FR3033840A1 - METHOD FOR MANAGING THE POWER SUPPLY OF AN INTERNAL COMBUSTION ENGINE - Google Patents

Abstract

A method of managing the power supply of an internal combustion engine (2) with a mixture of air and fuel defining a richness, said engine (2) being included in a group (1) motor-propeller comprising a computer (11) ) for controlling the injection of the fuel into the engine (2), the computer (11) also being connected to a probe (10), this method comprising: a wealth regulation phase, performed by measuring the probe (10) and making it possible to obtain the term wealth regulation; - a phase of adaptation of the wealth to obtain adaptive; - a learning phase of the ethanol rate; a phase of calculating the quantity of fuel to be injected into the engine (2); - a phase of securing adaptive devices; - a phase of distribution of wealth excesses; - an adaptive validation phase.

Description

The invention relates to the learning of the ethanol content in the fuel of an internal combustion engine, in order to calculate the quantity of fuel. to be injected into the internal combustion engine. The invention particularly relates to the so-called "flex-fuel" poly-fuel engines according to the English terminology, capable of operating indifferently with gasoline, bioethanol or a mixture of gasoline and bioethanol, with rates of ethanol which can vary between, for example, 22% (fuel E22) and 100 (3/0 (fuel E100). [0003] For an engine capable of operating with several types of fuels, the level of ethanol must be measured, to modify the amount of fuel to be injected into the engine Thus, the ethanol content must be filled in after each filling of the fuel tank. [0004] Mismanagement of the quantity of fuel to be injected into the engine can lead to overconsumption of the fuel. Some engines are equipped with an ethanol rate sensor, for other engines that are not equipped with ethanol, the ethanol content is obtained by analyzing the fuel. wealth drift , the richness being a ratio defining the excess or the defect of fuel with respect to oxygen in a fuel / oxygen mixture. The rate of ethanol, on vehicles without a sensor, must be learned after each filling of the tank by the analysis of the wealth drift. It is known to measure the wealth drift of the engine exhaust with the aid of a lambda probe, a wealth control function of the calculator to then analyze this drift and calculate the appropriate amount fuel to be injected into the engine, in accordance with a predetermined air / fuel mixing objective. From the analysis of wealth results a term of wealth regulation. This analysis of the wealth regulation term is also used for a second function, called adaptation, which makes it possible to calculate adaptives in order to compensate for the differences between the models coded and calibrated in the computer (deviations related to the 3033840 2 manufacturing dispersion, wear, aging), and physical components, for example injectors. These adaptives are applied to the models and make it possible to refocus the richness of the mixture according to a given objective. A third, so-called learning function, makes it possible to inform the calculator of the rate of ethanol included in the fuel present in a reservoir. It is then necessary to manage the interaction between these three functions in order to establish an optimal correction of the richness of the mixture. [0010] According to a conceivable technique for defining the respective working areas of the learning function of the alcohol content and the wealth adaptation function, when the first filling of the reservoir with a fuel whose rate of ethanol is known, this rate is indicated on a calculator, the wealth drift being then considered as solely due to the differences between the coded and calibrated models in the calculator and the actual physical components. The wealth adjustment function will then try to refocus wealth to its goal. At the next tank fill, the ethanol level may be different, the adaptation function is inhibited and the alcohol learning function is activated until the wealth drift is canceled. The estimated alcohol level is then considered known, the learning function is stopped and the wealth adaptation function is reactivated. FR 2998925 discloses a method for assisted learning of the ethanol content in the fuel of an internal combustion engine having an air / fuel mixture maintained at the stoichiometric level (optimal mixture for which the reduction of the oxides of nitrogen is at its maximum), by a permanent control of the richness, 30 in closed loop. This control is achieved through a wealth sensor, placed in the flow of engine exhaust. This method makes it possible to memorize the value of the ethanol content and the fuel quality, before filling the tank, storing the fuel level in the tank after filling, monitoring the drift of fuel. richness by the wealth sensor and the correction of the ethanol content to return to the stoichiometric level of the air / fuel mixture. However, there are certain cases of life where previously known methods are not effective and provide an erroneous ethanol level to the calculator. For example, during a second filling with a fuel different from the first, the filling taking place before the adaptation phase has had time to learn adaptive for this new fuel, the fuel phase. ethanol rate learning will be activated and will work on a wealth drift comprising a drift related to the unknown ethanol level, but also a drift related to a gap between the models and the physical components. The ethanol content learned will not be that of the fuel present in the tank. In another example of life, if the first filling of the tank is left to the free choice of the customer and the customer decides to fill the tank with a fuel different from that indicated in the calculator, the wealth drift reports a drift both related to the difference between the models and the physical components, but also to a poorly estimated ethanol rate. The adaptation function will then offset all the wealth drift, but it will be affected by an error related to the incorrect estimation of the ethanol content. The known processes of the prior art thus have many drawbacks: risks of overconsumption of fuel, large rejection of pollutant gases, or the risk of being unable to start the engine. [0016] One objective is to propose a method for managing the supply of an internal combustion engine that makes it possible to overcome the disadvantages presented above. [0017] For this purpose, it is proposed, in the first place, a method for managing the power supply of an internal combustion engine with a mixture of air and fuel defining a richness, this engine being included in a power train comprising: - an engine block defining a cylinder, a combustion chamber, an intake duct and an exhaust duct; - an injector; A piston ; A lambda probe; a fuel pump connected to a tank and the injector; A calculator connected to the pump and the injector for controlling the injection of fuel into the engine via the intake duct, the computer also being connected to the lambda probe; this process comprising: a wealth regulation phase, carried out by measuring the lambda probe and making it possible to obtain the wealth regulation term; an adjustment phase of the wealth, to compensate for the differences measured by the wealth regulation phase thanks to adaptives; A phase of learning ethanol rate, to inform this rate of ethanol to the calculator; a phase of calculating the quantity of fuel to be injected into the engine; a phase of securing adaptive, to memorize 15 adaptives that have been validated; a distribution phase of the wealth drift, making it possible to modify the wealth regulation term in order to obtain a modified wealth regulation term; an adaptive validation phase, making it possible to confirm the coherence of these adaptives with respect to secure adaptives in the computer. This method makes it possible to have a perfect interaction between each of the phases, thus correctly distributing the wealth drifts measured and avoiding overconsumption of fuel by the engine. Various additional features may be provided, alone or in combination: the adaptive validation phase uses the secure adaptives, for the computer to establish its validation process; the distribution phase of the wealth drifts uses the adaptives, the secure adaptives and the wealth regulation term, so that the calculator determines the modified wealth regulation term; The phase of calculating the quantity of fuel to be injected into the engine uses the term "richness regulation", "adaptives" and "ethanol content", so that the calculator calculates the quantity of fuel to be injected into the engine. engine. It is proposed, secondly, a computer controlling the injection of fuel according to the management method as presented above. It is proposed, thirdly, a powertrain comprising the computer. Other objects and advantages of the invention will emerge in the light of the description of an embodiment, given below with reference to the accompanying drawings, in which: FIG. 1 is a schematic view showing the connections between a computer and the various components of an internal combustion engine; Fig. 2 is an operating diagram of a power management method of an engine; Figure 3 is a timing chart showing the state of the different phases and equivalences of wealth of Figure 2 according to a configuration of use. In Figure 1 is shown a group 1 power plant 20 including an internal combustion engine 2 comprising a motor block 3 defining at least one cylinder 4, a chamber 5 in which a piston 6 slides, the block 3 also comprises two ducts 7, 8 opening into the cylinder 4, namely an intake duct 7 and an exhaust duct 8 after combustion. [0024] In addition, the block 3 comprises a fuel injector 9 for injecting, into the inlet duct 7, a quantity of fuel L, so that it is mixed with air before entering the engine. 5. The block 3 also includes a lambda probe 10 for measuring the richness F of the fuel L contained in the exhaust gas, conveyed from the combustion chamber 5 to the exhaust duct 8. The richness F of a fuel is a ratio defining the excess or the fuel fault L with respect to oxygen in a fuel / air mixture, which is expressed as follows: DmL F = (-) xr Dma 3033840 6 Where DmL is the flow rate of the fuel mass mL, Dm, is the flow rate of the air mass ma, r is the stoichiometric ratio, ie the ratio of the air mass m and that of the fuel mL to obtain a complete combustion of the air / fuel mixture, expressed as follows: As shown in FIG. 1, the engine 2 also comprises a calculator 11. [0027] computer 11 is connected to the motor 2, and more precisely to the injector 9 and to the probe 10, physically, that is to say by means of cables, or wirelessly, for example by bluetooth, infrared or Wireless. The computer 11 then receives information from the probe 10 analyzing the richness F of the fuel L and then transmits an injection instruction, via a calculation method, a quantity of fuel L to the injector 9. The injection setpoint is in the form of an electrical signal determining an opening time of the injector 9. The injector 9 is physically connected to a reservoir, conveying the fuel L from the reservoir to the fuel tank. injector 9, the fuel L being extracted from the tank by a fuel pump 12. In order to determine the quantity of fuel L to be injected into the engine 2, the computer 11 uses a power management method of the engine 2 to determine the exact level of ethanol TX included in the fuel L , illustrated in Figure 2 by an operating diagram. [0031] Each time the tank is filled with a fuel different from the preceding fuel, it is necessary to learn the new level of ethanol TX contained therein to determine the quantity of fuel L to be injected to reach the objective. of predetermined wealth F. [0032] The operating diagram of a power management method of the engine 2 shown in FIG. 2 describes the different phases allowing the exact calculation of fuel L to be injected into the engine 2, and shows the interaction between each phases. The diagram has two branches. The first branch relates to phases that are in constant operation.

The other branch concerns phases that are triggered by particular conditions. During operation of the engine 2, a first phase 100 of the power management of the engine 2 is to measure the richness 5 F of the fuel L, by the probe 10. This measurement is made continuously over time. With this richness F, the calculator 11 is able to calculate the RF wealth regulation term, at a phase 200. The RF wealth regulation term makes it possible to estimate the wealth drift that it leads and allows to increase or decrease the amount of fuel L injected, so as to ensure a wealth consistent with a goal. The phase 200 for calculating the wealth regulation is calculated continuously, it is not subjected to the condition in phase 700. On the other hand, the adaptation phase 300 is no longer activated when the phase 700 becomes it ( tank fill detection) and as long as the 900 phase has not completed the learning of a new ethanol level. The wealth drift F has been calculated and given by the RF richness control term, an adaptive is calculated to compensate for this drift and refocus it to its wealth objective F. A phase of adaptation 300 is then initiated, to enable the establishment of ADR richness adaptives. A validation phase 400 consists in confirming that the ADR adapts calculated correspond to a wealth drift F linked to deviations between the models calibrated in the calculator 11 and the physical components, and not at a rate of ethanol TX. badly estimated. The validation phase 400 then compares the ADR-calculated adaptives with ADRS secure adaptives, the ADRS secure adaptives corresponding to the adaptives calculated for a known and referenced TX ethanol level, these ADR adaptives being recorded in the calculator 11. Secure adaptives ADRS are then subtracted from ADR adaptives. The value obtained is analyzed during a phase 500. If the value is too important, then it is estimated that the drift of wealth 35 DEV_AD learned by the adaptive becomes dangerous, and ADR adaptive are not validated, phases 800 correction and 900 learning are then initiated. In the opposite case, a security phase 600 can be started, if a condition defined below of a phase 1000 is validated. The security phase 600 consists in recording the ADR adaptives that have been validated, in order to transform them into ADRS secure adaptives and to be able to reuse them for other phases. The phases 800 and 900 can also be activated provided that filling of the reservoir is detected at a phase 700. The correction phase 800 consists in updating the term of RF richness regulation 10 to From the ADR adaptives, ADRS secure adaptives and the RF richness control term, this phase results in the modified RFM richness regulation term, calculated from the following formula: RFM = RF + ADR - ADRS [0042] The modified regulation term RFM makes it possible to determine at phase 900 the level of ethanol TX of the fuel L present in the tank. The training phase 900 then makes it possible to inform the calculator 11 of the level of ethanol TX which will subsequently enable it to calculate, in phase 1100, the adequate quantity of fuel L to be injected into the engine 2 in order to reach the objective of predetermined wealth F. Once the rate of ethanol TX filled, it is analyzed in phase 1000 by the computer 11. It is possible that this TX rate is recognized as close to a known value and stored in the computer 11, this concerns in particular the E22 and E100 fuels composed respectively of 22% and 100% of ethanol. If the TX ethanol level learned is close to these values, then a request is made to phase 600 to secure the previously calculated ADR adaptives which made it possible to establish this TX ethanol level. If, and only if, the conditions of the phases 500 and 1000 are validated, then the security phase 600 is initiated, and the adaptive ADR are stored in the computer 11 as ADRS secure adaptives. If the TX ethanol level is not recognized as close to a known value, then the phase 1100 is started, which consists in calculating the quantity of fuel L to be injected into the engine 2 in order to 3033840 9 bring wealth F closer to his goal. Phase 1100 works from the RF richness control term, the ADR adaptives, and the TX ethanol level learned in phase 900, even if it is not recognized as close to a known value. Ultimately, the first branch of the diagram aims to monitor and validate adaptive ADR, while the second branch aims to learn the rate of ethanol TX, from a regulatory term of modified wealth RFM. An example of power management of an internal combustion engine is shown in FIG. 3. FIG. 3 comprises a timing diagram and a legend for defining each block present in the timing diagram. The timeline is divided vertically into five columns, each referring to a step. In what follows, reference will be made to steps E0 to E4. The chronogram of FIG. 3 is also shared horizontally in six lines, the first four referring to the different equivalences of wealth, notably: the wealth equivalence of the wealth regulation term, denoted EQ_RF; The richness equivalence of the modified wealth regulation term, denoted EQ_RFM; - the richness equivalence of adaptive richness, noted EQ_ADR; - the equivalence in richness of the rate of ethanol, noted EQ_TX.

The last two lines of the timing diagram represent the activation state of the learning phases AP and adaptation AD. The blocks of the timing diagram refer to: adaptive richness ADR, represented by hatching; drifting wealth adaptive DEV_ADR, represented by 30 alveoli; at the level of ethanol TX, represented by triangles; drifting wealth of ethanol DEV TX, represented by vertical lines; drifting from the learning phase DEV_AP, represented by double hatching. The initial step EO corresponds to a case of life where several refills of the fuel tank have already been carried out, with consecutive fuel changes, thus distorting the value of the ethanol level TX, which is learned by the phase d. AP learning. In addition, the learning phase AP generates a drift of the TX ethanol rate learned DEV_AP, related to its performance. In this case of life, the adaptation phase AD does not make it possible to establish a complete correction of the wealth drift, and this one only learns a part ADR of the drift, the residue DEV_ADR, 10 not yet learned by the adaptation phase AD, is compensated by the RF richness regulation phase. The TX ethanol level is also not learned in full and the unexamined part DEV_TX is compensated by the RF richness control phase. The initial value of EQ_RFM is calculated as such: We set ADRS = ADR + DEV_ADR (total deviation of the components) we know that EQ_RFM = EQ_RF + ADR - ADRS from the initial data of the graph we have: 20 EQ_RFM = DEV_TX + DEV_AP + DEV_ADR + ADR - ADRS = DEV_TX + DEV_AP + DEV_ADR + ADR - ADR - DEV_ADR Let EQ_RFM = DEV_TX + DEV_AP [0054] The following step El starts with an AP learning phase of the ethanol level TX, when the detection of a filling of the tank for example. The learning phase AP works from the modified RFM richness regulation term and learns a part of the DEV_AP drift and then changes the TX ethanol level. The rich equivalence of the modified RFM richness regulation term always includes the DEV_AP richness drift, related to the performance of the AP learning phase. The next step E2 begins with an adaptation phase AD of the richness, it works from the RF richness regulation term and learns the entire wealth drift DEV_ADR and DEV_AP 35, thus evolving adaptive richness ADR. As all the wealth has been absorbed, the RF richness control term becomes zero. In contrast, the RFM modified richness control term 3033840 11 still includes a DEV_AP richness drift. The RFM modified wealth regulation term is calculated as follows: RFM = (RF + ADR) - ADRS RF = DEV_AP + DEV_ADR is known to be RFM = (DEV_AP + DEV_ADR + ADR) - (ADR + DEV_ADR) if RF = 0 then RFM = 0 + DEV_AP + DEV_ADR + ADR - ADR - 10 DEV_ADR one thus obtains RFM = DEV AP [0056] The following step E3 begins with a learning phase AP of the ethanol rate TX (new filling of the tank ), this one working from the RFM modified wealth regulation term it then learns the DEV_AP wealth drift. At this moment, the wealth drift DEV_AP is both learned by the adaptation phase AD and by the learning phase AP, the wealth is then deviated from a drift DEV_AP too much. This drift is compensated by the RF richness control phase. The last step E4 begins with an adaptation phase AD of the wealth, working from the RF richness regulation term, to learn the entire wealth drift DEV_AP. The RF richness regulation term and the RFM modified wealth regulation term are then zero, so the richness F is to its purpose, the ADR richness adapters only learned the differences between the calibrated models in the calculator. and the physical components, and the TX ethanol level is very close to the ethanol content of the fuel in the tank. The method thus described has several advantages, among which: the detection of life events or the TX ethanol level is known and the storage of the ADR adaptives associated with this TX ethanol level, the AP learning of the rate TX ethanol with RFM modified wealth drift information, monitoring and limiting the value of adaptive ADR. All these advantages make it possible to inform the calculator 11 an exact TX ethanol level, the computer 11 then injecting the appropriate quantity of fuel L into the engine 2, thus avoiding any overconsumption of fuel, reducing the rejection of pollutant gases and ensuring a correct calculation of the ignition advance of the engine 2 at each fuel change. The invention makes it possible to secure the adaptive richness, in the case of life where the ethanol content is safe, for example when the alcohol level is very close to the physical limit of 100% (fuel 10 E100 ), or the physical limit of 22% (fuel E22), or even when it is detected that the first filling of the tank is carried out with an identified fuel. In these situations, Wealth Adapts only correct the wealth drifts related to the difference between the models and the physical components of the vehicle. The invention furthermore makes it possible to feed a learning function with a modified wealth regulation term, distributing the wealth drift between the learning of the ethanol content and the wealth adaptives. Advantageously, a permanent monitoring is performed of the difference between the current richness adaptive and the secure richness adaptive. If this difference becomes greater than a threshold value, reactivation of the learning function will be required. This case may correspond for example to the filling of a tank with a fuel having an ethanol content very different from that known by the computer, as long as the learning function could not identify this rate.

Claims (6)

REVENDICATIONS1. Method for controlling the supply of an internal combustion engine (2) with a mixture of air and fuel (L) defining a richness (F), this engine (2) being included in a power unit (1) comprising: - a motor block (3) defining a cylinder (4), a combustion chamber (5), an intake duct (7) and an exhaust duct (8); an injector (9); a piston (6); a probe (10) lambda; a fuel pump (12) connected to a reservoir and to the injector (9); a calculator (11) connected to the pump (12) and to the injector (9) for controlling the injection of fuel (L) into the engine (2) via the intake duct (7), the computer (11) ) being also connected to the lambda probe (10), this method comprising: a wealth regulation phase (F), performed by measuring the lambda probe (10) and making it possible to obtain the wealth regulation term (RF); - an adaptation phase (AD) of the richness (F) to compensate for the differences measured by the wealth regulation phase (F), using adaptive (ADR); a learning phase (AP) ethanol rate (TX), to inform this rate of ethanol (TX) to the computer (11), a calculation phase of the amount of fuel (L) to inject into the motor (2); this method being characterized in that it comprises: a phase of securing the adaptives (ADR), making it possible to memorize those adaptives (ADR) that have been validated; a wealth distribution (F) distribution phase, making it possible to modify the wealth regulation (RF) term, in order to obtain a modified wealth regulation (RFM) term; 3033840 14 an adaptive validation phase (ADR), to confirm the consistency of these adaptive (ADR) compared to secure adaptive (ADRS) in the calculator (11). 2. Management method according to claim 1, characterized in that the adaptive validation phase (ADR) uses the secure adaptive (ADRS) so that the computer (11) establishes its validation process. 3. Management method according to claims 1 and 2, characterized in that the distribution phase of the wealth drifts 10 (F) uses the adaptive (ADR), the adaptive secure (ADRS) and the wealth regulation term (RF) ), so that the calculator (11) determines the modified wealth regulation (RFM) term. 4. Management method according to claims 1, 2 and 3 characterized in that the phase of calculating the amount of fuel (L) 15 to be injected into the engine (2) uses the term wealth regulation (RF), the Adaptive (ADR) and ethanol (TX), so that the computer (11) performs the calculation of the amount of fuel (L) to be injected into the engine (2). 5. The fuel injection (L) control computer (11) according to the management method as set forth in any one of claims 1 to 4. 6. Group (1) propulsion motor comprising the computer (11) according to claim 5.