GB2566919A

GB2566919A - Method of determining the closing response of a solenoid actuated fuel injector

Info

Publication number: GB2566919A
Application number: GB1710797.0A
Authority: GB
Inventors: L P Michel Eric
Original assignee: Delphi Automotive Systems Luxembourg SA
Current assignee: BorgWarner Luxembourg Automotive Systems SA
Priority date: 2017-07-05
Filing date: 2017-07-05
Publication date: 2019-04-03
Also published as: GB201710797D0

Abstract

A method of determining the closing time of a solenoid actuated valve of a fuel injector comprising, in an operational cycle, during the closing phase, by a) sampling the value of voltage across the solenoid of said valve, b) plotting the voltage of step a) against time, c) with respect to each sampling point on a plot determined from step b), determining the angle a subtending a first line and a second line, said first line connecting the sampling point and the preceding sampling point, and said second line connecting the sampling point and the subsequent sampling point, d) determining the sampling point where the said angle is at a maximum or minimum, e) determining the closing time from the time of the sampling point found in step d). Preferably step d) comprises plotting the angle with respect to said sampling point against time.

Description

METHOD OF DETERMINING THE CLOSING RESPONSE OF A

SOLENOID ACTUATED FUEL INJECTOR

FIELD OF THE INVENTION

This disclosure relates to controlling fuel injectors. It has particular application to solenoid actuated fuel injectors, and ensuring such fuel injectors are operated and calibrated correctly, and that corrections in the control of injection to compensate for variations and wear, are checked. In particular it relates to a method of determining the closing response i.e. the closing time of a fuel injector valve by accurately determining the point of inflection from a solenoid current/voltage trace of a solenoid operated actuator of a fuel injector.

BACKGROUND OF THE INVENTION

Solenoid actuated fuel injectors typically are controlled by pulses sent to a solenoid controlled actuator of a fuel injector which act to open a fuel injector valve and allow fuel to be dispensed. Such actuators act to displace (via the armature of the actuator) a pintle and needle arrangement of the valve, to move the needle away from a valve seat. In such a state, the valve is open and when the pulse falls there is no power to the actuator and the valve is forced to a closed position.

Pulse profiles may vary and may comprise a series of pulses to operate the solenoid. There may be an initial activation (boost) pulse, provided in order to start to move the needle away from the valve seat, thereafter the pulse and thus power to the actuator is reduced - so therefore after a short while this may be followed by a “hold” phase where a reduced level of power is applied to keep the valve in the open position. These pulses may be regarded as fueling pulses. Thereafter the pulse and this voltage is reduced to close the valve. This may be followed by one or more braking pulses which act to slow the movement of pintle and needle when closing. So to recap, in order to allow a robust opening, solenoid driven (e.g. gasoline direct) injectors are typically powered up with a slight excess of electric energy to the solenoid coil. The coil is energized in a first phase with a boost voltage to accelerate the armature from close to open. Typically such first phase is followed by a second well defined energy supply (or “hold”) phase, which is characterized to hold the reached open position of the valve for a desired time.

A development trend is to reduce the time from close to open and vice versa of the solenoid driven valve and imitate the performance of competing piezo driven injector valves at significant lower cost. The objective is to dispense precisely lower fuel mass quantities. At very low fueling instances the solenoid driven valve operates in a so called transitional mode as opposed to ballistic or linear mode, which means that the valve will not settle in open position but moves partially towards closing prior to reach steady-state open position. If the closing is initiated during such bouncing it causes dynamically varying closing speeds of the pintle and armature. As a consequence it causes non-linear fueling in relation to the stimulus. Furthermore speed depending dynamic friction is considered to be one cause of accelerated wear, stimulating observable stick-slip effects of the moving parts and can be caused by varying closing speeds stimulated by bouncing. Likewise it comprises insuperable part to part variations if not addressed with significant computational efforts (ICLC). These prior art apparatus have significant part to part fuel variations during this so called transitional phase which limits the usability under these conditions and draw thereby a distinct line of differentiation to competing (piezo) injector propulsion technologies. The technical aspect to address is to control the supply driving schedule of the coil and thereby the speed of the armature and pintle during the transition from close to open and thereby reducing the momentum for bouncing.

As mentioned solenoid operated fuel injectors work on the principle of sending an injection pulse to a solenoid actuator so as to activate it, which in turn controls the opening and closing of valves and hence allows fuel to be injected. The longer the pulse, generally the longer the valve is open and the more fuel (under pressure) is injected in the combustion chamber. An important parameter in the control is the Minimum Delivery Pulse (MDP), that is the minimum length of solenoid drive pulse which will ensure that some fuel is injected. Obviously if the pulse is too small the solenoid actuator will not open enough or long enough for before closing the valve and so no fuel to be injected.

A further important parameter for controlling fuel injectors is thus the Closing Response of the injector which is the time taken for the injector valve to close (i.e., the time when the valve member such as pintle moves back to its closed position after opening), from the time of the end of the activation pulse. Thus methodology has been used to determine the closing time of the actuator valve by detecting the point of inflection in the voltage trace across the solenoid of the actuator.

A signal of the voltage across the injector can be provided to by the ECU to activate it. After the end of the command pulse, when there is no voltage applied by the high power driver, the voltage variation is then only caused by injector needle/armature movement. The inflexion point in the measured voltage trace corresponds to the injector closing (needle touching its seat). So the closing time of said valve can s determined from the point of inflection of said measured voltage trace.

The target on fuelling accuracy is every more demanding to provide CR detection which is very accurate, robust and repeatable, on the whole injector operating area.

The current CR detection method presents several downsides: the specific calibration required for each ECU interface is time consuming. Further specific calibration is required for test rig data which is time consuming and the calibration is often a compromise. The low pass filters typically used introduce some delay and thus an offset in the detected CR = not good when comparing to absolute hydraulic reference data.

It is an object of the invention to provide an method of determining the closing time/closing response by providing a more accurate way to determine the point of inflection on the voltage trace due to valve closing. Aspects improved CR detection accuracy and robustness/repeatability should be significantly improved.

SUMMARY OF THE INVENTION

In one aspect is provided a method of determining the closing time of a solenoid actuated valve of a fuel injector comprising, in an operational cycle, during the closing phase, a) sampling the value of voltage across the solenoid of said valve;

b) plotting the voltage of step a) against time; c) with respect to each sampling point on a plot determined from step b), determining the angle a subtending a first line and a second line, said first line connecting the sampling point and the preceding sampling point, and said second line connecting the sampling point and the subsequent sampling point, d) determining the sampling point where the said angle is at a maximum or minimum;

e) determining the closing time from the time of the sampling point found in step

d).

Step d) may comprises plotting the angle with respect to said sampling point against time.

The angle a at sampling point η (a_n) may be determined from the following equation:

a_n = 180 + arctan (y_n+i - y_n)/(x_n+i-Xn)*Ts -arctan (y_n- y_n-i)/(xn-Xn-i)*Ts where x_n and y_n are the plot Cartesian co-ordinates of a sampling point n; where x_n-i and y_n-i are the plot Cartesian co-ordinates of the sampling point preceding sampling point n; where x_n+i and y_n+i are the plot Cartesian co-ordinates of the sampling point following preceding sampling point n; and Ts is the sampling period.

180 + arctan (y_n - y_n-i)/(xn-Xn-i)*Ts - arctan (y_n+i - y_n)/(xn+i-x_n)*Ts where x_n and y_n are the plot Cartesian co-ordinates of a sampling point n; where x_n-i and y_n-i are the plot Cartesian co-ordinates of the sampling point preceding sampling point n; where x_n+i and y_n+i are the plot Cartesian co-ordinates of the sampling point following preceding sampling point n; and Ts is the sampling period.

BRIEF DESCRIPTION OF THE DRAWINGS

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

Figure 1 shows a typical activation pulse sent to a solenoid controlled fuel injector

Figure 2 shows a plot of the voltage across the solenoid in the closing phase enlarged and the area around the point of inflection

Figures 3 and 4 shows a plots of the voltage in the closing phase as well as the computed value of angle a , for different resolutions and signal conditions.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Figure 1 shows a typical activation pulse sent to a solenoid controlled fuel injector and shows the voltage 1 across the solenoid of a solenoid actuated fuel injector during a fuelling (operating ) cycle, consequent to a fuelling pulse sent from a controller such as an ECU. The parameter shown is voltage e.g. applied across the solenoid terminals. As can be seen there is an initial high activation or “boost” pulse 2. This pulse acts to provide the force needed to move/accelerate the armature/pintle arrangement away from its closed position to an open position. After this is a lower hold phase (pulse) 3 where a low voltage is applied to keep the valve in the open position. This can be provided by chopping. After this the voltage is reduced to zero and the voltage become negative. After this and the valve starts to close, shown by the trace in region A. During this time, the closing phase the voltage across the solenoid terminals decays. So after the current throught the solenoid (applied to open the valve) us switched off, the voltage falls sharply and then slowly rises again due to the induced voltage as the valve closes. As can be seen during the closing response time period a point of inflection can be seen as the valve closes; this is caused by the valve member/actuator member contacting the solenoid armature hits a stop. In methodology according to the invention, the position/point of inflection is accurately determined. Since the closing response corresponds to a specific shape in the voltage trace, aspects of the method are is based on signal shape recognition rather than signal filtering and derivatives.

Figure 2 shows a plot of the voltage 1 across the solenoid in the closing phase enlarged and the area around the point of inflection and shows consecutive sampling points Al, A2, A3, A4, A5, A6. At each sampling point, two join lines 5a, 5b are plotted, one 5a between the sampling point and the immediately preceding sampling point and the other 5b between the sampling point and the sampling point immediately following. These two (connecting) lines can be regarded as a virtual scissor, articulated on each measurement point. The angle between (i.e. subtending) the lines is then determined; this process is repeated at each sampling point, and this subtending angle (opening or scissor angle) is plotted with time; i.e. plotted with to each sampling point.

So with respect to the sampling point A2 the scissors comprises the straight lines A1-A2 (5a) and A2-A3 (5b), the angle therebetween is oiA2. With respect to point A5 the scissors comprises two straight lines A4-A5 and A5-A6 and the angle therebetween is aA5. The angles a are determined plotted with respect to each sampling point.

In the above examples the angle subtending the connecting lines is that below the graph. In an alternative embodiment the angle β that is the ange subtending the lines above the plot may alternatively be recorded and processed similarly. This angle is effectively 360-a.

Figure 3 shows a plots of the voltage in the closing phase 1 as well as the computed value of angle a , for different resolutions (i.e. different scissor lengths

i.e. connecting line sizes) denoted with reference numeral 6a and 6b against sampling point (which is effectively equivalent to time). The plot may be formed by joining discrete values of a, or forming a curve to fit the values of a. The plots shows the values of a with respect to two different scissors lengths which is equivalent to sampling frequency /resolution. As mentioned also shown is the plot of voltage against time from which the angles a are calculated. As can be seen the value of angle a reaches a maximum at the point of inflection, at time = Tm.

Thus by plotting the angles determined as described above an accurate point of inflection can be determined, Thus to recap the opening angle of the scissor peaks at the location of the injector closing (inflexion on the voltage trace), thus providing directly the desired information.

It is to be re-iterated noted as mentioned that alternative to the plotting the angle a subtending the two joining lines (scissors) with below the plot of sample points, the angle β subtending the join lines (scissors) above the plots may be determined (where β = 360°-α). In this case when β is plotted against time the value thereof reaches a minimum when at the point of inflection.

Figure 4 shows a similar plot to figure 3 for the value of a (again at two different resolutions denoted 7a and 7b) with high frequency sampling and a noisy voltage trace. As can be seen even here the method allows god accuracy in determining the point of inflection.

The scissor length (sampling interval therefore) may be adapted to reject high frequency signal noise and preserve only the relevant information.

When determining the angle a, since the X- and Y- axis units are different, this is not a geometrical angle as one would compute in a 2D Cartesian system where the X- and Y- units are identical. So some special care may in cases be necessary to compute this angle

Angle &A2 = 180 + atan ( ^{yA3 yA2}—) — atan ( ^{yA2 yA1}—) ° \(xA3-xA2)*Ts/ \(xA2-xAl)*Ts/ where Ts is the sampling period [s], atan is the arctangent function. This may be provided by an approximation from a 2D look-up table for easy implementation in an electronic control module.

In general the equation may be stated:

The angle may be alternatively stated as β_η = 180 + arctan (y_n - y_n-i)/(xn-Xn-i)*Ts - arctan (y_n+i - y_n)/(x_n+i-Xn)*Ts

For improved accuracy, a second smaller scissor (hence more reactive to shape, but also more sensitive to noise) provides a secondary information for improved peak location detection. The location of the first scissor angle peak is used to define a limited search area for the peak of the second scissor angle. This addresses the issue when having several peaks due to weak voltage inflexion on some injectors.

An advantage over the existing method is it gives improved accuracy and robustness over previous methods. There are only a few parameters to calibrate, such as e.g. scissor lengths and thresholds. Results showed that some generic values can be applied to all data, regardless of sampling frequency, this results in significant time e saving in calibration and validation phase. There is no compromise between low pulse width and high pulse width CR detection performance. The method is also easy to implement in ECUs. And does not need several Low Pass filters to compute derivatives as is performed currently.

Claims

CLAIMS:

1. A method of determining the closing time of a solenoid actuated valve of a fuel injector comprising, in an operational cycle, during the closing phase,

a) sampling the value of voltage across the solenoid of said valve;

b) plotting the voltage of step a) against time;

c) with respect to each sampling point on a plot determined from step b), determining the angle a subtending a first line and a second line, said first line connecting the sampling point and the preceding sampling point, and said second line connecting the sampling point and the subsequent sampling point,

d) determining the sampling point where the said angle is at a maximum or minimum;

e) determining the closing time from the time of the sampling point found in step d).

2. A method as claimed in claim 1 wherein in step d) comprises plotting the angle with respect to said sampling point against time.

3. A method as claimed in claims 2 where the angle a at sampling point n (an) is determined from the following equation:

an = 180 + arctan (yn+i - yn)/(xn+i-Xn)*Ts -arctan (yn - yn-i)/(xn-Xn-i)*Ts where xn and yn are the plot Cartesian co-ordinates of a sampling point n; where xn-i and yn-i are the plot Cartesian co-ordinates of the sampling point preceding sampling point n;

where xn+i and yn+i are the plot Cartesian co-ordinates of the sampling point following preceding sampling point n; and

Ts is the sampling period.

4. A method as claimed in claims 2 where the angle a at sampling point n (an) is determined from the following equation:

180 + arctan (yn - yn-i)/(xn-Xn-i)*Ts - arctan (yn+i - yn)/(xn+i-xn)*Ts where xn and yn are the plot Cartesian co-ordinates of a sampling point n; where xn-i and yn-i are the plot Cartesian co-ordinates of the sampling point preceding sampling point n;

where xn+i and yn+i are the plot Cartesian co-ordinates of the sampling point

10 following preceding sampling point n;

and Ts is the sampling period.