METHOD FOR DETERMINING A CHARACTERISTIC VARIABLE OF A SOLENOID VALVE AND METHOD FOR TRAINING A PATTERN RECOGNITION METHOD BASED ON ARTIFICIAL INTELLIGENCE

Abstract

A method for determining a characteristic variable for opening and/or closing a flow-through opening of a solenoid valve, in which solenoid valve a solenoid coil is energized to raise an armature to open the flow-through opening for a fluid. During operation of the solenoid valve, a profile of a current in the solenoid coil being determined, and using a pattern recognition method based on artificial intelligence, the characteristic variable(s) is/are determined based on at least one section of the profile or a profile derived therefrom using a neural network. A method for applying and for training a pattern recognition method based on artificial intelligence are also described.

Claims

1-13. (canceled)

14. A method for determining a characteristic variable for opening and/or closing a flow-through opening of a solenoid valve, the solenoid valve including a solenoid coil which is energized to raise an armature for opening the flow-through opening for a fluid, the method comprising the following steps: determining, during operation of the solenoid valve, a profile of a current in the solenoid coil; and determining a characteristic variable using a pattern recognition method based on artificial intelligence, based on at least one section of the profile or on a section of a profile derived from the profile.

15. The method as recited in claim 14, wherein the characteristic variable is selected from: an opening point in time of the flow-through opening, a closing point in time of the flow-through opening, a value which indicates whether the flow-through opening has been opened or closed or not, and a value which indicates the probability with which the flow-through opening has been opened or closed.

16. The method as recited in claim 14, wherein an operation of the solenoid valve is carried out and/or adapted and/or a diagnosis is carried out, based on the determined characteristic variable.

17. The method as recited in claim 14, wherein the solenoid valve is used for introducing fuel into cylinders of an internal combustion engine or reducing agent or reducing agent solution into an exhaust tract of an internal combustion engine.

18. The method as recited in claim 14, wherein the solenoid valve is configured for metering or measuring the fluid into a volume.

19. The method as recited in claim 18, wherein the fluid is a liquid or a gas.

20. A method for training a pattern recognition method based on artificial intelligence, which is used for determining a characteristic variable for opening and/or closing of a flow-through opening of a solenoid valve, the solenoid valve including a solenoid coil which is energized to raise an armature for opening the flow-through opening for a fluid, the method comprising: for each of multiple profiles of a current profile occurring during operation of the solenoid valve in the solenoid coil or a profile derived from the current profile, supplying at least one section as an input value to the pattern recognition method based on artificial intelligence; and based on characteristic variables obtained for the sections from the pattern recognition method based on artificial intelligence as an output value and comparison values thereto, adapting the pattern recognition method based on artificial intelligence.

21. The method as recited in claim 20, wherein the multiple profiles of the current profile are selected for various values of parameters of the solenoid valve.

22. The method as recited in claim 20, wherein the multiple profiles of the current profile are selected for various values of parameters of the solenoid valve from input vectors which are provided via a remote processing and/or memory system.

23. The method as recited in claim 20, wherein the various parameters are selected from: operating temperatures of the fluid and the solenoid valve, runtimes and wear of the solenoid valve, a number of opening/closing cycles which have already taken place, activation voltages for the solenoid coil, spring strengths of springs which press against the armature, dimensions of the flow-through opening, a functionality of the solenoid valve.

24. The method as recited in claim 20, wherein the pattern recognition method based on artificial intelligence uses an artificial neural network or a “Support Vector Machine”.

25. A processing unit configured to determine a characteristic variable for opening and/or closing a flow-through opening of a solenoid valve, the solenoid valve including a solenoid coil which is energized to raise an armature for opening the flow-through opening for a fluid, the processing unit configured to: determine, during operation of the solenoid valve, a profile of a current in the solenoid coil; and determine a characteristic variable using a pattern recognition method based on artificial intelligence, based on at least one section of the profile or on a section of a profile derived from the profile.

26. A non-transitory machine-readable memory medium on which is stored a computer program for determining a characteristic variable for opening and/or closing a flow-through opening of a solenoid valve, the solenoid valve including a solenoid coil which is energized to raise an armature for opening the flow-through opening for a fluid, the computer program, when executed by a processing unit, causing the processing unit to perform the following steps: determining, during operation of the solenoid valve, a profile of a current in the solenoid coil; and determining a characteristic variable using a pattern recognition method based on artificial intelligence, based on at least one section of the profile or on a section of a profile derived from the profile.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0027] FIG. 1 schematically shows a solenoid valve in which a method according to the present invention.

[0028] FIG. 2 shows a schematic profile of a current in the solenoid coil of a solenoid valve for explaining a method according to the present invention in one preferred specific embodiment.

[0029] FIG. 3 shows profiles of a current in the solenoid coil of a functional solenoid valve and a blocked solenoid valve in the attraction phase and associated time derivatives for explaining a method according to the present invention in one preferred specific embodiment.

[0030] FIG. 4 shows a sequence of a method according to the present invention in one preferred specific embodiment on the basis of a circuit.

[0031] FIG. 5 shows a sequence of a method according to the present invention in another preferred specific embodiment.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

[0032] FIG. 1 schematically shows a solenoid valve 100, which is used by way of example in a fuel injector 101 and in which a method according to the present invention may be carried out. Solenoid valve 100 includes an electromagnet 110 including a solenoid coil 111, which may be designed to be annular, for example. Upon application of an (activation) voltage U, for example by an executing processing unit 180 designed as a control unit, current I flows in solenoid coil 111.

[0033] Furthermore, an armature 120 is provided, which is moreover used as a valve needle, using which a flow-through opening 150 may be closed or unblocked (opened). Furthermore, a spring 130 is provided, which engages on the armature 120 and, without energization of solenoid coil 111 and thus without magnetic force, presses armature 120 into or against flow-through opening 150 and closes it. Spring 130 may be in contact on its side facing away from the armature on a suitable component of solenoid valve 100.

[0034] Upon energization of solenoid coil 111, a magnetic force is built up and armature 120 is raised against the spring force of spring 130 and pulled in the direction of solenoid coil 111 or electromagnet 110. Flow-through opening 150 is unblocked. Upon corresponding energization of the solenoid coil, armature 120 may be raised up to stop 115.

[0035] FIG. 2 shows a schematic signal profile V of a current I over time t in the solenoid coil of a solenoid valve for explaining a method according to the present invention in one preferred specific embodiment.

[0036] At point in time t=0, the activation begins by applying an activation voltage to the solenoid coil, the current increases (attraction phase). At time t.sub.o, an inflection is apparent in current profile V; this is the opening point in time of the solenoid valve corresponding to the mentioned “Begin of Injection Pulse” method (BIP). This inflection results due to the movement of the armature (or the valve needle), which is raised by the solenoid coil at sufficiently high magnetic force.

[0037] The current then increases further up to a maximum; from then, a change typically takes place from the attraction phase into a holding phase having lower current, until the energization is ended after passage of the activation period of time Δt.sub.A. This may take place due to removal of the voltage or also due to application of an extinction voltage. The armature then falls back and closes the flow-through opening again at point in time t.sub.s. This is the closing point in time of the solenoid valve corresponding to the mentioned “End of Injection Pulse” method (EIP). Total opening time Δt.sub.o results from the difference of closing and opening points in time and may be used, for example, for determining the amount of fluid injected here.

[0038] FIG. 3 shows various profiles of a current I in the solenoid coil of a solenoid valve and associated second time derivatives I" over time t (thus d.sup.2I/dt.sup.2) for explaining a method according to the present invention in one preferred specific embodiment; in particular, only sections of the profiles which include the opening point in time are shown here.

[0039] Profile V.sub.1 corresponds here to a current profile as may occur in a properly functioning solenoid valve; in particular, the profile is comparable to profile V from FIG. 2 or the corresponding section thereof. Due to the high activation voltage used here, the inflection at opening point in time t.sub.o is less strongly pronounced than in the schematic representation from FIG. 2.

[0040] Profile V".sub.1 corresponds to the second time derivative of profile V.sub.1 and thus indicates curvature changes. In this case, a maximum is apparent clearly at opening point in time t.sub.o. The conventional detection methods use, for example, a pronounced maximum which is located after a minimum as a feature for the opening point in time of the valve. Profile V".sub.1 shown here, however, is filtered; otherwise, the maximum would be more difficult to determine or it could be determined less accurately.

[0041] Profile V.sub.2 corresponds to a current profile as may occur in a solenoid valve which is blocked or not functioning properly; there is accordingly no inflection for an opening point in time here. Profile V".sub.2 corresponds to the second time derivative of profile V.sub.2 and thus indicates curvature changes. Accordingly, a sharp maximum which follows a minimum for an opening point in time is not to be ascertained here; however, under certain circumstances conventional evaluation methods may misinterpret the profile before the chronologically expected opening point in time as a less pronounced minimum-maximum and incorrectly report back an opening point in time.

[0042] As mentioned, however, within the scope of the present invention, the current profiles are in particular used directly, thus without derivative, thus, for example, profile V.sub.1 or V.sub.2, in order to determine a characteristic variable for the opening and/or closing of the flow-through opening of the solenoid valve.

[0043] FIG. 4 shows a sequence of a method according to the present invention in one preferred specific embodiment on the basis of a circuit. Current I flowing during the energization of the solenoid coil is detected in that measuring voltage u dropping across a measuring shunt 400 is detected and amplified, for example, with the aid of an amplifier 410. This measuring voltage or its profile thus corresponds to the current profile in the solenoid coil as indicated, for example, in FIG. 2 with V or in FIG. 3 with V.sub.1 or V.sub.2. In executing processing unit 180, the measuring voltage may be converted, for example, via an analog-to-digital converter 420 into measured values processable by software. These measured values and possibly information about the point in time of the measurement are then fed in as input values 425 to the receptors of a (previously trained) artificial network as a pattern recognition method 430 used here.

[0044] Artificial neural network 430 then outputs as output value(s) 435 at least one characteristic variable, for example, opening point in time t.sub.o or closing point in time t.sub.s or both. The characteristic variable may also include a value which indicates whether or not the solenoid valve has opened (yes/no); in the case of profile V.sub.2 of FIG. 3, the solenoid valve would not have opened, for example, 0 (digital for “not open”) could be output as the value.

[0045] The characteristic variable (or several of them) may then be transferred, for example, to a correction, replacement, emergency, and/or diagnosis function 440, with the aid of which, for example, activation times or activation voltages for the solenoid coil may be corrected in the next activation cycle or an auxiliary injection. These may then be implemented in activation software 450 to activate the solenoid coil, for example, by applying a specific activation voltage; this is illustrated as an example by a switch 460.

[0046] FIG. 5 schematically shows a sequence of a method according to the present invention in another preferred specific embodiment, specifically for training the pattern recognition method, for example, artificial neural network 430. For this purpose, multiple different current profiles are supplied as input values 525 to neural network 430, which generates one or multiple output values 535 in each case. The neural network or its weights may then be adapted, for example, in a step 545 by a comparison to comparison values 540.

[0047] As already mentioned, it is advantageous if multiple profiles of the current profile are selected for various values of parameters of the solenoid valve; three parameters P.sub.1, P.sub.2, and P.sub.3 are shown as examples, which may stand, for example, for activation voltages for the solenoid coil, spring strengths of springs which press against the armature, and dimensions of the flow-through opening.

[0048] In the case of a neural network in the cloud, the input vectors used may be used as parameters P.sub.1, P.sub.2, P.sub.3 or other parameters for the permanent training thereof.

[0049] The pattern recognition method trained or taught in this way may then be used as explained with reference to FIG. 4, for example, to determine opening and closing points in time of a solenoid valve.