Method for automatically controlling an internal combustion engine

The invention concerns a method for automatically controlling an internal combustion engine, in which an individual accumulator pressure of a common rail system is detected in a measurement interval and stored, and a start of injection and an end of injection are determined on the basis of the stored pressure values of the individual accumulator.

A common rail system with individual accumulators is disclosed by DE 103 44 181 A1. A common rail system with individual accumulators differs from a conventional common rail system in that the fuel to be injected is taken directly from the individual accumulator and not from the rail. During the injection pause, just the right amount of fuel flows back into the individual accumulator from the rail, so that the individual accumulator is again full at the start of the next injection. In the method described in DE 103 44 181 A1, an injection end is determined from the individual accumulator pressure values that have been detected and stored, and a virtual start of injection is then determined from the end of injection by a mathematical function, for example, a straight line. The virtual start of injection is then set as the actual start of injection for the further automatic control of the internal combustion engine. Since the raw values of the individual accumulator pressure are used in this method, the interfering frequencies present in the system are superposed on the measuring signal. This can cause an incorrect interpretation of the end of injection and bring about a virtual start of injection that deviates from the ideal state. Therefore, this method is still unsatisfactory with respect to accuracy.

DE 10 2004 006 896 A1 discloses a method for evaluating an injector that employs the method described in DE 103 44 181 A1 to determine a virtual injection start. In this method, the end of injection is compared with a setpoint injection end, and the virtual injection start is compared with a setpoint injection start. If their deviation falls within a tolerance range, the injector is evaluated as accurate. If the deviation is too great, the injector is rated inaccurate, and either its control parameter is adjusted or this is deactivated. Of course, the previously described problems with respect to accuracy also apply here.

Therefore, the objective of the invention is to provide a method for an internal combustion engine with a common rail system as well as individual accumulators that allows more precise determination of the start of injection and the end of injection.

The method includes using the individual accumulator pressure values stored in the measurement interval to determine an evaluation window within which an injection was brought about. The term injection as used in the present invention is understood to mean both an individual injection and a multiple injection in the form of a preinjection, a main injection and a post-injection. For example, if the evaluation window of the preinjection was determined, then in a first step, both a representative injection start and a trial injection end are determined in this evaluation window. In a second step, both a trial injection start and a representative injection end are then determined in this evaluation window. The representative injection start is then checked for plausibility against the trial injection start, and the representative injection end is checked for plausibility against the trial injection end.

A plausible injection start is set as controlling for the further automatic control of the internal combustion engine, for example, as an input variable for an injection delay input-output map by which the energization duration of an injector is adjusted. Correspondingly, a plausible injection end is set as an input variable for a closure delay input-output map. If the start of injection is automatically controlled, and the end of injection is automatically controlled, then a plausible injection start or a plausible injection end represents the controlled variable in this closed-loop control system. An injection start or an injection end that is not plausible is rejected, and, if a closed-loop control system for the start of injection or a closed-loop control system for the end of injection is being used, a change is subsequently made to an open-loop operation.

In the method described here, it is advantageous that precisely in the case of multiple injections, even small amounts of fuel can be adjusted. Change in the injectors in the course of the operating time, i.e., aging, can also be detected by the method and compensated. Since the method is redundant in itself, it is stable. If the rail pressure sensor fails, the method described here guarantees continued operation of the internal combustion engine.

FIG. 1 shows a system diagram.

FIG. 2 shows a graph of a multiple injection.

FIG. 3 shows a family of characteristic curves for determining the filter cutoff frequency.

FIG. 4 shows a graph.

FIG. 5 shows a program flowchart (main program).

FIG. 6 shows a first subroutine.

FIG. 7 shows a second subroutine.

FIG. 8 shows a third subroutine.

FIG. 1 shows a system diagram of an electronically controlled internal combustion engine 1 with a common rail system along with individual accumulators. The common rail system comprises the following mechanical components: a low-pressure pump 3 for pumping fuel from a fuel tank 2 ; a variable suction throttle 4 on the low-pressure side for controlling the fuel volume flow flowing through the lines; a high-pressure pump 5 for pumping the fuel at increased pressure; a rail 6 for storing the fuel; and injectors 7 for injecting the fuel into the combustion chambers of the internal combustion engine 1 . An individual accumulator 8 is integrated in the injector 7 as an additional buffer volume. A common rail system with individual accumulators differs from a conventional common rail system in that the fuel to be injected is taken directly from the individual accumulator 8 and not from the rail 6 . During the injection pause, just the right amount of fuel flows back into the individual accumulator 8 from the rail 6 , so that the individual accumulator is again full at the start of the next injection.

The operating mode of the internal combustion engine 1 is determined by an electronic engine control unit (ECU) 10 . The electronic engine control unit 10 contains the customary components of a microcomputer system, for example, a microprocessor, interface adapters, buffers, and memory components (EEPROM, RAM). Operating characteristics that are relevant to the operation of the internal combustion engine 1 are applied in the memory components in the form of input-output maps/characteristic curves. The electronic control unit 10 uses these to compute the output variables from the input variables. FIG. 1 shows the following input variables as examples: the rail pressure pCR, which is measured by means of a rail pressure sensor 9 ; an engine speed nMOT; an individual accumulator pressure pE; and an input variable IN, which represents additional sensor signals, for example, an accelerator signal or the charge air pressure of an exhaust gas turbocharger. FIG. 1 also shows the following as output variables of the electronic control unit 10 : a PWM signal PWMSD for controlling the suction throttle 4 as a pressure regulator; a signal ye for controlling the injectors 7 (injection start/injection end); and an output variable OUT, which represents additional control signals for automatically controlling the internal combustion engine 1 , for example, a control signal for activating a second exhaust gas turbocharger during a register supercharging.

FIG. 2 shows graphs of multiple injection and comprises individual FIGS. 2A to 2C . The graphs show the individual accumulator pressure pE ( FIG. 2A ), the gradient variation GRAD pE(F 1 ) of the filtered individual accumulator pressure ( FIG. 2B ), and an enlarged segment of the gradient variation GRAD pE(F 1 ) ( FIG. 2C ), in each case as a function of the crank angle KW. FIG. 2A shows a preinjection VE, a main injection HE, and a post-injection NE. The solid line shows the individual accumulator pressure pE, and the broken line shows the filtered individual accumulator pressure pE(F 1 ). FIGS. 2B and 2C show the gradient variation of the filtered individual accumulator pressure as a solid line, while the broken line parallel to the x-axis represents a setpoint value VW.

In a measurement interval, the pressure distribution of the individual accumulator pE is recorded. In practice, the measurement interval corresponds to an operating cycle of the internal combustion engine. For reasons of better clarity, a measurement interval in the range of 340° to 375° crank angle is shown in FIG. 2A . in this measurement interval, the maximum value, point A, and the minimum value, point B, of the recorded individual accumulator pressure pE are determined, and the difference dpE is computed. The difference dpE is used to determine a first filter cutoff frequency fG 1 by means of a first characteristic curve KL 1 (see FIG. 3 ). This filter cutoff frequency fG 1 is then used to filter the pressure distribution of the individual accumulator pE (dot-dash line pE(F 1 ) in FIG. 2A ). The filtered individual accumulator pressure pE(F 1 ) is then used to compute the gradient variation GRAD pE(F 1 ) ( FIG. 2B ). The gradient variation GRAD pE(F 1 ) is then used to determine an evaluation window for each injection. In FIG. 2C , the first window F( 1 ), i.e., the evaluation window of the preinjection VE, is shown as an enlarged segment. Within the evaluation window F( 1 ), the minimum MIN( 1 ) is determined as a reference point. Starting from this minimum MIN( 1 ), in the crank angle range before the minimum MIN( 1 ), i.e., towards smaller crank angles, in a first step, the point at which the gradient variation GRAD pE(F 1 ) equals the setpoint value VW is determined. This corresponds to the point P 1 in FIG. 2C . In a second step, a derivative action dW, for example, dW=3° crank angle, is deducted from the crank angle corresponding to point P 1 , which in the present case is 345°. This corresponds to the crank angle W 1 , which is the first boundary of the evaluation window F( 1 ). Starting again from the minimum MIN( 1 ), in the crank angle range after the minimum MIN( 1 ), i.e., towards larger crank angles, in a first step, the point at which the gradient variation GRAD pE(F 1 ) equals the setpoint value VW is determined. This corresponds to the point P 2 in FIG. 2C . In a second step, the derivative action dW is added to the crank angle corresponding to point P 2 , which in the present case is 350°. This corresponds to the crank angle W 2 , which is the second boundary of the evaluation window F( 1 ). Therefore, the evaluation window F( 1 ) of the preinjection VE corresponds to the crank angle range from W 1 to W 2 . The evaluation window for the main injection HE and the evaluation window for the post-injection NE are then computed in analogous fashion.

FIG. 3 shows a family of characteristic curves for determining the filter cutoff frequencies. The input variable is the difference of the individual accumulator pressure pE. The output variable is the filter cutoff frequency fG for the filtering of the individual accumulator pressure pE. Three characteristic curves KL 1 to KL 3 are shown. A first characteristic curve KL 1 is used in the determination of the evaluation windows. If the difference is, for example, dpE=150 bars (see FIG. 2A ), then a first filter cutoff frequency fG1=2200 Hz is computed by means of the first characteristic curve KL 1 . A second characteristic curve KL 2 is used in the determination of the representative injection start and the trial injection end. If the difference is dpE=dpE( 1 ) (see FIG. 2A ), then a second filter cutoff frequency fG 2 is computed by means of the second characteristic curve KL 2 . The third characteristic curve KL 3 is used in the determination of the trial injection start and the representative injection end. If the difference is dpE=dpE( 1 ), then a third filter cutoff frequency fG 3 is computed by means of the second characteristic curve KL 3 . A more exact explanation of when the second filter cutoff frequency fG 2 and the third filter cutoff frequency fG 3 are used is given below with reference to FIG. 4 .

FIG. 4 shows the gradient variation of the filtered individual accumulator pressure as a function of the crank angle KW. The solid line denotes a gradient variation GRAD pE(F 2 ) obtained when the individual accumulator pressure pE is filtered with the second filter cutoff frequency fG 2 . The broken line denotes a gradient variation GRAD pE(F 3 ) obtained when the individual accumulator pressure pE is filtered with the third filter cutoff frequency fG 3 . The central idea here is to filter the individual accumulator pressure pE in the evaluation window with the second filter cutoff frequency fG 2 and then to filter the individual accumulator pressure with the third filter cutoff frequency fG 3 and to determine the characteristics of the injection, in other words, the start of injection and the end of injection, from the respective gradient variation.

In a first step, the maximum pressure and the minimum pressure are determined for the preinjection VE, and the pressure difference dpE( 1 ) is computed (see FIG. 2A ). The pressure difference dpE( 1 ) is then used in a second step to determine the second filter cutoff frequency fG 2 by means of the second characteristic curve KL 2 ( FIG. 3 ). In a third step, the individual accumulator pressure pE is filtered with the second filter cutoff frequency fG 2 , and the gradient variation GRAD pE(F 2 ) is computed from this (see the solid line in FIG. 4 ). The minimum MIN( 1 ) within the evaluation window F( 1 ), here the evaluation window of the preinjection, is then determined in a fourth step. In a fifth step, starting from this minimum MIN( 1 ), in the crank angle range before the minimum MIN( 1 ), the point at which the gradient variation GRAD pE(F 2 ) equals the setpoint value VW is determined. This point is the representative injection start SB. In a sixth step, again starting from the minimum MIN( 1 ), in the crank angle range after the minimum MIN( 1 ), the point at which the gradient variation GRAD pE(F 2 ) equals the setpoint value VW is determined. This point is the trial injection end SEp. The representative injection start SB and the trial injection end SEp within the evaluation window F( 1 ) are thus defined.

The individual accumulator pressure pE is then filtered with the third filter cutoff frequency fG 3 . The third filter cutoff frequency fG 3 is computed from the difference dpE( 1 ) by means of the third characteristic curve KL 3 . Proceeding as described above, the gradient variation GRAD pE(F 3 ) is computed, the minimum is determined, the point before the minimum at which the gradient variation GRAD pE(F 3 ) equals the setpoint value VW is determined, and the point after the minimum at which the gradient variation GRAD pE(F 3 ) equals the setpoint value VW is determined. The point thus found before the minimum is the trial injection start SBp, and the point thus found after the minimum is the representative injection end SE. The representative injection start SB is then checked for plausibility against the trial injection start SBp, for example, by taking the quotient or the difference. If the difference is less than a limit, then the representative injection start is plausible and is set as controlling for the further automatic control of the internal combustion engine. Similarly, the representative injection end SE is checked for plausibility against the trial injection end SEp. If it is plausible, then the representative injection end SE is set as controlling, for example, as an input variable for a closure delay input-output map.

FIG. 5 is a flowchart of the main program for the method. At SI the individual accumulator pressure is detected in the measurement interval and stored. The measurement cycle is understood to mean, for example, one operating cycle of the internal combustion engine, i.e., 0° to 720° crank angle. Naturally, instead of a range of crank angles, it is also possible to use a time frame. At S 2 the maximum individual accumulator pressure and the minimum individual accumulator pressure are determined, and the difference dpE is computed (see FIG. 2A ). At S 3 the first filter cutoff frequency fG 1 is determined as a function of the difference dpE by means of the first characteristic curve KL 1 ( FIG. 3 ). At S 4 the distribution of the individual accumulator pressure pE is filtered with the first filter cutoff frequency fG 1 , and at S 5 the gradient variation GRAD pE(F 1 ) is computed from this. At S 6 the program is routed to a first subroutine UP 1 , by which the evaluation windows are determined. The first subroutine UP 1 is shown in FIG. 6 and will be described in conjunction with that figure. After the evaluation windows have been determined, the program is routed at S 7 to a second subroutine UP 2 , by which a representative injection start and a trial injection end are determined within each evaluation window. The second subroutine UP 2 is shown in FIG. 7 and will be described in conjunction with that figure. After the return from the second subroutine UP 2 , the program is routed at S 8 to a third subroutine UP 3 , by which a trial injection start and a representative injection end are computed within each evaluation window. The third subroutine UP 3 is shown in FIG. 8 and will be described in conjunction with that figure.

At S 9 the representative injection start SB(i) is then checked for plausibility against the trial injection start SBp(i), for example, by taking the difference. If the difference is less than a limit GW (interrogation result S 9 : yes), then at S 10 the representative injection start SB(i) of the evaluation window, for example, the evaluation window F( 1 ) of the preinjection, is set as the controlling injection start SB(i) for this evaluation window. If the representative injection start SB(i) is not plausible (interrogation result S 9 : no), then at S 11 the representative injection start SB(i) and the trial injection start SBp(i) of this evaluation window F(i) are rejected. At S 12 the representative injection end SE(i) is then checked for plausibility against the trial injection end SEp(i) of this evaluation window F(i), for example, the evaluation window F( 1 ) of the preinjection. If the difference is less than the limit GW (interrogation result S 12 : yes), then at S 13 the representative injection end SE(i) is set as the controlling injection end SE(i). The term “set as controlling” means that the representative injection end SE(i) is an input variable of a closure delay input-output map. If, on the other hand, the injection end is automatically controlled, then the representative injection end SE(i) represents the controlled variable of this closed-loop control system. If the representative injection end SE(i) is not plausible, then at S 14 the representative injection end SE(i) and the trial injection end SEp(i) are rejected. The program then ends.

FIG. 6 shows a first subroutine UP 1 for determining the evaluation windows. At Si the number n of injections in the interval is read in. This can be one injection or a multiple injection. As shown in FIG. 2A , a multiple injection comprises a preinjection, a main injection, and a post-injection. Therefore, n = 3 . At S 2 a running variable is set to the value i=1. Then at S 3 the gradient variation GRAD pE(F 1 ) is used to determine the minimum MIN(i), in this case MIN( 1 ). At S 4 the point P 1 and the point P 2 are determined. The point P 1 corresponds to the point in the range of crank angles before the minimum MIN(i), i.e., towards smaller crank angles, at which the gradient variation GRAD pE(F 1 ) equals the setpoint value VW minus the derivative action dW, for example, dW=3° crank angle. The point P 2 corresponds to that point in the range of crank angles after the minimum MIN(i), i.e., towards larger crank angles, at which the gradient variation GRAD pE(F 1 ) equals the setpoint value VW plus the derivative action dW. The crank angle W 1 corresponds to point P 1 , and the crank angle W 2 corresponds to point P 2 . The window F(i) is thus defined as the range of crank angles W 1 to W 2 . At S 7 a check is made to determine whether the evaluation windows of all injections have been computed, i.e., Whether i=n. If this is not the case (interrogation result S 7 : no), then at S 8 the running variable i is increased by one, and at S 3 the next evaluation window is determined. If, on the other hand, all of the evaluation windows have been determined (interrogation result S 7 : yes), the program returns to the main program of FIG. 5 , where the next step S 7 is executed.

FIG. 7 shows a second subroutine UP 2 for determining a representative injection start SB(i) and a trial injection end SEp(i) within an evaluation window F(i). At Si a running variable i is set to the initial value i=1. At S 2 the individual accumulator pressure pE within the time window F(i), in this case F( 1 ), is read in, and at S 3 the maximum and minimum pressure values are determined and their difference dpE(i) is computed. At S 4 the second filter cutoff frequency fG 2 is determined as a function of the pressure difference dpE(i) by means of the second characteristic curve KL 2 . At S 5 the individual accumulator pressure pE is then filtered with the second filter cutoff frequency fG 2 . At S 6 the gradient variation GRAD pE(F 2 ) is computed from the filtered distribution pE(F 2 ) of the individual accumulator pressure pE. At S 7 the minimum in the gradient variation GRAD pE(F 2 ) is then determined, and that point before the minimum at which the gradient variation GRAD pE(F 2 ) equals the setpoint value VW is determined. This point is the representative injection start SB(i) in the window F(i). The point after the minimum at which the gradient variation GRAD pE(F 2 ) equals the setpoint value VW is then determined. This point is the trial injection end SEp(i) in the window F(i). At S 8 an interrogation determines whether the representative injection start SB(i) and the trial injection end SEp(i) have been determined for all injections n. If this is not the case (interrogation result S 8 : no), then the running variable i is increased by one. Otherwise, the program returns to the main program.

FIG. 8 shows a third subroutine UP 3 for determining a trial injection start SBp(i) and a representative injection end SE(i) within an evaluation window F(i). At S 1 a running variable i is set to the initial value i=1. At S 2 the individual accumulator pressure pE within the time window F(i), in this case F( 1 ), is read in, and at S 3 the maximum and minimum pressure values are determined and their difference dpE(i) is computed. At S 4 the third filter cutoff frequency fG 3 is determined as a function of the pressure difference dpE(i) by means of the third characteristic curve KL 3 . At S 5 the individual accumulator pressure pE is then filtered with the third filter cutoff frequency fG 3 . At S 6 the gradient variation GRAD pE(F 3 ) is computed from the filtered distribution pE(F 3 ) of the individual accumulator pressure pE. At S 7 the minimum in the gradient variation GRAD pE(F 3 ) is then determined, and that point before the minimum at which the gradient variation GRAD pE(F 3 ) equals the setpoint value VW is determined. This point is the trial injection start SBp(i) in the window F(i). The point after the minimum at which the gradient variation GRAD pE(F 3 ) equals the setpoint value VW is then determined. This point is the representative injection end SE(i) in the window F(i). At S 8 an interrogation determines whether the trial injection start SBp(i) and the representative injection end SE(i) have been determined for all injections n. If this is not the case (interrogation result S 8 : no), then the running variable i is increased by one. Otherwise, the program returns to the main program.