EP0175596B1 - Adaptive method for regulating the injection in an injection motor - Google Patents

Description

The invention relates to an adaptive method of regulating the injection of an injection engine, according to which the injection time (Ti) is permanently determined by conventional regulation as a function of the intake pressure (P) or of the air flow at the intake from a control line (D) defined by its slope (f) and its ordinate at the origin (b), apart from other possible corrections (c, c ' ...), and we periodically readjust, by successive adaptation cycles according to a determined period, the values of the slope (f) and the ordinate at the origin (b) of the regulation line as a function differences in possible wealth observed by an exhaust gas analysis probe, the pressure space being divided into a certain number (n) of zones to each of which corresponds a central value (P _j ) of the pressure.

It is known to permanently determine the injection time by conventional regulation as a function of the pressure in the intake manifold, or also of the air flow at the intake, from a regulation line defined by its slope and its ordinate at the origin in the diagram of the injection times as a function of the pressures. These values are calculated as accurately as possible during the engine development period; but it is known that they vary randomly over time as a function of various parameters, for example the clogging of the air filter which reduces the air flow for the same pressure. It is therefore necessary for precise regulation to periodically readjust the parameters of this regulation line.

For this, there is a process, called "American process", described in particular in the article "A Closed-Loop A / F Control Model for Internai Combustion Engines" by Douglas R. HAMBOURG and Michael A. SHULMAN, published in 1980 by the "Society of Automotive Engineers, Inc." This process consists in using a probe known as "Lambda probe of analysis of exhaust gases which gives a signal which varies when fon has a lack of oxygen in the exhaust gases, which testifies to a richness crossing the value 1 corresponding to the stoichiometric mixture. When this occurs, the parameters on the control line are corrected as follows: if the inlet pressure P is less than a determined threshold, a correction is applied only to the ordinate at the origin of the line , while if the pressure value is greater than this threshold, a correction is applied only to the slope of the line. This process is therefore approximate. Indeed, if the current conditions are maintained, the recalculated line always ends up passing through the current operating point, but a local anomaly can distort the calculation of all the other points. In addition, this process only works with unit richness, while the problems of energy saving and pollution lead more and more to use wealth less than unity.

We also know the process called "overinjection", described in French patent application 8317538 in the name of the applicant, and which consists, when the injection is regulated at a richness less than unity, to periodically increase progressive richness until the gas analysis probe is triggered, then return to the initial richness while retaining the value of the relative increase in injection time which was thus necessary , which, compared to the theoretical increase resulting from the desired wealth, gives the necessary correction. The aforementioned patent application indicates how the jolts resulting from this momentary foray into wealth can be avoided by action on the ignition advance. However, these over-injections must be sufficiently spaced in time, with a period for example of 10 minutes.

One could naturally depending on this state of the art consider using the American process, even with a richness less than 1, by combining it with the over-injection process. However, in this case, the lack of precision of this American process would be further increased by the significant increase in the readjustment period due to the over-injection process.

The object of the invention is to eliminate the above drawbacks by carrying out an adaptive method of regulating injection, the adaptation of which is more precise in all its points and less sensitive to local anomalies, while accommodating n ' no matter what periodicity.

• - en fonction des indications de la sonde et de la richesse désirée, on détermine le facteur correctif (1 + a) par rapport à la droite de régulation actuelle (D) ; en confondant la valeur mesurée (P) de la pression d'admission avec la valeur la plus voisine (P_j) du centre de la zone dans laquelle se trouve (P).
• - on calcule et on attribue individuellement pour chaque zone un facteur correctif (β_j), par rapport à une droite initiale (D_o, f_o, b_o) définie en mémoire morte, par un calcul purement linéaire en fonction du facteur correctif (1 + α) défini précédemment, et en fonction des paramètres (f, b) de la droite actuelle, tel que

• - en fonction des diverses valeurs (β_i) en mémoire vive de ce dernier paramètre correctif, on calcule et on attribue de nouvelles valeurs des paramètres (f, b) de la droite de régulation (D) par des formules purement linéaires telles que :
  
  
  b_o, k_; et k'; étant des coefficients de pondération constants en mémoire morte,
• - les coefficients de pondération (k_i, k'_;), utilisés pour le calcul des paramètres (f, b) de la droite de régulation au cours de chaque cycle d'adaptation, étant déterminés de manière que cette droite passe le plus près possible de tous les points théoriques (M) de fonctionnement définis à chaque cycle d'adaptation.

The invention is characterized in that at each adaptation cycle of a fixed period, after having determined a theoretical point (M) of engine operation, of abscissa (P) corresponding to the current intake pressure measured, as well as the corresponding zone number (j), the following operations are carried out successively:

• - according to the indications of the probe and the desired richness, the corrective factor (1 + a) is determined relative to the current regulation line (D); by confusing the measured value (P) of the intake pressure with the closest value (P _j ) from the center of the zone in which is located (P).
• - a corrective factor (β _j ), relative to an initial straight line (D _o , f _o , b _o ) defined in read-only memory, is calculated and individually assigned for each zone, by a purely linear calculation as a function of the corrective factor ( 1 + α) defined previously, and according to the parameters (f, b) of the current line, such that

• - as a function of the various values (β _i ) in random access memory of this last corrective parameter, new values of the parameters (f, b) of the control line (D) are calculated and assigned by purely linear formulas such as:
  
  
  b _o , k _; and k '; being constant weighting coefficients in read-only memory,
• - the weighting coefficients (k _i , k '_; ), used for the calculation of the parameters (f, b) of the control line during each adaptation cycle, being determined so that this line passes closest possible for all theoretical operating points (M) defined in each adaptation cycle.

Other particularities will appear in the description which follows of an embodiment taken as an example and represented in the appended drawing, in which:

FIG. 1 is a diagram of the control lines in the pressure / time space, and

FIG. 2 represents the flow diagram of the process.

c, c'... étant des corrections en fonction de paramètres mesurés, tels que température de l'eau, température de l'air, etc. et f étant le coefficient d'échelle. Les dénominateurs 256 sont des valeurs arbitraires correspondant de préférence à la capacité de stockage d'un octet pour que les faibles valeurs de correction soient ramenées à des valeurs entières. Les valeurs f et b peuvent être considérées comme représentant respectivement la pente et l'ordonnée à l'origine de la droite de régulation, compte non tenu des autres corrections.As long as the engine is running, and at a higher speed than that of the idle, the injection time T is permanently determined _; by a conventional regulation as a function of the intake pressure P, or even in certain cases, of the air flow at the intake measured by a flow meter, and from a regulation line which can be expressed by the equation:

c, c '... being corrections according to measured parameters, such as water temperature, air temperature, etc. and f being the scale coefficient. The denominators 256 are arbitrary values preferably corresponding to the storage capacity of a byte so that the low correction values are reduced to integer values. The values f and b can be considered as representing respectively the slope and the ordinate at the origin of the control line, without taking into account the other corrections.

In addition, and this is why the regulation process is adaptive, the values of f and b are readjusted periodically as a function of the differences in richness observed by an exhaust gas analysis probe. It can be a Lambda zirconium oxide probe sensitive to an excess of oxygen, or any other probe or analysis method.

If the engine operates with a unit richness, that is to say in stoichiometric mixture, according to the most frequent standards in use in the United States, the signal from the probe immediately indicates whether the richness should be increased or reduced , i.e. the injection time.

If on the contrary, the engine operates with a constant or variable richness depending on the circumstances but less than unity, for example 0.8, as it is. more and more frequently the use according to European standards in order to reduce consumption and pollution, the value of the correction is a little more complex to establish. We can in particular use the so-called overinjection process, described in the aforementioned French patent, which consists, with each adaptation cycle, of gradually increasing the injection time until the analysis probe switches, then to quickly return to the previous wealth. If for example the richness is fixed at 0.8, it suffices, from the current injection time, to increase it by 25% to theoretically obtain this changeover. If this changeover takes place sooner or later, a simple rule of three gives the value of the correction to be made.

On the diagram of FIG. 1, the theoretical operating point M is thus determined from point A with the same abscissa on the current operating line D by a corrective term a such that the ordinate of N is equal to the ordinate of A multiplied by (1 + α). The known method of over-injection also includes measures to prevent the incursion into higher wealth, although brief, from introducing a jerk into the operation of the vehicle, and this by temporarily altering the ignition advance of proportionately.

Each adaptation cycle therefore determines a theoretical operating point M at the abscissa P corresponding to the current inlet pressure. If this point M is on the regulation line D, of course there is no correction to be made. If, on the contrary, the point is outside the line, it may be necessary to correct it.

For this, according to the state of the art, and in particular according to the American method indicated above, a medium pressure threshold is determined, and if the current pressure P is less than this threshold, only the ordinate is corrected at l origin b of the line D without modifying the slope f of this line so that it progressively passes through the theoretical point M, whereas on the contrary, if this pressure is greater than the threshold, only the slope f is corrected without modifying the ordinate at the origin b so that this straight line progressively passes through the new point M. This process is therefore simple but not very precise and is very sensitive to possible local anomalies.

On the contrary, according to the invention, the space of pressures P is divided into a certain number n of zones, for example four in the example of FIG. 1, and for each zone of rank j the average pressure P _j is defined corresponding to the abscissa of the center of the zone.

During the development of the engine, the ideal initial regulation line D _{o is determined} , the parameters f _o and b _{o of which} are loaded into read-only memories. On the contrary, the parameters f and b of the current regulation line D are loaded into random access memories and contain the values resulting from the previous use. Failing this, that is to say in the event of erasing the RAMs, they are loaded with the values f _o and b _o .

The adaptation cycles follow one another, which can be relatively short (a fraction of a second) if unit richness is used, and which have an interest in being more spaced apart, for example 10 minutes, if one uses a richness lower than the unit and the method of overinjection for the reason indicated above.

At each new adaptation cycle, illustrated by the flow diagram of FIG. 2, the current intake pressure P is measured and the number j of the zone in which this pressure P is found is determined. For this, we usually operates digitally and just perform a full division or rounding.

Having determined j, we proceed to the analysis of the signal from the probe and to the calculation of the corrective term 1 + α with respect to the current regulation line D. This implies in particular, in the case of using a lower richness individually, the application of the over-injection process as a whole. It is indeed from point A that the wealth incursion is carried out up to point M and return to point A, the ratio of the ordinates of M and A, compared to the fixed wealth, making it possible to determine directly 1 + a. These calculations are carried out by confusing the value P of the pressure with the closest value, for example P2 in the example of FIG. 1 if j = 2.

There are also n random access memories containing various values of β _{j, j} varying from 1 to n, the coefficients β being defined as the coefficients a but from the initial regulation line D _o . In other words, we pass from point B on this straight line to point M by multiplying the ordinates by the factor 1 + β.

valeur qui résulte d'une expression purement linéaire en fonction de α, puisque 1/f_o et 1/f_oPj sont des constantes, ainsi que b_o, tandis que f et b sont les valeurs actuelles en mémoires vives des paramètres de la droite de régulation D. Ce calcul purement linéaire est donc facile et rapide. Bien entendu, il n'affecte que le β_j, tandis que les autres β_i, pour i différent de j demeurent à leur ancienne valeur.For the value of j calculated at the start of the cycle, the value indicated in FIG. 2 is calculated and assigned to the memory β _j ,

value which results from a purely linear expression as a function of α, since 1 / f _o and 1 / f _o Pj are constants, as well as b _o , while f and b are the current values in RAM of the parameters of the regulation line D. This purely linear calculation is therefore quick and easy. Of course, it affects only the β _j, while the other β _i , for i different from j remain at their former value.

Continuing the adaptation cycle, we then calculate and assign to the memories f and b also purely linear values expressing themselves as a function of β _i, for all the values of i from 1 to n, with weighting coefficients k _j and k ' _i :

These 2n constants k _j and k 'are naturally contained in read-only memories and are determined, experimentally or by calculation, in such a way that the new line D thus determined passes as close as possible to all the points such as M previously calculated.

The regulation of the injection time continues with the new values of the parameters f and b of the regulation line, while independently the adaptation cycle continues with a waiting loop for the fixed period before starting again at the beginning of the cycle.

During the operation of the engine, the intake pressure P naturally varies and passes more or less frequently through all the values of the space provided, which makes it possible to update successively. periodically the various points corresponding to the various zones. But it is clear that each adaptation cycle takes into account not only the operating point M of the zone j considered, but also all of the other points calculated previously, that is to say the history which precedes. In particular, each new straight line D does not generally pass through all the points but consequently attenuates the influence of any local anomalies.

The calculator uses only a few variables: P, j, a, f, b, β _i (n values) and few constants: 1 / f _o , 1 / f _o P _j , b _o , k _i (n values), k '(n values), richness, periodicity. In addition, the calculations are extremely simple, since they are all linear and with a small number of terms, and nevertheless precise enough to ensure rapid convergence possibly accommodating a high cycle period.

Naturally, the process applies equally to the stoichiometric mixture or to the different riches of the unit, even variables as we have seen, and it is always possible to add a additional weighting by performing only a fraction of the calculated corrections each time, or by increasing the coefficients only by one unit at a time in the calculated direction, and this in a known manner. Thus, between two adaptation cycles during which the numbers (n and n + 1) of the zones where the measured intake pressure P is measured are determined respectively, the parameters f and b of the line of regulation D according to a low-pass type curve:

Claims (4)

1. An adaptive process for regulating the injection of an injection engine comprising permanently determining the injection time (Ti) by conventional regulation in dependence on the intake pressure (P) or the intake air flow rate from a regulating straight line (D) defined by its slope (f) and its ordinate at the origin (b), disregarding other possible corrections (c, c'...), and re-adjusting periodically by successive adaptation cycles in accordance with a predetermined period, the values of the slope (f) and the ordinate at the origin (b) of the regulating straight line in dependence on any differences in richness detected by an exhaust gas analysing probe, the space of the pressures being divided into a certain number (n) of zones, to each of which corresponds a central value (P_j) of the pressure, the process being characterised in that in each adaptation cycle of a fixed period, after having determined a theoretical point (M) of operation of the engine, of an abscissa (P) corresponding to the current intake pressure measured, as well as the corresponding zone number (j), the following operations are successively performed :

- in dependence on the indications from the probe and the desired richness, determining the corrective factor (1 + a) with respect to the current regulating straight line (D) ; by merging the measured value (P) of the intake pressure with the closest value (P_j) of the centre of the zone in which (P) occurs,

- calculating and individually attributing for each zone a corrective factor (p_j), with respect to an initial straight line (Do, f_o, b_o) which is defined in a read only memory, by a purely linear calculation in dependence on the previously defined corrective factor (1 + a) and in dependence on the parameters (f, b) of the current straight line, such that :

- in dependence on the various values (β_;) in a random access memory in respect of said last corrective parameter, calculating and attributing fresh values of the parameters (f, b) of the regulating straight line (D) by purely linear formulae such that :

b_o, k_i and k'_; being constant weighting coefficients in the read only memory,

- the weighting coefficients (k_i, k'_;) used for calculation of the parameters (f, b) of the regulating straight line in the course of each adaptation cycle being so determined that said straight line passes as closely as possible to all the theoretical points (M) of operation defined in each adaptation cycle.

2. A process according to claim 1 characterised in that the engine operates with a degree of.richness equal to unity (stoichiometric mixture) and a short period for the adaptation cycle.

3. A process according to claim 1 characterised in that the engine operates with a degree of richness of less than unity, which is constant or variable, and a relatively high period, which are compatible with the use of the known over-injection process for evaluation of the corrective term (1 + a) with respect to the current straight line.

4. A process according to one of the preceding claims characterised in that between two adaptation cycles in the course of which the numbers (n and n + 1) respectively of the zones in which the measured intake pressure (P) occurs were determined, the parameters (f and b) of the regulating straight line (D) vary in accordance with a curve of the low pass filter type :

with x and x' being between 0 and 1.

Images