WO2016091848A1

WO2016091848A1 - Fuel injection control in an internal combustion engine

Info

Publication number: WO2016091848A1
Application number: PCT/EP2015/078917
Authority: WO
Inventors: Sebastien COUET; Thierry Cochet; Philippe ARCHAMBAULT; Christian FASOLO; Fabien DE OLIVEIRA; Guillaume Meissonnier
Original assignee: Delphi International Operations Luxembourg S.À R.L.
Priority date: 2014-12-09
Filing date: 2015-12-08
Publication date: 2016-06-16
Also published as: GB201421853D0

Abstract

A method of fuel injection control in an internal combustion engine having at least one cylinder with an associated solenoid actuated fuel injector for performing injector events is presented. The injector has a valve group with a needle, which is at least indirectly controlled by a pressure force in a control chamber, which is in turn controlled by a control valve operated by the solenoid actuator. The method comprising the steps of: a) performing at least one injector event by applying, to said electromagnetic actuator, a drive pulse having a predetermined pulse width (PW); b) recording the voltage across said electromagnetic actuator, after the end of the drive pulse; c) processing the recorded voltage data to identify a predetermined waveform pattern in the corresponding secondary voltage derivative, following the end of the drive pulse; and d) computing a valve indicator (FWM) representative of the magnitude of said predetermined waveform pattern. This valve indicator can be used in a variety of applications related to injection control, and over the full injector operating range, e.g. for monitoring viscosity or lacquering, determining MDP, correcting for drift or voltage changes, determining the control valve lift.

Description

FUEL INJECTION CONTROL IN AN

INTERNAL COMBUSTION ENGINE

FIELD OF THE INVENTION

The present invention generally relates to internal combustion engines and more generally to injection control in such engines.

BACKGROUND OF THE INVENTION

The contemporary design of internal combustion engines must cope with the increasingly stringent regulations on pollutant emissions. Accordingly, automotive engineers strive for designing engines with low fuel consumption and low emission of pollutants, which implies including electronic devices capable of monitoring the combustion performance and emissions.

In order to comply with this emissions standard, a proper operation of a fuel-injected engine requires that the fuel injectors and their controller allow for a timely, precise and reliable fuel injection, whatever the operating condition and during all injector lifecycle (at least 240 000 km). Indeed, it is well known that problems arise when the performance, or more particularly the timing, and the quantity of fuel delivered by the injectors diverge beyond acceptable limits. For example, injector performance deviation or variability will cause different torques to be generated between cylinders due to unequal fuel amounts being injected, or from the relative timing of such fuel injection. This problem is particularly acute when injecting small fuel quantities, due to response delays at opening and closing. The injector performance is also affected by a number of uncontrolled parameters such as for example: injector ageing, deposit, fuel properties (especially viscosity)...

In order to take into account the individual specificities of an electromag- netically actuated fuel injector, it has been proposed to associate to a given fuel injector a number of performance parameters thereof. These performance parameters are, e.g., encoded in a bar code applied to the injector, so that the performance parameters can be retrieved by a bar code scanner at the time of installation in the engine and transferred to the engine control unit (ECU).

Such method for fuel injector parameters installation is for example de- 5 scribed in US 7,136,743.

Another method of fuel injector installation has been disclosed in WO201 1/073147, which uses a segmented master performance curve. Each fuel injector to be installed in the engine is provided with specific fuel injector parameters in a machine-readable format, and these parameters are transi t) ferred to the engine ECU. Fitting information, preferably coefficients for a characteristic equation attributed to each respective segment of the master flow curve, are contained in these fuel injector specific parameters.

EP2375036 discloses a method for determining an injector closing time based on the voltage trace measured at the coil of the electromagnetic actuator 15 of the injector.

Although the above-mentioned methods are beneficial in that they allow more appropriately describing the flow performance per injector and provide finer control in the ballistic zone, it is still desirable to find ways of improving the control of fuel injection in an internal combustion engine. In particular, it is 20 desirable to make sure that an injector has properly opened in the ballistic area, or to be able to properly detect fuel viscosity, or lacquering in order to take appropriate measures.

SUMMARY OF THE INVENTION

The present invention provides an improved method of controlling fuel injection in an internal combustion engine having at least one cylinder with an associated electromagnetically actuated fuel injector for performing injector events, wherein for each injector event a drive signal is applied to said fuel injector. The fuel injector comprises: a valve group having a needle slideably arranged therein in order to control at least one spray orifice through its displacement; and a control chamber arranged to be filled with high pressure fuel so as to exert, directly or indirectly, a pressure force on the needle in its closing direction, an outlet path of the control chamber being controlled by a control valve operated by an electromagnetic actuator with a coil.

The method comprises the steps of: a) performing at least one injector event by applying, to said electromagnetic actuator, a drive pulse having a predetermined pulse width; b) recording the voltage across said electromagnetic actuator, after the end of the drive pulse; c) processing the recorded voltage data to identify a predetermined waveform pattern in the corresponding secondary voltage derivative, following the end of the drive pulse; and d) computing a valve indicator representative of the magnitude of said pre- determined waveform pattern.

The present invention relies on the finding that the secondary voltage derivative reflects an individual characteristic of the control valve, which is related to the movement of the valve member of the control valve. When the control valve has been opened and returns to its valve seat (closed position), it tends to rebound on the valve seat and cause a glitch on the voltage trace. It has been found that the timing of this control valve closing can be accurately determined from the secondary derivative of the coil voltage of the electromagnetic actuator. More specifically, it has been found that the timing of closing coincides with a predetermined waveform pattern in the corresponding secondary voltage derivative, and that the magnitude of this waveform can be used as valve indicator that can be employed in a variety of applications, and this over the full operating range of the injector.

Preferably, the waveform pattern is the first wave, in particular the first half- wave, of the waveform described by the secondary voltage derivative following the end of the drive pulse. The first wave or half-wave may be detected as a predetermined increase in the secondary voltage derivative.

The predetermined waveform pattern, respectively the first wave or half-wave, is advantageously amplified. The valve indicator is then determined by compu- ting the magnitude of the amplified waveform pattern, respectively of the amplified first wave or half-wave.

In one embodiment, the amplification is carried out by convoluting the measured voltage data with a predetermined function of similar shape, preferably a triangle function. The triangle function may be configured so that the triangle has a width corresponding to the width of the first half-wave.

Method Step c) advantageously involves processing said recorded voltage data to compute filtered secondary derivative voltage data, and identifying the predetermined waveform pattern from the filtered secondary derivative voltage data. The filtering step preferably involves a low pass filter, an in particular a weighted moving average calculation.

To filter out noise and non-significant actuations, the determined valve indicator is used for injection control only if it exceeds a predetermined threshold.

Once a valid valve indicator value has been obtained (i.e. above threshold), the closing time of the control valve can be determined as the timing of the local extremum of the predetermined waveform pattern. Also, the control valve closing delay may then be generally calculated as the difference between the closing time and the length of the drive pulse (pulse width).

The valve indicator, control valve closing time and closing time constitute so- called glitch data that can be used in part or as a whole for a variety of injection control strategies in the engine, namely for correcting injection parameters such as e.g. a pulse width of an injection drive signal, an injector needle opening or closing delay... Examples of control strategies based on the glitch data are given below.

1. Viscosity. The valve indicator correlates satisfactorily with the fuel viscosity and may thus be used to monitor the latter. The actual knowledge of the fuel viscosity while the engine is running permits implementing correction strategies. In particular the pulse width, injector needle opening delay or injector needle closing delay may be corrected by an offset determined from a mapping having viscosity as input, the valve indicator being dependent on the viscosity.

2. Lacquering. The valve indicator may also be used to monitor control valve lacquering. As a matter of fact, glitch event data of injectors affected by lacquering significantly differ from nominal values for all the injector family. Injector lacquering can thus be detected by monitoring the glitch event data. 3. MDP. A further use of the valve indicator is to determine the minimum delivery pulse of the control valve. This can be done by repeating the routine of steps a) to d) with increasing pulse width, starting from a non-injecting pulse width, until the valve indicator exceeds a predetermined threshold, and computing the minimum delivery pulse MDP based on the difference between the pulse width PW and closing delay CD (MDP_V|_V = PW - CD_V|_V).

4. Voltage change. The difference between minimum delivery pulse at a reference and current voltage may be used to correct the injection pulse width for battery voltage changes.

5. Drift. The injection pulse width may be corrected for injector drift based on a difference between current and reference closing delays.

6. Valve lift. The time period for opening the control valve may be calculated based on the difference between the timing of fully open control valve and minimum delivery pulse. The knowledge of the time period for opening the control valve allows reading a correction value APW from a mapping designed for the injector family: APW = f(time period for opening), which can be used as offset to correct the pulse width on an injection drive signal.

The present invention thus also concerns a variety of fuel injection control method implementing correction strategies based on the above-mentioned applications. According to a further aspect, the present invention concerns a system for controlling fuel injection in an internal combustion engine having at least one cylinder with an associated electromagnetically actuated fuel injector with a coil, wherein the system comprises a programmable computing device config- ured for implementing the above method and applications thereof.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

FIG. 1 : shows two graphs: a) current and voltage evolution vs. time at the coil of the injector's electromagnetic actuator, and b) voltage trace vs. time after the end of the pulse width;

FIG. 2: shows two graphs: a) coil voltage after the end of the pulse width and b) the filtered second derivative of the coil voltage during a time window about the glitch;

FIG. 3: shows three graphs: a) the filtered second derivative of the coil voltage during a time window about the glitch - similar to Fig.2 b); b) is the triangle function used for convolution; c) is the processed voltage waveform resulting from the convolution of the two signals a) and b);

FIG. 4: shows three graphs: a) coil secondary voltage vs. time after the PW and at different temperatures; b) the filtered and convoluted signals of Fig. 4a) vs. time; c) viscosity vs. glitch magnitude evidencing the substantially linear correlation;

FIG. 5: is (a) a graph of the closing delay vs. time and of (b) the valve indicator vs. time for a fuel injector affected by lacquering;

FIG.6: is a graph illustrating the variation of current and valve lift vs. time for a corresponding pulse witdh;

FIG. 7: shows two graphs of injector flow performance curves (fuel quantity vs.

pulse) at different battery voltages; where graph b) is corrected based on MDP_vaiv_e;

FIG.8: shows (a) a plot of the control valve closing delay vs. pulse width and (b) a detail of this graph around the first peak; and

FIG.9: is a graph of the injected fuel quantity vs. time required for the full valve lift of the control valve.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

The present invention addresses the problem of performance variability of fuel injectors, which is particularly acute in the ballistic region in the case of some modern designs of electromagnetically (solenoid) actuated fuel injectors. As it is known, an electromagnetically actuated fuel injector generally compris- es a valve group having a needle that is axially movable in order to open and close one or more flow orifices through which fuel is sprayed into an engine combustion chamber. In modern diesel fuel injectors, as used e.g. in common rail systems, the opening and closing of the needle is controlled by a hydraulic pressure force in a control chamber, the pressure force being adjusted by means of the actuator . This control chamber is arranged to be filled with high- pressure fuel so as to exert, at least indirectly, a pressure force on the needle in its closing direction. An outlet path of the control chamber is controlled by a control valve, which, when open, allows fuel from the control chamber to escape and thus creates a pressure drop in the control chamber, thereby causing the needle to lift and open the flow orifice in order to spray fuel into the engine combustion chamber. This control valve is operated by the electromagnetic actuator. In the context of the present invention the electromagnetic actuator includes a solenoid, which typically has an excitation coil arrangement and a movable armature. For actuation of the control valve, the movable armature is connected to the movable valve member (obturating member) of the control valve. A fuel injector comprising a control chamber with a valve member operated by means of a solenoid actuator is e.g. known from EP 2 647 826. The fuel injector, specifically the electromagnetic actuator is operated by a drive signal that is applied during a duration known as "pulse width" (PW). Generally, to inject a fuel amount Q, a value of pulse width is read from a table and possibly corrected, and the fuel injector is operated, for a given injector event, so that the drive signal is applied during a time period corresponding to the pulse width, to influence a desired injection time and normally inject a given fuel amount. Hence, for any electromagnetically actuated fuel injection to be performed, a PW is generated to command a corresponding injector opening duration in order to deliver fuel. During the PW, the electromagnetic actuator is energized and an electric current flows therethrough at a defined voltage. The mobile part of the electromagnetic actuator is magnetically attracted by the magnetic coil and lifts the valve member of the control valve to open the outlet path of the control chamber creating thereby a pressure drop therein, which in return lifts the injector needle against a closing spring. Once the pulse ends, the electromagnetic actuator is no more energized and the applied drive voltage is removed; as the magnetic attraction is removed, the control valve returns to its seat to thereby close the outlet path of the control chamber. Consequently the injector needle returns to its closing position under the action of the closing spring and fuel pressure. This leads to the build up of eddy currents through the magnetic coil due to the movement of the actuator mobile part. The electromagnetic actuator may be used as a sensor to measure the eddy current and its voltage. Especially, the control valve closure movement can be detected by monitoring the voltage across the actuator coil at the end of the injection. The voltage evolution during the PW and thereafter are shown in Fig .1 .

For the timing PW (between time t_start and t_End, the drive voltage is applied to the coil and current flows therethrough. At the end of the PW (i.e. after t_End) the measured voltage curve describes an exponential decay. This is well known in the art. The circle in Fig .1 identifies a discontinuity, also herein called "glitch", in the coil voltage curve or trace, occurring after the end of the PW. The glitch, which can be better seen in the voltage vs. time graph of Fig .1 b), corresponds to the closing time of the control valve, and the timing of the glitch can thus be considered as the closing time of the injector needle. The glitch is due to a modification of the eddy current induced in the coil when the valve member of the control valve hits its valve seats.

The determination of the glitch and of its timing allows computing the so- called "closing delay" (CD), which is the time period between the end of the pulse and the time of the glitch, when the closing member of the control valve reaches its closing position, causing the closing of the injector needle.

Fig .2 a) is a graph of the filtered coil voltage versus time after the end of pulse. Again, the circle indicates the glitch in the coil voltage trace, which occurs at the time of the closing of the control valve.

It has been found that the actual timing of the control valve closing (noted t_c) can be determined from the secondary derivative — of the recorded dt

voltage trace as measured at the coil of the electromagnetic actuator.

Fig .2 b) shows the secondary voltage derivative of the voltage curve of Fig .2 a) around the glitch. As can be seen, the voltage secondary derivative curve has a characteristic sine waveform, and more specifically that of a damping sine waveform. It may be noted here that the signal of Fig.2b) is a filtered second derivative of the voltage data; a low pass filter may be used. A possible calculation and filtering algorithm is proposed below.

The present inventors have observed that the local extremum (i.e. minimum or maximum) of the first half-sine waveform, following the end of the pulse width (t_End) corresponds to the closing time t_c of the control valve. With reference to Fig.2b), the local extremum is a minimum and the closing time is thus t_c=120 με.

It has further been found that the magnitude of this first half wave correlates with the speed of the valve member (i.e. closing member) of the injector control valve (i.e. the valve controlling the pressure applied on the injector needle in the control chamber). Since the valve member is typically elastically biased (e.g. by a spring) in the closing direction, the valve member reaches its maximal speed just before hitting its seat. The valve member further tends to rebound on its seat. This rebound phenomenon is believed to be the cause for the damping sine waveform observed in the voltage secondary derivative. As it will be understood, the greater the valve speed, the greater the magnitude of the first half-sine.

The magnitude of the first half-sine can then be used as a valve indicator that reflects the speed of the valve member when returning to its closed position. The valve indicator, i.e. a value reflecting the magnitude of the first half-sine, is hereinafter noted FWM. In the context of a generally sine-shaped waveform, the term magnitude herein designates the change of the signal over the first half-wave (it is also known as peak amplitude or semi-amplitude, i.e. the half of the peak-to-peak amplitude).

This valve indicator FWM can then be used in a variety of applications related to injection control, and over the full injector operating range, as will be discussed below.

It is however to be noted that for low pulse widths, namely in the ballistic region, the stroke of the valve member may be too small for the valve member to fully open, and under some circumstances, even too small for injection to occur. However, since the sine waveform of the voltage secondary derivative is caused by the valve member hitting the valve seat, such waveform is likely to be observed on the voltage secondary derivative curve also where actually no injection occurred, provided the valve member has been displaced.

To be able to filter out such situations where the valve opening was too small and/or too short for injection to occur at the spray tip, the determined magnitude of the first half-sine can be compared to a predetermined magnitude threshold. This magnitude threshold is calibrated in function of the injector technology and based on testings. Hence, in practice, one will preferably consider that the measured magnitude of the first half-sine is significant, and hence can be used as valve indicator FWM, when it is greater than the calibrated, predetermined magnitude threshold.

1. Practical embodiment

In order to implement the present method in the engine, it should preferably require only a small amount of computing resources and be adapted to handle the measured voltage signal, which may be relatively weak. It is there- fore preferable to amplify this signal in the region of interest, i.e. around the first half-sine of —- .

dt²

1.1 Derivative calculation routine

In the engine, voltage data measured at the injector actuator coil after the end of the pulse width are generally stored in a table with the corresponding timings. The second derivative can e.g. be computed using the following approach.

- Computation of the first derivative:

V (t) - V (t - l)

A t

- then the first derivative data Di(t) are filtered with a weighted moving average:

n

∑ z^'( (t - i) + D_l(t + ) + (n + l)Dl(t)

where n is a parameter of the filter.

- The obtained filtered first derivative data are then derivated again with the same algorithm.

These 3 steps can in fact be combined as a simple algorithm:

^-ζ- (t) = Constl^■ (V(t - Const!) + V(t + Const!) - !^■ V(t))

dt where Constl and Constl depend on filter weights and sampling frequency.

1.2 Convolution

In order to obtain robust data, the half-sine waveform is advantageously amplified by convolution with a selected function, and preferably a triangle function. This treatment is illustrated in Fig.3.

Fig.3 a) shows the voltage secondary derivative waveform as in Fig.2 b). This waveform is convoluted with the triangle function of Fig.3 b), leading to the amplified secondary voltage waveform shown in Fig.3c).

The magnitude of this amplified first half-sine is then the value which is advantageously used as valve indicator FWM.

Again, to avoid false detection, the valve indicator FWM is validated only if it is above a predetermined magnitude threshold. In the affirmative, the timing of the maximum of the first half-sine is used to compute the closing delay:

CDviv = t_c - PW

2. Monitoring fuel viscosity

Fuel viscosity is a critical parameter in cold conditions which affects the injected fuel quantity. But fuel viscosity can be very different from one fuel to another and the fuel property is a priori unknown to the engine control unit (ECU). Moreover, a significant increase in fuel viscosity leads to changes in the injector flow curve, as well as in the opening and closing delays. The knowledge of the fuel viscosity would allow adapting relevant injection parameters.

Figure 4 a) shows the actuator coil voltage trace after t_End for several fuel temperatures. As can be seen, as the fuel temperature drops, the glitch timing, i.e. the closing time t_c, is delayed (shift to the right) and its intensity appears to be reduced. This can be better observed on the corresponding secondary voltage derivative curves shown in Fig.4b): the magnitude of the first half-sine decreases with increasing viscosity.

It has thus been found that the present valve indicator FWM satisfactorily correlates with the fuel viscosity, as shown in Fig.4 c).

In a preferred method using the present valve indicator FWM for estimating the fuel viscosity, it is preferable to use a variation of the valve indicator to take into account the differences between each injector, in particular with the help of a reference point done in hot conditions (where fuel viscosity is known as very low).

A proposed algorithm for fuel viscosity estimation is the following:

• Determining reference values of the glitch event in hot conditions: closing delay in hot conditions is noted CD_hot The valve indicator in hot conditions is noted FWM_hot

• Determining the glitch event values in the desired operating conditions at lower temperature: CD_COid and FWM_COid

• Determining the difference between hot and cold conditions:

ACD= CDcoid - CD_hot; AFWM= FWM_COid - FWM_hot · Estimating the fuel viscosity using a mapping in function of the glitch event data comparison:

Visco = F(ACD, AFWM) The actual knowledge of the viscosity then allows modifying injection parameters such as: o correcting the Pulse Width by a pulse offset value determined from a mapping: PW₀ff_Set = f(Visco) o correcting the needle opening delay using by an offset determined from a map: OD₀ff_Set = f(Visco) o correcting the needle closing delay using an offset determined from a map: CD_INJ₀ff_Set = f(Visco)

By being able to modify the pulse length, the opening and closing delay as a function of viscosity variation, it is possible to significantly improve the accuracy of the delivered fuel quantity, which is also favourable with regard to the cold start time as well as cold idle stability.

3. Injector lacquering monitoring

Another phenomenon affecting the injection of an engine is the deposit in the control valve. In fact, while it is desirable that injectors have a steady performance, fuel quality is not the same around the world. Poor quality fuels can produce deposit (e.g. lacquering) inside the injector and modify the injector performance. Deposit at the control valve slows down the valve, and in worst case situations leads to sticking of the valve. At high temperature, injection still works up to a certain amount of deposit. However, this phenomenon becomes of great importance when operating at cold temperatures or engine start, where the injector simply fails to deliver fuel. When there is deposit on the control valve, the glitch event data (control valve closing delay and valve indicator FWM) are strongly modified. The striking drift of the glitch event data can be observed from Fig.5, wherein the grey rectangles 2 and 4 show the nominal operating regions of the injectors. As it appears, the injectors in the test operate clearly out of the nominal regions: region 2 indicates the nominal values of valve closing delays; region 4 indicates the nominal values of the valve indicator FWM.

As can be seen, the glitch event data of injectors affected by lacquering significantly differ from nominal values for all the injector family. This confirms that injector lacquering can be detected by monitoring the glitch event data.

A possible algorithm for valve deposit detection is proposed:

• determining the current glitch event data CD and FWM

• comparing the current glitch event data with reference (nominal) operating thresholds of closing delay CD_Ref and valve indicator FWM_Ref

• Concluding to a valve deposit situation where both CD is higher than CD_Ref and the valve indicator FWM is lower than FWM_Ref

Early detection of deposit inside the injector can be alerted with a warning light before injector failure and possibly adapt earlier injection pattern / parameters when deposit occurs.

4. Control valve opening

Another aspect of interest for controlling the injection of fuel in an engine is the detection of the control valve opening delay, in particular for small pulses such as in the in the ballistic operating region. As is it known in the art, the term "ballistic" is generally used to designate needle movements for which the needle essentially opens and closes, without remaining in (or even reaching) the fully open position. Such ballistic behavior also exists for the control valve, specifically its valve member. The problem of operating in the ballistic zone is that the global behavior is particularly affected by opening and closing responses/delays.

It has been observed in the context of the present invention that in the ballistic region the valve closure lasts the same time as the valve opening (CDviv=OD_Viv), since the valve lift curve generally describes a bell shape or the like.

In order to avoid inertia effects and to be able to do carry out learning without fuel injection, the learning of the valve opening delay is preferably done at the smallest possible pulses.

The following algorithm is proposed to determine the valve's minimum drive pulse needed to move the injector control valve (the valve MDP is illustrated in Fig.6): a) Set the PW at a known non-injecting pulse b) perform injector event (pulse) for the set PW c) Record voltage trace and compute valve indicator FWM: a. If measured valve indicator FWM is less than the predetermined threshold (FWM < FWM_th): go back to step b with in- creased pulse: PW=PW+5PW b. If glitch magnitude is found (i.e. FWM > FWM_th), then compute MDP_Viv = PW - CDviv

This MDPviv can then be used for pulse width adaptation over the full injector operating range.

5. Battery Voltage Compensation

When the battery voltage changes, the time needed to establish the electromagnetic force changes as well. In other words, for a same pulse length at different battery voltages, the injected fuel quantity is not the same because the valve opens later or sooner. This phenomenon is conventionally compensated by an "average" map for all injectors.

MDP valve learning according to the present method can also be useful to compensate for battery voltage changes. Fig. 7 a) below shows the flow performance curve for one injector at 4 different rail pressures, and at 10 V and 14 V respectively (10V curves are indicated 6, 14V curves indicated 8). It can be observed that the injected quantity is lower at lower voltage, for a same pulse width. Conversely, to inject a given fuel quantity a larger pulse width is required at lower voltages.

It has been found that the higher and lower voltage curves are in fact offset by an amount corresponding to MDP_V|_V. The following routine can thus be used to compensate for battery voltage: · Store MDP valve at nominal battery voltage to be used as reference: MDP_ref

• Learn MDP value: MDP_cur

• Compare learned value to reference: AMDP= MDP_cur - MDP_ref

• Adapt pulse length: Pulse_new = Pulse_n0minai + AMDP The pulse length to be used for injection is thus calculated as the difference between the nominal pulse width, less the difference between the current and reference MDPs.

The result is illustrated in Fig. 7b), where it can be observed that the flow curves obtained by applying the above correction routine leads to flow curves at 10V that substantially correspond to those obtained at 14V.

The benefits are the ability to compensate the battery voltage effect for each injector, decrease cylinder to cylinder dispersion and compensate partially injector drift during its lifetime.

6. Valve lift estimation

It has been observed that the control valve closing delay depends on the pulse width. Figures 8 a) is a graph of the closing delay vs. pulse width; and Fig.8 b) illustrates a detail of Fig.8a). Looking closer at the curve of Fig.8a), one can observe that the closing delay first increases with the pulse width. In this linear region, the valve operates in the ballistic zone. Around PW=260 s, the CD reaches a peak at 136 με, undergoes a small decrease to subsequently reach a steady level at about 140 με. First peak illustrated in Fig.8 b) is due to control valve inertia and rebound of the valve member against the top seat (fully open).

The longer the stroke of the control valve, the longer the time required to reach the valve full lift. Accordingly, the time needed to reach the valve's fully open position (noted T_viv_fuii_iift) is an image of the actual lift (total stroke) of the valve lift (noted VL). It is thus possible to estimate the valve lift by computing the time required to move from the fully closed to fully open positions of the valve member.

The following algorithm may be used for determining the lift of the control valve: · Determined the MDP of the valve (as in section 4 above)

• Find the PW corresponding to of the top of the first wave:

Pulse_top_ wave

• Compute time needed to reach the valve full lift:

Tvivjuiijift = Pulse_t0 wave MDP

The control valve lift is known as an important parameter influencing injector mass flow. Moreover, the control valve lift is also known as a parameter moving during the injector life. Fig.9 illustrates the influence of valve full lift on the injected fuel quantity. The graph is obtained for 12 different injectors to which a PW of δθθμε is applied. A correlation between the injected fuel quantity and full lift is clearly observed.

The following algorithm can be used to correct the injected fuel quantity:

• Determine time needed to attempt valve full lift (see previous algo- rithm T_V|_V__fUi_Uift)

• Read a correction value APW from a mapping designed for the injector family: APW = f(Tviv_fuii_iift)

• Adapt the nominal pulse PW„_om: Pulse = PW„_om + APW

There are several benefits to this correction; it allows:

• estimating a physical injector parameter and its evolution during injector life;

• compensating partially injector wearing; · decreasing cylinder to cylinder dispersion.

7. Injector drift estimation

During injector life, the control valve closing delay is likely to vary due to wear of the control valve seat. This will alter the injector closing delay as well as injected fuel quantity.

As explained above, the control valve closing delay can be monitored. Accordingly, partial compensation of the injector drift can be achieved by the following algorithm:

• determine and store the valve closing delay when the injector is new (stored as CD_ref)

• regularly determine the valve closing delay: CD_cur

• compare the current CD_cur value to the reference: ACD_V|_V = CD_cur - CD_ref

• Adapt PW if the difference between the learned value and the reference is higher than a threshold: PW = PW„_om + f(ACD_V|_V > CD_thr)

This algorithm is beneficial for partial compensation of injector ageing and can be used to adapt compensation strategy depending on injector timing.

Claims

A method of determining a parameter of a fuel injector in an internal combustion engine having at least one cylinder with an associated electro- magnetically actuated fuel injector for performing injector events, wherein for each injector event a drive signal is applied to said fuel injector, said injector comprising: a valve group having a needle slideably arranged therein in order to control at least one spray orifice through its displacement; a control chamber arranged to be filled with high pressure fuel so as to exert, at least indirectly, a pressure force on said needle in its closing direction, an outlet path of said control chamber being controlled by a control valve operated by an electromagnetic actuator with a coil; said method comprising the steps of: a) performing at least one injector event by applying, to said electromagnetic actuator, a drive pulse having a predetermined pulse width (PW); b) recording the voltage across said electromagnetic actuator, after the end of the drive pulse; characterized by the steps of c) processing the recorded voltage data to identify a predetermined waveform pattern in the corresponding secondary voltage derivative, following the end of the drive pulse; and d) computing said parameter, said parameter being a valve indicator (FWM) representative of the magnitude of said predetermined waveform pattern.

The method according to claim 1 , wherein said waveform pattern is a first wave, in particular a first half-wave, of the waveform described by the secondary voltage derivative following the end of the drive pulse.

The method according to claim 1 or 2, wherein said predetermined wave- form pattern, respectively said first wave or half-wave, is amplified and said valve indicator is determined by computing the magnitude of the amplified waveform pattern, respectively of said amplified first wave or half-wave.

4. The method according to claim 3, wherein said amplification is carried out by convoluting the measured voltage data with a predetermined function of similar shape.

5. The method according to claim 4 wherein said shape is a triangle function

6. The method according to claim 5, wherein said triangle function is configured so that the triangle has a width corresponding to the width of said first half-wave.

7. The method according to any one of claims 2 to 6, wherein said first wave or half-wave is detected as a predetermined increase in the secondary voltage derivative.

8. The method according to any one of the preceding claims, wherein the timing of said predetermined waveform pattern, of a local extremum of said predetermined waveform pattern defines the closing time (t_c) of said control valve.

9. The method according to any one of the preceding claims, said valve indicator is used for injection control only if it exceeds a predetermined threshold.

10. The method according to any one of the preceding claims, wherein said valve indicator is used to monitor the fuel viscosity.

1 1 .The method according to claim 10, wherein an injection pulse width, injector needle opening delay or injector needle closing delay is corrected by an offset depending on fuel viscosity.

12. The method according to any one of the claims 1 to 8, wherein said valve indicator is used to monitor control valve lacquering.

13. The method according to any one of the claims 1 to 8, wherein said valve indicator is used to determine the minimum delivery pulse of the control valve, in particular by repeating the routine of steps a) to d) with increasing pulse width, starting from a non-injecting pulse width, until said valve indicator exceeds a predetermined threshold, and computing the minimum deliv- ery pulse based on the difference between the pulse width and closing delay.

14. The method according to any one of the claims 1 to 9, wherein the timing of the fully open control valve is determined, and the time period for opening the control valve is calculated based on the difference between the timing of fully open control valve and the minimum delivery pulse.

15. The method according to claim 14, wherein an injection pulse width is adapted based on a map having as input the time period for opening the control valve.

16. The method according to any one of the claims 1 to 9, wherein said valve indicator is used to determine the minimum delivery pulse of the control valve, wherein the difference between the minimum delivery pulse at a reference and current voltage is used to correct the injection pulse width for voltage changes.

17. The method according to any one of the claims 1 to 9, wherein an injection pulse width is corrected for injector drift based on a difference between current and reference closing delays.

18. The method according to any one of the preceding claims, wherein said step c) involves processing said recorded voltage data to compute filtered secondary derivative voltage data, and identifying said predetermined waveform pattern from said filtered secondary derivative voltage data.

19. The method according to claim 18, wherein said filtering involves a low pass filter,

20. The method of claim 19 where said low pass filter is a weighted moving average calculation. A system for controlling fuel injection in an internal combustion engine having at least one cylinder with an associated electromagnetically actuated fuel injector, wherein said fuel injector has a valve group having a needle slideably arranged therein in order to control at least one spray orifice through its displacement; a control chamber arranged to be filled with high pressure fuel so as to exert, at least indirectly, a pressure force on said needle in its closing direction, an outlet path of said control chamber being controlled by a control valve operated by an electromagnetic actuator with a coil; wherein the system comprises a programmable computing device configured for implementing the method according to any one of the preceding claims.