is constant and differentiating (1), we get p = RT* Vim (mth - me ) where m_e is the mean value of the flow into the engine cylinders and m_th is the flow via the throttle 6 (see Fig. 1) as measured by means of the MAF sensor 4.
Furthermore, the flow m_e into the engine can be calculated on the basis of the well-known speed-density model: me = η(p, ne ) ne 2 Vd p RT* where n_e is the engine speed, η is the volumetric efficiency and V_d is the total displaced cylinder volume. It can be noted that (3) is not realizable due to the uncertain parameter *, which could be associated with an uncertain ratio η(p, n_e)/T*.
According to the preferred embodiment of the invention, an estimation algorithm (t) is chosen such that (t) → * as t → ∞, using the pressure measurement as provided by the MAP sensor 9 and the air flow measurement as provided by the MAF sensor 4.
It can be assumed that the measured intake manifold pressure satisfies the following equation, which comes from (2) and (3): p = -η(p, ne ) ne 2 Vd Vim p + RT Vim *mth Consider the following adaptive observer:
where  = (t) is an adjustable parameter. It can be noted that (5) has the same structure as (4) and uses the measured pressure as an input to volumetric efficiency.
A tracking error can be defined as follows: e(t) = p(t) - p (t)
The adaptive control aim according to the invention is to drive the tracking error to zero. In other words, the correction value  should be adapted in a manner so that the difference between p(t) and (p hatt)(t) is minimal. Using (4) and (5) we get the following error model:
where
Define the prediction error (i.e. the error between the unknown parameter and its estimate)
and prediction error estimate as follows ees = e - ε where ε and ϕ are adjusted as follows ε = -η(p, ne ) ne 2 Vd Vim ε + α0 (e - ε) - ϕ ϕ = -(η(p, ne ) ne 2 Vd Vim + α0) ϕ - RT Vim mth where α₀ is a positive design parameter and  ˙ is defined below.
To show the convergence of the prediction error (8) to its estimate (9) we consider the following error:
Straightforward calculations show that e₁ satisfies the following equation: e 1 = -(η(p, ne ) ne 2 Vd Vim + α0)e 1 and
Consider the following adaptation law, which was presented in the paper "Composite Adaptive Engine Load Estimation", Proc. of the 10th Mediterranean Conference on Control and Automation - MED 2002, Lisbon, Portugal, July 9-12, 2002, Stotsky A. and Eriksson S. :  = -γ(α0ϕ(e - ε) - RT Vim mthe + σ( - )) where  is a priori value of * with adjustable σ modification σ = β(e - ε)2 where β>0.
In the following, the case in which the a priori value is constant, i.e.  = 1, will first be considered. After that, the case in which an adapted feedforward part is introduced in the algorithm instead of a single a priori value will be considered.
The adjustment law  (t) represents the composite adaptive law driven by the tracking error e, the prediction error estimate (e - ε) and the adjustable σ modification.
For implementation, it is preferable to use a calibratable table with prediction error estimate (e - ε) as an input instead of (16). Such a table should have a dead-zone which determines the steady-state condition, a condition where a compromise between the speed of adaptation and the signal quality can be achieved without σ modification. Such a dead-zone may generally be used as the above-mentioned limit value when determining whether the transient or steady state condition occurs. The use of such a table allows also the separation between positive and negative transients.
According to the invention, the adaptation of the correction value  (equations (29)-(33) below) converges to a stable value. A theorem related to the convergence properties of the adaptation will now be presented. Consider the system (7), (11), (10), (15). If the system parameter α₀ satisfies the following constraints: α0 > 4β(C + 2 V(0)) where
and
Then

and all the signals of the system remain bounded.
Proof: Consider the following Lyaponov function candidate
Evaluating its derivative along the solutions of the system we have
By decomposing the second term we get
and by taking into account (17) we finally get
Now we conclude that e and ((e - ε) - ϕ
) are bounded and squareintegrable, is bounded, and ϕ and (e - ε) are squareintegrable. From (7) we conclude that e ˙ is also bounded, and (19) is achieved. Using (10) one can show that ε ˙ and ε are also bounded. Then (20) and hence (21) are achieved. The convergence of the parameter (the control aim (22)) can be proved using the following equation
which represents a stable time varying lowpass filter with squareintegrable and convergent input and by noting that there exists a positive constant C such that ϕ² ≥ C.
Then
and the speed-density flow (3) can be realized as follows me = η(p, ne ) ne 2 Vd p RT(t) where the parameter (t) is adjusted according to (15), (11), (10).
It can be noticed that m_e = m_th under steady-state conditions.
In summary, all the equations for engine load estimation are listed in the following: me = η(p, ne ) ne 2 Vd p RT(t) p = -η ne 2 Vd Vim p + RT Vim mth ε = -η ne 2 Vd Vim ε+ α0(p - p - ε) - ϕ ϕ = -(η ne 2 Vd Vim + α0) ϕ - RT Vim mth
As shown in equation (29), the value p of the intake manifold pressure as provided by the MAP sensor 9 (Fig. 1) is used when determining the air charge. The correction value  is also used in this regard. The value m_th as provided by the MAF sensor 4 is furthermore used when determining the correction value, see equation (33).
Simulation results of the system (29)-(33) with  = 1 using measurements from a real vehicle show that the adjustable parameter  converges to different values depending on the engine working point (intake manifold pressure) and the feedforward part of the algorithm  should preferably be improved. To this end, this feedforward part of the algorithm  is presented as a static map whose input is the intake manifold pressure, i.e.,  = f(p) with breakpoints p_mp1,  _mp1 and p_mp2,  _mp2. Breakpoints of this map  ₁ and  ₂ which correspond to the fixed points p_mp1 and p_mp2 are updated. Assuming that the feedforward part is a linear function of the intake manifold pressure,  = k 1 p + k 2 where k₁ and k₂ are unknown parameters to be defined.
As mentioned above with reference to Fig. 4, the correction value  can also be defined as a polynomial. In the following, however, it is assumed to be a linear function.
We divide the whole intake manifold pressure region into two well separated parts, i.e., low and high pressures: 0 < p_low < p_b and p_b < p_high < p_amb, where p_amb is the ambient pressure, p_b is a boundary value of the intake manifold pressure, and p_b = 550hPa.
At steady-state we memorize the values  _low and  _high which correspond to < p_low and p_high. Then the coefficients for the linear model can be found as solutions of the following equations  high = k 1(t)phigh + k 2(t)  low = k 1(t)plow + k 2(t) and the breakpoints of the static map are adapted under the steady-state condition as follows  mp 1 = k 1(t)pmp 1 + k 2(t)  mp 2 = k 1(t)pmp 2 + k 2(t) where k₁ and k₂ are computed from (35) and (36).
For later convenience we list all the equations for engine load estimation with an improved feedforward part. The estimator is described by the equations (29)-(33) with feedforward part  = f(p) where the breakpoints  _mp1 and  _mp1 in the static map  = f(p) are adapted according to (37) and (38) with the following coefficients k 1(t) =  high -  low phigh - plow k 2(t) =  low - k 1(t)plow where  _high and  _low are adapted for low pressure:

and for high pressure:

where Δ, γ_ and γ_p are positive design parameters.
Notice that the values of high and low pressures should be well separated, i.e., the adaptation is allowed if and only if there exists a positive constant C such that (p_high - p_low) > C > 0.
The main improvement of the algorithms (29)-(33) is the adjustable feedforward part  = f(p). This feedforward part is assumed to be a linear function of intake manifold pressure (34). To validate this model, two equations for low and high pressures are needed. The learning process can be described as follows. During steady-state conditions if the intake manifold pressure is low (p < p_b),  _low is adapted according to (42) and  _low converges to the adaptive parameter , holding  _high constant.
Notice that  does not depend on  under steady-state conditions since the term σ( - ) disappears from equation (33) (since σ(t) = 0 under steady-state conditions) and the adaptation law is driven by tracking and prediction errors only.
The coefficients of the model (34) are computed from (40) and (41) and the breakpoints of the static map  = f(p) are adapted according to (37) and (38). Suppose that a positive transient comes and the intake manifold pressure grows up to the high level. Then, at steady-state,  _high is adapted according to (44), thereby  _low is frozen. Again, the coefficients of the model (34) are computed from (40) and (41) and breakpoints of the static map  = f(p) are adapted according to (37) and (38). Now the mapping  = f(p) is based on two recently learned values  _low and  _high. Notice that the feedforward part can also be updated in the least-squares sense. Then the next transient uses the recently updated feedforward part  = f(p), whose values are close to the true values. The improvement of the feedforward part of the algorithm speeds up the convergence of the estimated parameters to their true values.
In summary, the present invention relates to an improved method and system for estimating the engine load in a precise manner, in particular during transient conditions. The engine load, i.e. the value m_e, is determined by means of a number of signals, e.g. a measured pressure value p from the MAP sensor 9 and by means of the correction value , see equation (29) during steady state and transient conditions. During steady state conditions, the value m_e will be equal to the measured value m_th, due to the adaptation of the correction value  during steady state conditions. Furthermore, the correction value  is adapted by means of equations (30)-(33).
The invention provides a theoretically justified, robust adaptive estimation scheme which is driven by both tracking and prediction errors with adjustable σ-modification and adaptive feedforward part for engine load, which allows the use of the speed-density flow under transient conditions and measured flow via the throttle under steady-state. The result allows a fast and robust engine load estimation.
The invention is not limited to the embodiment described, but can be modified within the scope of the appended claims.

Claims

Method for estimation of air charge of an engine comprising at least one cylinder (1), an air inlet (3) and an intake manifold (8), said method comprising:

measuring the air flow via said air inlet (3), and

measuring the pressure in said intake manifold (8),

characterized in that said method comprises:

determining whether the engine is operated at a steady state condition or a transient condition by measuring the pressure in the intake manifold (8),

determining a measure of the air charge of said cylinder (1) by measuring the air flow via said air inlet (3) during said steady state condition and by measuring said pressure during said transient condition,

correcting said measure of said air charge by means of a correction value () during said transient condition, and

adapting said correction value () during said steady state condition.
Method according to claim 1, characterized in that said correction value () is determined by means of said measuring of the pressure in the intake manifold (8).
Method according to claim 1 or 2, characterized in that the step of determining whether the engine is operated at a steady state or transient condition is carried out by comparing the absolute value of the derivative of the pressure in the intake manifold (8) with a predetermined limit value (L).
Method according to any of the preceding claims, characterized in that said step of correcting said measure of the air charge is carried out only during said transient condition, and that said step of adapting said correction value () is carried out only during said steady state condition.
Method according to any of the preceding claims, characterized in that said step of adapting said correction value () comprises storing a value representing said correction value () as a function of the load of said engine when said steady state condition occurs.
Method according to claim 5, characterized in that a plurality of stored values are used for determining a relationship between said correction value ().
System for estimation of air charge of an engine comprising at least one cylinder (1), an air inlet (3), an intake manifold (8), a first sensor (4) for measuring the air flow via said air inlet (3) and a second sensor (9) for measuring the pressure in said intake manifold (8), characterized in that said system comprises a control unit (5) to which said first sensor (4) and said second sensor (9) are connected, said control unit (5) being arranged for determining whether the engine is operated at a steady state condition or a transient condition via a signal from said second sensor (9) and for determining a measure of the air charge of said cylinder (1), said control unit (5) also being arranged for correcting said measure of said air charge by means of a correction value () during said transient condition, and for adapting said correction value () during said steady state condition.