Method of determining a closing time of a needle valve of a fuel injector and controlling the injector

Abstract

A method of determining a closing time of a needle valve of a fuel injector and controlling the injector. The method comprises: a) providing a reference signal (7); b) activating the injector by sending an activation pulse to the solenoid and measuring a voltage signal on the solenoid starting from a point in time after the end of the activation pulse to provide a solenoid voltage signal (8); c) determining a difference signal (9) based on the difference between the signal of step a) and the signal determined in step b); d) differentiating the difference signal to provide a first derivative signal (11); e) providing a normalized signal from step d); f) differentiating the signal of step d) to provide a second derivative (17); g) multiplying the values of the signals from steps f), e) and c) to provide a curve (19); h) determining the maximum peak of the curve or a filtered version and determining the valve closing time from the point in time of the maximum peak; i) subsequently controlling the fuel injector based on the result of step h).

Description

Technical Field

[0001] This invention relates to a method for determining the closing time of a needle valve in a fuel injector. Therefore, this invention relates to determining, for example, the point in time when the needle of a fuel injector's needle valve contacts the valve seat to prevent fuel from flowing into the combustion chamber. Background Technology

[0002] Modern fuel injectors typically use electric actuators (such as piezoelectric or solenoid-operated actuators) to operate or control a needle valve, which opens and closes to distribute fuel into the combustion chamber via the movement of a needle away from its seat. Generally, an activation pulse of a certain duration (pulse width) is sent to the electric actuator to operate the fuel injector. The amount of fuel injected into the combustion chamber depends on the pulse duration. Fuel injectors can be of the type where the actuator directly removes the needle from the valve seat; or against, for example, a bias spring device; this is called a direct injector, and such injectors are used for both gasoline and diesel fuel.

[0003] In alternative designs, many modern fuel injectors are hydraulically operated, where the actuator does not directly actuate the needle. Instead, the actuator operates a valve (system) to control the pressure in the fuel injector, thereby indirectly removing the needle from the valve seat and selectively dispensing fuel. Therefore, in such injectors, there is a distinction between the opening and closing of the actuator-operated valve and the opening and closing of the needle.

[0004] The characteristics of the fuel injector vary over time. Therefore, a closed-loop control and compensation strategy is required. Typically, this is achieved by implementing a learning strategy to compensate the injector over time, where the behavior and characteristics of the fuel injector are learned over time to calculate correction values ​​with respect to, for example, the duration of the activation pulse, and these correction values, or "fine-tuning," are applied during real-time injector operation. This strategy is commonly referred to as ICLC (Injector Closed-Loop Compensation).

[0005] The closing delay CD (or alternatively referred to as the closing response time CR) of a solenoid-operated fuel injector (such as a direct-acting gasoline injector or a hydraulic fuel injector) is defined as the time between the end of the activation pulse sent to the solenoid of the solenoid actuator and the needle closing time (i.e., when the valve needle reaches the valve seat to prevent fuel flow). Knowing this parameter is important for feedback control of the fuel injector. This is determined by determining the needle closing time (NCT); the needle closing time can be another important parameter in a feedback control system.

[0006] The needle closure time (NCT) can be determined by analyzing the differential voltage on the injector solenoid (i.e., the solenoid of the solenoid actuator). When the injector's needle strikes the needle valve seat during closure and closing, the voltage slope of the actuator solenoid coil changes and can be observed as a "false signal" in the time curve: therefore, the false signal time occurs during the needle closure time, and the time between this false signal and the end of the pulse is called the closure response (CR) or closure delay.

[0007] False signals are inflection points, etc. Therefore, in conventional systems, to determine the timing of false signals, the derivative method is used to differentiate the voltage on the solenoid; and the first (and / or preferably the second) derivative of the voltage signal is analyzed to more easily detect and identify false signals, thereby more easily detecting and identifying NCT. A more refined method uses trigonometric methods to detect large angle changes (in the voltage curve, this requires less ECU calculation time than the initial method).

[0008] The problem lies in the high measurement / estimation gap of the closure response (i.e., the time until valve closure) under low fuel supply, leading to poor response in accurate control. Therefore, it is necessary to accurately determine the NCT. However, existing methods are poorly robust when it comes to detecting small spurious signals. Low-amplitude spurious signals can be due to injector design, injection pressure, or environmental conditions such as temperature and fluid characteristics. Furthermore, for certain injector definitions (e.g., M14), there may be two consecutive events. For example, in an M14 injector design, there are two physical events: a first event occurs when the needle contacts the seat, and a second event occurs as the armature continues its movement toward the stop ring. Both produce voltage inflection points of similar amplitude, and the detection criteria are diffuse, as either the first or second event can be detected in a shot-to-shot sequence. Summary of the Invention

[0009] In one aspect, a method is provided in a solenoid-controlled fuel injector comprising a solenoid-controlled actuator, for determining the closing time of a needle valve when the valve needle contacts the valve seat during a duty cycle of the fuel injector, and subsequently controlling the injector based on the closing time, the actuator being adapted to control the needle valve by movement of the needle towards and from the valve seat of the needle valve, the method comprising the steps of:

[0010] a) Provide a reference bend / signal Ref(7);

[0011] b) The injector is activated by sending an activation pulse to the solenoid, and the voltage signal on the terminals of the solenoid is measured from the time point after the end of the activation pulse to provide a solenoid voltage signal Raw signal (8).

[0012] c) Based on the difference between the signal in step a) and the signal determined in step b), determine the difference signal Delta_signal(9);

[0013] d) Differentiate the signal obtained in step c) to provide a first differential signal Diff1(11);

[0014] e) Based on the value provided in step d), the signal Diff1_Normalised decreases as the value of Diff1 moves further away from zero.

[0015] f) Differentiate the curve / signal from step d) to provide the second derivative of Delta_signal, Diff2(17);

[0016] g) Multiply the values ​​of the curves / signals from steps f), e), and c) to provide another curve, Mix(19);

[0017] h) Determine the maximum peak value of the curve from step g) or a filtered version of the curve, and determine the valve closing time based on the time point of the determined maximum peak;

[0018] i) The fuel injector is then controlled based on the result of step h).

[0019] The reference bend / signal Ref(7) can represent the voltage on the solenoid of the injector or reference injector relative to time during a work cycle in which no injection occurs.

[0020] The reference bending / signal Ref(7) can be learned by activating the injector under test without injection, wherein the pulse length in the activation is less than the minimum drive pulse MDP.

[0021] The method may include: determining a cutoff time t after the end of the activation pulse. co And only execute the method steps after that time, or set the values ​​of Diff1Normalised, Diff2, or Delta to zero before that time.

[0022] The method may include the intermediate step of providing a curve / signal Diff1Sat, which includes a curve of Diff1, wherein any value of Diff1 less than a threshold X is set to -X, and wherein step e) includes providing a signal Diff1_Sat Normalised, wherein the further the value of Diff1_Sat is from zero, the more Diff1_Sat Normalised decreases.

[0023] The threshold X can be determined based on the value of the peak positive value of curve Diff1.

[0024] The value of step e) is preferably a dimensionless value.

[0025] The value is determined based on the following formula:

[0026] |(Z–D1(t))| / Z or |(D1(t)-X)| / X

[0027] Where D1 represents Diff1(D1) or Diff1Sat(D1S),

[0028] |(Z–D1S(t))| / Y or |(D1S(t)-Z)| / Y

[0029] Z and Y are arbitrarily chosen values.

[0030] Following step c), the difference signal Delta_signal(9) can be filtered before processing in subsequent steps.

[0031] Following step d), the first differential signal Diff1(11) can be filtered before processing in subsequent steps.

[0032] The phrase “activate the injector” can be interpreted as the work cycle of implementing the injector. Attached Figure Description

[0033] The invention will now be described by way of example with reference to the accompanying drawings, in which:

[0034] - Figure 1 Simplified curves are shown of the time of the activation (logic) pulse sent to the solenoid of the fuel injector activated by the solenoid and the fuel injection rate (i.e., from the needle valve).

[0035] - Figure 2a The timeline of voltage 3 on the solenoid of the actuator is shown;

[0036] - Figure 2b A similar curve with a low voltage signal and a low amplitude spurious signal is shown.

[0037] - Figures 3 to 7 The curves obtained in the steps of a method according to an example are shown.

[0038] - Figure 8 The overall process is outlined based on an example. Detailed Implementation

[0039] Figure 1A simplified curve of activation (logic) pulse 1 sent to a solenoid-activated fuel injector, including a needle valve known in the art, is shown as a function of time, and curve 2 shows the fuel injection rate (i.e., from the needle valve). The pulse width (PW) of the activation pulse is shown; it begins at t1 and ends at t3. The opening and closing of the needle valve is generally hysteretic; it begins / ends from the activation pulse, respectively. The needle valve opens at point t2 and closes at point t4; therefore, fuel is injected between these times; HO refers to the time between these times, which is called the hydraulic opening or needle valve opening time, during which fuel is injected into the combustion chamber. The closing response (CR) (often alternatively called the closing delay (CD)) is the time between points t3 and t4. t4 is the needle closing time (NT). The opening delay (OD) is between t1 and t2, from the start of the activation pulse to the start of the needle valve opening. This parameter is important for control. The total operating duration (TOD) is defined as the time between the start of the opening pulse t1 and the valve closing time t4 (closing time CT).

[0040] Figure 2a The timeline of voltage 3 on the actuator solenoid is shown, starting from time t3 after the end of the activation pulse. It can be seen that a spurious signal is observed in region 4, where the needle valve is assumed to be closed, i.e., the end of the closing delay at the needle closing time (NCT). Figure 2b A similar curve is shown, featuring a low voltage signal 5 and a low amplitude spurious signal 6.

[0041] The terms “plot,” “signal,” and “curve” are interchangeable and are used interchangeably, meaning they refer to the same thing.

[0042] invention

[0043] The steps involved in implementing embodiments of the present invention are described in detail below. As will become clear later, in the examples, not all of these steps need to be performed and may be in a different order.

[0044] Step 1

[0045] In an example of the present invention, a reference solenoid voltage signal (i.e., a time curve) is first provided. The reference signal (Ref) or curve can be stored in a MAP, for example, in an engine ECU known in the art. The reference signal can represent the voltage appearing on the actuator solenoid of the nominal / standard / reference fuel injector based on a small pulse without injection. Thus, the reference bend / signal Ref (7) is learned by activating the injector under test without injection, in which the pulse length is less than the minimum drive pulse (MDP). For example, a typically very short PW, such as 100 μs when the MDP is approximately 200 μs on the current injector.

[0046] Therefore, the voltage is the difference between the low-side drive and high-side drive of the injector's traction pin. It typically comprises a steeply sloping curve whose gradient becomes shallower over time. In other words, the curve can be one where the voltage rises with a decaying gradient (from a base level (e.g., zero) over time) (i.e., reaching a near-asymptotic value). This curve can be learned by measuring and recording the voltage signal after activating the fuel injector, preferably the one being tested, to have an individual reference for each injector as needed, rather than a universal reference map / curve. This curve (Ref) is... Figure 3 It is shown as reference numeral 7 in the attached figure.

[0047] Step 2

[0048] The injector under test is activated with an activation pulse; that is, the characteristics of the injector will be determined for use in closed-loop control. This activation pulse can differ from the length / duration used to obtain the reference bend. In practice, for the reference bend, we need to apply a pulse with a very small pulse width (PW) without fuel, for example, 100 μs (MDP, for example, about 200 μs). For step 2, it can be performed on any pulse on which the closing response / closing time is desired, thus ranging from the smallest drive pulse to a larger PW, and obtaining / determining, for example, recording the voltage signal on the solenoid of a solenoid actuator. The curve can begin at a (pre-)set time after the end of the activation pulse. It is in Figure 3 It is shown as curve / signal 8 and is designated as "Raw signal".

[0049] Step 3

[0050] In the next step, the reference bend 7 (value) is subtracted (or vice versa) from the actual measurement (voltage) of signal 8 determined in step 2 to provide a differential voltage signal / curve, referred to as "delta_signal" or D1, as shown below. Figure 4 As shown in curve 9.

[0051] Therefore, the difference between curves (signal 7 and signal 8) is determined to find curve 9, which is Delta_signal = Raw_signal(8) - Ref(7) (reference signal); that is, subtracting each other. The resulting curve 9 can preferably be filtered to provide curve 10 (filtered delta signal 10).

[0052] Either or both of the reference curve / signal 7 and the actual curve / signal 8 (Ref signal, Raw signal) can be appropriately scaled to match and have the same starting reference time point, which can be any time after the end of the activation pulse (t3).

[0053] Therefore, in retrospect, subtracting the curves from each other (order irrelevant) yields a curve of the difference D1 (“delta_signal”), which is further analyzed / processed in the following steps to determine the needle valve closing time. Figure 3 The measured (current) voltage on the solenoid is shown (the curve starts at or shortly after the end of the activation pulse, or is generally measured as the start)8 (which can be considered as...). Figure 2b The curve 5 is equivalent to the reference curve (signal / curve) 7. This reference signal or curve can be stored, for example, in the MAP mentioned above, and can be stored as a series of values; or as a formula. Methods for storing valid reference curves / signals are known to those skilled in the art and will not be described in further detail. The signal Delta_signal(D1) is further processed as follows:

[0054] Step 4

[0055] In this step (step 4), the signal 9 or the filtered delta signal 10 with respect to the difference (Delta_signal) D1 is differentiated relative to time to obtain the signal Diff1. Figure 5 It is shown by reference numeral 11 in the attached figure. It can also be filtered to produce a filtered differential signal 12 (filt Diff1). Figure 5 In the diagram, the y-axis unit is, for example, mV / s. Note that the unit indicates a step size of 1 mV / s. It can be seen that at point 13, the peak value of Diff1 (relative to the maximum value) is close to 1 mV / s. It initially has a sharp negative peak at 14.

[0056] Step 5 (optional)

[0057] Based on signal Diff1 (or filt Diff1), another signal called Diff1_sat is calculated; Figure 6 It is shown in Figure 15. It is the same as Diff1, except that the first part of the Diff1 curve is truncated because it is set to zero, or actually at the cutoff time t. co The previous curve was set to zero (which effectively means that the method ignores this period in advance to exclude anomalies caused by the initial sharp negative peak 14).

[0058] In one approach, to calculate this curve, a value "X" is chosen. This value (in mV / s) can be a nominal value, but should be positive and generally higher than the positive peak value 13 obtained in curve 11 / 12; that is, higher than Figure 5 The peak value of Diff1 is at point 13. In this case, the value X = 1 is chosen.

[0059] This value is used to provide a "saturation effect" on the curve Diff1 to produce the curve Diff1_Sat; this is to eliminate, for example, the abrupt change in gradient that occurs in the first time portion of the curve Diff1, as shown at point 14. This continues until the cutoff time point t. co In fact, it is less than -X (i.e., until time point t). co Any value of ) is set to -X.

[0060] Therefore, in bending 6, the calculated value of Diff1_Sat below the negative nominal value X is set to -X. Therefore, from t=0 to t=t co Between these points, the value of Diff1_Sat is set to -X. Therefore, in summary, the focus of signal analysis is on the zero-crossing of Diff1. Diff1 then saturates around 0 at + / -X. This signal is now referred to as "Diff1_sat".

[0061] Therefore, in retrospect, the value X is appropriately chosen by the technician, and it can be reasonably arbitrary and vary upwards from zero, but preferably in the region of the maximum value in the Diff1 curve, thus at point 13, and preferably slightly above it. It should be remembered that the curve can vary with different activations. If multiple curves are considered, with an envelope, it may be preferable that the maximum value of Diff1 never rises to near the chosen X value.

[0062] Step 5 is optional and is implemented to ignore the effects of large negative values ​​that may occur at the beginning of a time window some time after the activation pulse. In an alternative to overcome (i.e. ignore) these effects, this can be achieved simply by considering only time point t. co The method steps (i.e., previous and remaining method steps) are then executed in place of this step; that is, a time window that has passed the large negative peak is selected for the overall processing / method. Technicians are able to determine a time window sufficient to ensure that the large negative peak has passed (i.e., effectively eliminated). co value.

[0063] Step 6

[0064] In the next step, the curve / signal Diff1_Sat(D1S) [or t] coThe signal Diff1(D1) is then converted to an absolute signal value and normalized around 0 to produce a normalized curve signal Diff1_Normalised(D1N) (whose value decreases as the value of Diff1 / Diff1_sat moves further away from zero). When step 5 is performed, Diff1Normalised can be referred to as Diff1Sat Normalised(D1SN).

[0065] This can be achieved by setting 100% [or 1] if |D1| = 0, and Diff1_normalised to 0% [or 0] if |D1| = X, exhibiting linear behavior between 0 and X.

[0066] And that means that if |D1S|=0, then 100%, and if |D1S|=X, then Diff 1SatNormalised(D1SN) is 0%, exhibiting linear behavior between 0 and X.

[0067] It should be noted that in the claims, references to Diff1_Normalised should be interpreted as above, for example, the signal Diff1_Normalised includes positive values ​​whose values ​​decrease as the value of Diff1 or Diff1_Sat moves further away from zero.

[0068] Therefore, the output signal in this step is called Diff1_Normailised, which will be used in the final multiplication expression; and when step 5 is executed, the output is Diff 1Sat Normalised (D1SN); Figure 7 It is shown in the figure with reference numeral 16.

[0069] In other words, at time point "t", the percentage or fraction value Diff1_Normalised or Diff1_Sat Normalised (DISN) is determined by the following formula:

[0070] |(Z–D1(t))| / Z or |(D1(t)-X)| / X

[0071] Where D1 represents Diff1(D1) or Diff1Sat(D1S).

[0072] Therefore, in this example, the value varies from 0% to 100% (0 to 1 as factors).

[0073] This formula is not exhaustive; those skilled in the art can conceive of other formulas. A more general formula could be:

[0074] |(Z–D1(t))| / Y or |(D1(t)-Z)| / Y

[0075] Z and Y can be any values. Therefore, the factor can range from 0 to 2 (200%).

[0076] The result is a curve that produces a dimensionless (factor) value, the magnitude of which decreases depending on the absolute difference between zero and the value of Diff1 / Diff1_Sat. In the penultimate step, there is a multiplication process using this factor, and the shape of the subsequent curve is analyzed, so the ratio of the Diff1_Normailised / Diff1_sat_normalised (D1_SN) curve is not important.

[0077] Step 7

[0078] In this step, the differential curve / signal (Diff1)11 (or the filtered curve 12) is then further differentiated relative to time to determine Figure 8 The second derivative signal / curve (Diff2) shown in Figure 17 can be optionally filtered to provide curve 18.

[0079] Step 8

[0080] Then there exists a combined processing method to provide curve Mix(19) by multiplying signals Delta_signal(9 / 10), Diff2, and Diff1_Normailised / Diff1_Sat Normalised(D1SN), see Figure 8 In other words, the signal values ​​at the same time point are multiplied to generate another signal curve 19.

[0081] In other words, generating this curve Mix involves multiplying the signal values ​​Delta_signal (9 / 10), Diff2, and [Diff1_Normailised / Diff1 or Sat Normalised (D1SN)] at the same time point.

[0082] Mix=Delta_signal(D1)*Diff 1_Normailised(D1N)*Diff2

[0083] or

[0084] Mix=Delta_signal(D1)*Diff 1Sat Normalized(D1SN)*Diff2

[0085] It should be noted that any of the parameters being multiplied can be a filtered version of the signal.

[0086] Step 9

[0087] Analyze the signal / curve Mix19 to find its maximum value, i.e., by identifying the highest peak (which is peak 20 in curve 19). This peak occurs at the "tg" time when a false signal appears and represents the needle closure time.

[0088] Figure 8 Some of the steps described above are illustrated.

[0089] other

[0090] As described above, preferably, the method only starts from time point t. co Execution; that is, selecting a time window that has passed 100 seconds past the large negative peak for the overall processing / method. Technicians can determine a time window sufficient to ensure that the large negative peak is passed through (i.e., effectively eliminated). co value.

[0091] other

[0092] It should be noted that the steps for providing D1S actually state the process of multiplying the values ​​of Delta_signal(9 / 10), Diff2, and Diff1_Normailised(DIN) only at time t. co Then it will proceed, at time t. co It is the chosen cutoff time that is desired to occur after the initial negative peak. This is because at T... co The previous value of Norm_Diff_Sat was set to zero, so the product from the multiplication will also be zero.

[0093] In other words, in the preferred example, the multiplication of values ​​only occurs in t. co After it ended.

[0094] The method of multiplying the values ​​can preferably be performed at another cutoff time (t) within the time window of the desired false signal. co2 This should occur before, and preferably far from, that time, because abrupt changes are caused by poor bending, not by needle closure.

[0095] As a result, the robustness and determinism of CR are greatly improved compared with previous methods:

[0096] The required protection method can now be used to determine the CR of an injector with a very flat spurious signal.

[0097] In addition, the reference shape can be derived from a general shape (stored in the ECU) or can be measured directly on an injector-by-injector basis, thus providing a separate and adaptive reference for each component.

[0098] In a region with two consecutive low events, the first low event is consistently detected using this method.

Claims

1. A method for determining the closing time of a needle valve of a fuel injector and controlling the fuel injector based on the closing time, the fuel injector being a solenoid-controlled fuel injector comprising an actuator controlled by a solenoid, the actuator being adapted to control the needle valve by movement of a needle toward and from a valve seat of the needle valve, and the closing time of the needle valve being when the needle contacts the valve seat during a duty cycle of the fuel injector, the method comprising the steps of: a) Provide a reference signal (7), wherein the reference signal (7) represents the voltage on the solenoid of the fuel injector or reference injector relative to time during a work cycle in which no injection occurs; b) The fuel injector is activated by sending an activation pulse to the solenoid, and the voltage signal at the terminals of the solenoid is measured from the time point after the end of the activation pulse to provide a solenoid voltage signal (8). c) Determine the difference signal (9) based on the difference between the reference signal in step a) and the solenoid voltage signal determined in step b). d) Differentiate the difference signal from step c) to provide a first differential signal (11); e) Based on the first differential signal from step d), provide a standardized first differential signal, wherein the value of the standardized first differential signal decreases the further the value of the first differential signal is from zero; f) Differentiate the first differential signal from step d) to provide a second differential signal (17) of the difference signal. g) Multiply the values ​​of the signals from steps f), e), and c) to provide a blending curve (19); h) Determine the maximum peak value of the mixture curve from step g) or a filtered version of the mixture curve, and determine the closing time of the needle valve based on the time point of the determined maximum peak value; i) The fuel injector is then controlled based on the result of step h).

2. The method according to claim 1, wherein, The reference signal (7) is learned by activating the injector under test without injection, wherein the pulse length in the activation is less than the minimum drive pulse MDP.

3. The method according to claim 1 or 2, wherein the method comprises: Determine the cutoff time (t) after the end of the activation pulse. co The method steps are performed only after the deadline, or the value of the standardized first differential signal, the second differential signal, or the difference signal is set to zero before the deadline.

4. The method according to claim 3, wherein the method comprises the intermediate step of: providing a saturated first differential signal, the saturated first differential signal comprising a curve of the first differential signal, wherein, Any value of the first differential signal less than a threshold X is set to -X, and wherein step e) includes providing a standardized saturated first differential signal, the more the value of the saturated first differential signal is far from zero, the more the standardized saturated first differential signal decreases.

5. The method according to claim 4, wherein, The threshold X is determined based on the positive peak value of the curve of the first differential signal.

6. The method according to claim 1, wherein, The value of the standardized first differential signal in step e) is a dimensionless value.

7. The method according to claim 6, wherein, The value is determined based on the following formula: | (Z – D1(t)) | / Z or | (D1(t) - X) | / X Wherein, D1 represents the first differential signal or the saturated first differential signal, or | (Z – D1S(t)) | / Y or | (D1S(t)-Z) | / Y Where X is the positive threshold for the first differential signal, and Z and Y are arbitrarily chosen values.

8. The method according to claim 1, wherein, Following step c), the difference signal (9) is filtered before processing in subsequent steps.

9. The method according to claim 1, wherein, Following step d), the first differential signal (11) is filtered before processing in subsequent steps.

Images