JP2008184943A - Fuel injection pressure control device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a fuel injection pressure control device capable of suppressing inconveniences such as engine stalling resulting from a sudden change in fuel pressure which may occur before learning correction processing of a control rule relating to fuel injection pressure control. <P>SOLUTION: In a common rail type fuel injection system for injecting and supplying fuel to an object engine, a fuel injection pressure control device for controlling fuel injection pressure is provided with a program for executing plural types of learning correction processings (temporary learning correction processing and actual learning correction processing) corresponding to plural types of execution conditions with different hardnesses in order from a condition milder than the other by taking as an object a drive current amount of a suction regulation valve utilized for the fuel injection pressure control and for correcting each time a prescribed control rule relating to the fuel injection pressure control on the basis of learning value of a drive current amount of a suction regulation valve, and a program for setting narrower a variable range of a gain integral term relating to a correction amount of the learning correction processing as a degree of learning progress is increased. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a fuel injection pressure control apparatus that controls a fuel injection pressure when fuel is supplied to a target engine, and more specifically, for example, is suitably used in a common rail fuel injection system of a diesel engine. The present invention relates to a fuel injection pressure control device.

  As is well known, for example, an engine (internal combustion engine) used as a power source of an automobile or the like ignites and burns fuel supplied by a fuel supply system to generate output torque. That is, the performance (characteristics) of the fuel supply system (fuel supply system) is one of important factors (factors) that determine the output characteristics of the engine when the engine is controlled. In recent years, in diesel engines and the like, as such a fuel supply system, high-pressure fuel (for example, light oil having a fuel pressure of about “1400 atm”) is accumulated and held in a common rail (accumulation piping), and the high-pressure fuel is supplied to the target engine as needed. Common rail fuel injection systems that supply fuel are being adopted. As a general configuration of this system, in addition to the common rail, a rail pressure sensor for measuring the pressure in the common rail (rail pressure) and a predetermined fuel (for example, light oil) are pumped from the fuel tank and pumped to the common rail. A fuel pump, a suction control valve (SCV) that makes the amount of fuel sucked into the pump variable, and an injector (fuel) that injects high-pressure fuel accumulated in the common rail to the target engine (specifically, the specified cylinder) There is known one that is further provided with an injection valve or the like.

  In such a common rail fuel injection system, it is important to manage rail pressure. For this reason, generally, based on the above configuration, feedback control (PID control) is performed so as to bring the rail pressure value (actually measured value) closer to the target value. Specifically, for example, by adjusting a supply current amount (corresponding to a drive amount of the suction adjustment valve) with respect to the suction adjustment valve while referring to a control map (adaptation map) created in advance by experiments or the like, The discharge amount of the fuel pump is variably controlled to a desired amount (target value). In this way, the rail pressure, which is a main parameter for determining the fuel injection pressure, is controlled to a target value, and fuel supply (injection supply) to the target engine is performed at an appropriate fuel injection pressure.

  However, even with such a method (pressure control method), the rail pressure and thus the fuel injection pressure cannot always be controlled with high accuracy. For example, when mass-producing and selling each element of the fuel supply system, for example, the fuel pump, the intake regulating valve, the ECU (electronic control unit), the battery, etc. There are some individual differences (machine differences) in the characteristics of various control parts including In particular, a general suction regulating valve is likely to have individual differences (variations) due to its structural characteristics, for example, in the shape of the valve opening or the biasing force of a spring that biases the valve body. However, for mass production of all the products, control maps and control formulas that take into account individual differences when mounted on the vehicle (for example, “the current amount of the intake control valve” and “the discharge amount of the fuel pump”) Creating maps and mathematical formulas) is not time-consuming when considered in current production systems. Therefore, even when using a map or mathematical expression in which an appropriate value is written in advance, it is difficult to perform control in which all the influences (variations) due to the individual differences are considered.

Therefore, conventionally, as described in, for example, Patent Document 1, learning is made of the deviation between the discharge characteristics of the fuel pump and the actual discharge characteristics defined on the control map as described above under stable operating conditions during idling operation. Thus, there has been proposed an apparatus for correcting the control map in order to compensate for the learned shift amount, that is, the map error including the shift caused by the individual difference (machine difference). According to such an apparatus, the control map can be corrected after the fuel supply system is mounted on the vehicle. That is, even if an error due to the above-described individual difference (machine difference) occurs in the control map, a control map (or a control equation or the like) in which the error (map error) is compensated by the above correction can be obtained. It becomes possible.
JP 2004-293540 A

  As described above, according to the device (fuel injection pressure control device) described in Patent Document 1, the learning correction process is performed to compensate the error (map error) due to the individual difference (machine difference). It becomes possible to obtain a map (or a control expression or the like). However, even when such an apparatus is used, the above error (map error) is not compensated if learning itself is not performed. Therefore, the fuel injection pressure is controlled with a control map including an error (map error) until learning is performed. Actually, in the initial stage of assembly of the engine, a more preferable time is the first engine after the engine and control parts including the fuel supply system are all mounted on the vehicle and accurate learning data is obtained. It is not easy to reliably perform such learning correction processing at start-up.

  In general, the manufacture of automobiles, etc. is carried out in cooperation with many parts manufacturers (or specialized departments), and the process of manufacturing and inspecting control parts including ECUs (electronic control units), and engines and controls for vehicles The process of assembling parts or the like and inspecting in the assembled state is performed by a different manufacturer (or specialized department). For this reason, it is not always easy for the manufacturer (or specialized department) in charge to perform the learning correction process at the initial stage of engine assembly. Also, in order to shorten the work time, there is a situation where it is desired to avoid taking time for such learning correction processing as much as possible. That is, in the present situation, the engine may be started before the learning correction process is completed. And as mentioned above, it is difficult to avoid this surely.

  The inventor paid attention to the occurrence of engine stall (engine stop), so-called engine stall, as one of the problems that are concerned by starting the engine before the completion of the learning correction process. That is, the learning correction process is started on the condition that, for example, the engine is warmed up and is in an idling operation state after the engine is started, and it takes about 7 to 10 seconds to complete. Cost. In order to complete the learning correction process normally, it is necessary to maintain the engine in a stable operation state (for example, idling operation state) at least during that period (a period during which learning is executed). However, in the process of performing the inspection with the engine assembled, there are many opportunities to move the vehicle, and the driver may step on the accelerator pedal even during the execution period of the learning correction process (during learning execution). is there. At this time, if an accelerator operation (acceleration racing) is performed in which the accelerator pedal is greatly depressed (instantaneously), the rail pressure suddenly rises and then drops rapidly, following the sudden drop. As a result, the injection pressure (rail pressure) also decreases. In particular, when a parameter related to the rail pressure (for example, the discharge amount of the fuel pump) is fed back by PID control, a gain integral term (so-called I term) that is one of gains (PID constants) at the time of sudden pressure rise is a negative pressure side. Over-accumulated state (in the direction of decreasing pressure). For this reason, even if the rail pressure suddenly starts to drop, the integral term continues to lower the rail pressure for a certain period, and during that period, the rail pressure drops rapidly. Normally, an engine fuel injection pressure control device is designed assuming such an irregular (anomalous) accelerator operation, and thus a rail pressure drop associated therewith. Fuel injection control is normally performed by absorbing such a drop in rail pressure. However, if the above-mentioned accelerator racing is performed before the completion of the learning correction process and the map error (control deviation) at that time is large, the rail pressure drop (corresponding to the injection pressure) cannot be absorbed. If the possible lower limit value is exceeded (goes out of the injectable region), there is a risk of engine stall due to poor injection.

  The present invention has been made in view of the above circumstances, and inconveniences such as engine stall due to a sudden change in fuel pressure, which are a concern before the completion of the learning correction process for the control law related to fuel injection pressure control, are generated. The main object is to provide a fuel injection pressure control device that can be suppressed.

  Hereinafter, means for solving the above-described problems and the effects thereof will be described.

  According to the first aspect of the present invention, the fuel supply system that injects and supplies fuel to the target engine performs the injection supply through variable control of a predetermined pressure parameter (for example, the pressure itself or other parameter that affects the pressure). In the fuel injection pressure control device for controlling the fuel injection pressure at the time, execution of a plurality of types having different strictness (for example, the number of conditions, parameter range, etc.) targeting at least one pressure parameter used for the fuel injection pressure control A plurality of types (twice or more) of learning correction processes corresponding to the conditions are performed in order from those with the milder conditions, and each time a predetermined control law (for example, the fuel injection pressure control) based on the learned value of the target pressure parameter (for example, Learning correction processing execution means for correcting a target pressure parameter and other parameters), and a learning pair For at least one of the parameters to be taken, as the degree of learning progress from the unlearned stage to the time when all of the plurality of types of learning correction processing are performed by the learning correction processing execution means becomes larger (more advanced), Correction range variable means for setting the variable range of the learning correction parameter related to the correction amount of the learning correction processing to a narrower range.

  With such a configuration, the learning correction processing can be performed in a plurality of stages by the learning correction processing execution means. That is, from simple learning performed under relatively mild conditions (for example, conditions in which the fuel injection pressure is stable with some variation), conditions more severe than the previous stage (for example, conditions under which the fuel injection pressure is more stable). ) The learning stage (corresponding to the degree of learning progress) is shifted (advanced) to the more precise learning performed below, and the control deviation related to the fuel injection pressure (the predetermined control law Error) can be reduced step by step.

  In addition, the above-described apparatus can suppress the occurrence of an engine stall or the like that is a concern before the completion of the above-described learning correction process, by including the correction range variable means. Specifically, in order to perform the learning correction process described above, it is necessary to change the fuel injection pressure within a certain range, but in this range (pressure range necessary for learning), the degree of learning progress becomes large (the learning stage is The more you proceed, the smaller. Therefore, by setting the variable range of the learning correction parameter related to the correction amount of the learning correction processing to a narrower range as the learning progress degree becomes larger (the learning stage further proceeds) by the correction range variable means, It is possible to accurately regulate excessive correction such as correction due to over-accumulation of the gain integral term described above, and to suppress the occurrence of inconvenience such as engine stall due to the sudden change in fuel pressure as described above. Become.

  In addition, as a configuration for suppressing the occurrence of inconvenience such as the engine stall described above, a configuration in which the accelerator operation during execution of the learning correction process is invalidated (including regulation and prohibition) is also conceivable. However, with such a configuration, the occurrence of inconvenience due to the accelerator operation is surely prevented, but the accelerator operation is continuously invalidated during the execution of the learning correction process, so that the work efficiency may be lowered this time. Become. Moreover, even if it is configured with a means for notifying learning completion with sound, light, or the like, or configured to always display whether learning has been completed, the work efficiency decreases due to the standby time spent for learning correction processing. Is inevitable. In this regard, in the apparatus according to the first aspect, since the occurrence of inconvenience is suppressed while allowing the accelerator operation during the learning process, the work efficiency is maintained high while suppressing the occurrence of inconvenience. Therefore, the significance of the above device is great also in this sense.

  As a more specific configuration of the apparatus according to the first aspect, as in the invention according to the second aspect, for example, the learning correction processing execution means may be a predetermined main learning condition (a fixed condition or a variable condition may be used). ) Before the execution of the main learning correction process performed based on the establishment of (), the temporary learning correction process performed based on the establishment of a predetermined temporary learning condition (either a fixed condition or a variable condition) that is looser than the main learning condition. A provisional learning stage in which the correction range variable means is for a predetermined pressure parameter among the pressure parameters to be learned, and the learning progress degree has advanced to the provisional learning correction process; In this case, the variable range of the learning correction parameter related to the correction amount of the provisional learning correction process is set to a narrower range than the unlearned stage, and the learning progress degree is further increased to the main learning correction process. In the main learning stage, it is effective to set the variable range of the learning correction parameter relating to the correction amount of the main learning correction process to a range that is further narrower than the provisional learning stage. is there. As described above, by providing the provisional learning correction process before the main learning correction process (final learning correction process), the configuration according to the first aspect is suitably realized.

  The configuration described in claim 2 includes a configuration in which another learning process is added to the two types of learning processes (the main learning correction process and the provisional learning correction process). However, in order to simplify the processing related to learning, a configuration in which only the above-described two types of learning processing, that is, only the main learning correction processing and the temporary learning correction processing are all of the plurality of types of learning processing is more effective. .

  In this case, the correction range varying means is used, and the pressure guard for defining the boundary of the variable range for the variable range of the learning correction parameter is variable as in the invention described in claim 3. By newly providing at least one of the upper limit and the lower limit of the range, or by changing the position of the pressure guard already provided, the variable range of the learning correction parameter becomes narrower as the learning progress degree increases. It is more effective to make the range. With such a configuration, for example, by providing the pressure guard on the control map, or by changing the position of the pressure guard provided on the control map, for example, the learning correction parameter variable according to the learning progress degree can be changed. The setting of the range is preferably performed, and as a result, the correction range varying means is preferably realized.

  In the device according to any one of claims 1 to 3, as in the invention according to claim 4, at least one of the pressure parameters to be learned is set to the fuel injection pressure. It is more effective to provide a feedback control means for performing feedback control based on a pressure deviation which is a deviation between the target value and the measured value, and the learning correction parameter is a gain related to the feedback control.

  When feedback control is performed on the pressure parameter, a gain (control gain) related to the feedback control is one of the main parameters that determine the amount of correction of the pressure parameter (control target) from time to time. Therefore, according to the invention described in claim 4, the apparatus described in any one of claims 1 to 3 is easily and accurately realized.

  Here, the type of gain used as the learning correction parameter is arbitrary, and for example, any one of PID constants (proportional term, integral term, derivative term) or any combination can be used. However, what is particularly important in the pressure control before learning is an integral term (I term) that is highly likely to be cumulatively added according to the pressure deviation as described above and is likely to be in an over-accumulated state. In order to further increase the practicality of the apparatus according to the fourth aspect, a configuration using a gain integral term indicating the strength of the integral operation as the learning correction parameter is particularly beneficial, as in the invention according to the fifth aspect. is there.

  According to a sixth aspect of the present invention, in the apparatus according to the fourth or fifth aspect, the gain related to the feedback control increases the correction amount as the pressure deviation increases. With such a configuration, it is possible to improve the convergence of the feedback control (to shorten the time until convergence).

  According to a seventh aspect of the present invention, in the apparatus according to any one of the fourth to sixth aspects, the learning correction parameter is prepared for learning separately from a gain used for normal fuel injection pressure control. It is a gain. With such a configuration, it is possible to accurately perform the learning correction process while minimizing the influence on the normal fuel injection pressure control using a gain different from that of the normal control.

  According to an eighth aspect of the present invention, in the apparatus according to the seventh aspect of the present invention, the learning gain is a gain that makes a correction amount larger than a gain used for normal fuel injection pressure control (for example, sensitivity of pressure deviation). High gain). With such a configuration, the variable range set through the correction range variable means increases the convergence of the feedback control while suppressing the occurrence of inconvenience such as engine stall due to the sudden change in fuel pressure described above, and learning time Can be shortened.

  In the apparatus according to any one of claims 1 to 8, the fuel injection pressure control can be accurately performed through pump control. In this case, that is, the fuel supply system includes a fuel pump that pumps the predetermined fuel and a fuel injection valve that injects and supplies the fuel pumped by the fuel pump to the engine. If it is, as in the invention according to claim 9, one of the pressure parameters to be learned is a parameter related to the discharge amount of the fuel pump, It becomes possible to realize the apparatus more easily and accurately.

  Further, as a more practical configuration of the device according to claim 9, as in the invention according to claim 10, the fuel supply system can vary the fuel intake amount to the fuel pump in accordance with the drive amount. It is advantageous to have a configuration in which the parameter relating to the discharge amount of the fuel pump is a parameter (current amount or voltage value, etc.) relating to the drive amount of the suction adjustment valve. With such a configuration, the controllability when controlling the fuel injection pressure is also high.

  On the other hand, in the apparatus according to any one of claims 1 to 10, as the execution condition of the learning correction process, a condition that makes the fuel injection pressure more stable is set stepwise according to the learning progress degree. It is effective to do. In particular, as in the invention described in claim 11, the first learning condition (for example, provisional learning condition) related to the first learning correction process (for example, provisional learning correction process) which is one of the plurality of types of learning correction processes. ) Includes one of the conditions that the engine is in an idling operation state, and a second learning correction process (learning correction process in which a learning progress degree is higher than the first learning correction process). For example, the second learning condition (for example, the main learning condition) related to the main learning correction process includes one or more conditions related to the fuel injection pressure in addition to the establishment of the first learning condition. Is particularly effective. Under such execution conditions, both the first and second learning correction processes are performed under a condition in which the fuel injection pressure is relatively stable, and the second learning correction process is the first learning correction process. In this case, the fuel injection pressure is more stable. And by these 1st and 2nd learning correction processes, it becomes possible to reduce the error of the control law made into the correction object stepwise with high accuracy.

  Regarding the apparatus according to any one of claims 1 to 11, it is beneficial that the fuel supply system is a common rail fuel injection system in view of practicality in the present situation. In the common rail fuel injection system, since the management of the fuel injection pressure is particularly important, in order to further enhance the practicality of such a device, as in the invention described in claim 12, any one of the above claims 1-11. The apparatus according to claim 1, wherein the fuel supply system is a common rail fuel injection system, rail pressure measuring means for measuring pressure in the common rail, and rail pressure measuring means for controlling the fuel injection pressure. It is more effective to include an injection pressure control unit that performs control to bring the pressure measurement value obtained by the pressure closer to the target value (which may be a fixed value or a variable value according to a predetermined parameter) (preferably matched) It is.

  In a thirteenth aspect of the invention, in the fuel supply system for injecting and supplying fuel to the target engine, the injection supply is performed through variable control of a predetermined pressure parameter (for example, the pressure itself or another parameter that affects the pressure). In the fuel injection pressure control device for controlling the fuel injection pressure at the time, execution of a plurality of types having different strictness (for example, the number of conditions, parameter range, etc.) targeting at least one pressure parameter used for the fuel injection pressure control A plurality of types of learning correction processes corresponding to the conditions are performed in order from the one with the milder conditions, and each time a predetermined control law (for example, the target pressure parameter and others) related to the fuel injection pressure control based on the learned value of the target pressure parameter is performed. A learning correction processing execution means for correcting the control map for associating the parameters with each other, and the learning target For at least one of the parameters, as the degree of learning progress from the unlearned stage to the time when all of the plurality of types of learning correction processes are performed by the learning correction process execution means increases (is more advanced), the pressure parameter And a regulation strengthening means for strengthening regulations concerning change.

  Even with such a configuration, similarly to the apparatus according to the first aspect, it is possible to reduce the control deviation (error of a predetermined control law) related to the fuel injection pressure in a stepwise manner. Also in this apparatus, by providing the restriction strengthening means, it is possible to suppress the occurrence of an engine stall or the like that is a concern before the completion of the learning correction process described above. Specifically, in order to perform the learning correction process for the control law related to fuel injection pressure control, it is necessary to change the fuel injection pressure within a certain range, but the learning progress degree becomes large in that range (learning stage). The smaller it is). Therefore, by increasing the degree of learning progress (the learning stage further progresses) by the above-mentioned regulation strengthening means, the regulation regarding the change of the pressure parameter related to the fuel injection pressure control is strengthened, for example, the variable range (changeable) of the same parameter. Range) and the ease of change (for example, the amount of change per unit time) are changed so that the change in the parameter is suppressed, so that learning problems can be kept low (or no problem at all). It is possible to suppress the occurrence of inconvenience such as engine stall due to the aforementioned sudden change in fuel pressure.

  As a more specific configuration of the apparatus according to the thirteenth aspect, as in the invention according to the fourteenth aspect, for example, the learning correction processing execution means may be a predetermined main learning condition (a fixed condition or a variable condition may be used). ) Before the execution of the main learning correction process performed based on the establishment of (), the temporary learning correction process performed based on the establishment of a predetermined temporary learning condition (either a fixed condition or a variable condition) that is looser than the main learning condition. In the provisional learning stage in which the restriction strengthening means targets a predetermined pressure parameter among the pressure parameters to be learned, and the learning progress degree has progressed to the provisional learning correction process. In some cases, the restriction on the change in the pressure parameter is strengthened more than in the unlearned stage, and when the learning progress level is further in the main learning stage that has advanced to the main learning correction process, It is effective to configure the one in which enhance than regulations relating to the change of parameters further the temporary learning phase. As described above, by providing the provisional learning correction process before the main learning correction process (final learning correction process), the configuration according to the thirteenth aspect of the present invention is preferably realized.

  Further, as in the invention according to claim 15, in the apparatus according to claim 13 or 14, the restriction on the change of the pressure parameter is, for example, a variable range of the pressure parameter, a pressure change rate, or a variable pressure change rate. This can be done by range. In the above apparatus, the restriction can be strengthened by making at least one of these parameters smaller by the restriction strengthening means. That is, according to such a configuration, the regulation is preferably strengthened, and the regulation strengthening means is suitably realized.

  In the apparatus according to claim 15, the restriction strengthening means is replaced with at least one of a variable range of the target pressure parameter and a variable range of the pressure change rate as in the invention according to claim 16. By newly providing a pressure guard that defines the boundary of the variable range for the target at least one of the upper limit and the lower limit of the target variable range, or by changing the position of the pressure guard that is already provided As the degree of learning progress increases, it is effective to make the target variable range (at least one of the variable range of the target pressure parameter and the variable range of the pressure change rate) a narrower range.

  In the apparatus according to claim 16, as in the invention according to claim 17, at least one of the pressure parameters to be learned is set to a target value and a measured value of the fuel injection pressure. It is effective to provide a feedback control means for performing feedback control based on a pressure deviation which is a deviation of the pressure, and the pressure change rate is a gain related to the feedback control.

  Further, with regard to the device according to claim 17, it is effective that the pressure change rate is a gain integral term indicating the strength of the integral operation as in the invention according to claim 18. is there.

  According to the configurations described in the fifteenth to seventeenth aspects, practicality and feasibility can be improved in the same manner as the apparatuses described in the third to fifth aspects.

  Hereinafter, an embodiment embodying a fuel injection pressure control device according to the present invention will be described with reference to the drawings. In addition, the apparatus of this embodiment is mounted in the common rail type fuel injection control system (high pressure injection fuel supply system) which made the control object the reciprocating type diesel engine as a car engine, for example. That is, this device, like the device described in Patent Document 1, directly injects high-pressure fuel (for example, light oil with an injection pressure of “1000 atm” or higher) directly into the combustion chamber in the cylinder of a diesel engine (internal combustion engine). This is a so-called fuel injection pressure control device for a diesel engine, which is used for feedback control (PID control) of the fuel injection pressure with respect to a target value when supplying (direct injection supply).

  First, an outline of a common rail fuel injection control system according to the present embodiment will be described with reference to FIG. Note that a multi-cylinder (for example, in-line 4-cylinder) engine for automobiles is assumed as the engine of the present embodiment.

  As shown in FIG. 1, in this system, the ECU (electronic control unit) 30 takes in sensor outputs (detection results) from various sensors and drives the fuel supply device based on the sensor outputs. Configured to control. The ECU 30 controls the drive of the fuel supply device, for example, to control the output (rotation speed and torque) of a diesel engine, and feedback control the fuel injection pressure (rail pressure) to the engine to a target value (target fuel pressure). (PID control).

  Here, various devices constituting the fuel supply device are arranged in order of the fuel tank 10, the fuel filter 12, the fuel pump 14, the common rail 16, and the injector 20 (fuel injection valve) from the upstream side of the fuel. The common rail 16 is provided with a fuel pressure sensor 22 for detecting the fuel pressure (rail pressure) in the common rail 16 so that the rail pressure can be detected and managed. With such a configuration, the fuel in the fuel tank 10 is pumped up by the fuel pump 14, and is pressurized (suppressed) through the fuel filter 12 to the common rail 16. The common rail 16 stores the fuel pumped from the fuel pump 14 in a high pressure state and supplies the fuel to the injector 20 of each cylinder through the high pressure fuel passage 18 provided for each cylinder.

  The injector 20 is a fuel injection valve for high pressure fuel. In this injector 20, drive power is transmitted through a hydraulic chamber (command chamber). Specifically, when the pressure in the hydraulic chamber is increased or decreased according to the energized state (energized / non-energized) of the injector 20, the needle reciprocates (up and down) in the valve cylinder (inside the housing). The nozzle hole is opened and closed. At this time, the pressure increasing process in the hydraulic chamber is performed by supplying fuel from the fuel tank 10 by the fuel pump 14. On the other hand, the decompression process of the hydraulic chamber is performed by returning the fuel in the hydraulic chamber to the fuel tank 10 through a pipe 20 a connecting the injector 20 and the fuel tank 10.

  Thus, in this system, the fuel pumped by driving the fuel pump 14 is directly supplied to each cylinder (inside the cylinder) of the engine by each injector 20 (direct injection supply).

  Next, a detailed configuration of the fuel pump 14 will be described with reference to FIG.

  As shown in FIG. 2, the fuel pump 14 is basically configured to pressurize and discharge the fuel pumped up from the fuel tank 10 by the feed pump 40 with the high-pressure pump 50. . At this time, the amount of fuel pumped to the high-pressure pump 50 is adjusted by a suction adjustment valve 60 (SCV: Suction Control Valve) provided on the fuel suction side of the pump 14 (particularly before fuel pumping by the high-pressure pump 50). It has come to be measured.

  Here, the feed pump 40 has an outer rotor on the outer side and an inner rotor on the inner side, and the space created by each of the rotors is increased / decreased according to the rotation of each rotor, and fuel is sucked and discharged according to the increase / decrease. This is a so-called trochoid pump. The pump 40 functions as a so-called low-pressure supply pump that sucks the fuel in the fuel tank 10 from the inlet 42 and sends the fuel to the high-pressure pump 50, and is driven by the rotation of the drive shaft 41. The drive shaft 41 is interlocked with the crankshaft 24 (FIG. 1), and is driven by power from the engine output. That is, the drive shaft 41 is driven (rotation driven) with the rotation of the crankshaft 24 and rotates at a ratio of “1/1” or “½” with respect to one rotation of the crankshaft 24, for example.

  The fuel sucked up by the feed pump 40 passes through the fuel filter 42a and is sent to the intake adjustment valve 60. At this time, the discharge pressure (fuel pressure) of the feed pump 40 is limited (adjusted) to a predetermined pressure or less by the regulator valve 43. The regulator valve 43 communicates the discharge side and the supply side of the feed pump 40 when the discharge pressure of the feed pump 40 exceeds a predetermined pressure. The temperature of the fuel sent to the intake adjustment valve 60 is detected by the fuel temperature sensor 43a.

  The intake adjustment valve 60 is configured to include a linear solenoid type electromagnetic valve, and adjusts the intake fuel amount of the high-pressure pump 50. The ECU 30 (FIG. 1) controls the energization time (supply current amount) for the intake adjustment valve 60 so that the amount of fuel drawn from the feed pump 40 to the high-pressure pump 50 through the fuel passage 44 can be adjusted. It has become. That is, the fuel sent by the feed pump 40 is adjusted to a required discharge amount (target fuel pressure feed amount) by the suction adjustment valve 60 and enters the high-pressure pump 50 through the suction valve 53 (suction valve).

  The high-pressure pump 50 is a plunger pump that pressurizes the fuel metered by the suction adjustment valve 60 and discharges the fuel to the outside. The high-pressure pump 50 is mainly configured to include a plunger 51 that is driven to reciprocate by a drive shaft 41, and a pressurizing chamber 52a formed between the inner wall 52b of the housing 52 and the top surface of the plunger 51. The volume (volume) of the pressurizing chamber 52a (plunger chamber) changes as the plunger 51 reciprocates in the axial direction.

  The plunger 51 is pressed against a ring cam 56 mounted around an eccentric cam 55 (eccentric cam) by a spring 57. Although not shown, in detail, a cylindrical shaft hole for assembling the drive shaft 41 is formed at the center of the rectangular ring-shaped ring cam 56. A cylindrical eccentric cam 55 corresponding to the shape of the shaft hole is attached to the drive shaft 41 so as to be eccentric. Then, the ring cam 56 is assembled on the eccentric cam 55 of the drive shaft 41 in such a manner that the drive shaft 41 passes through the shaft hole of the eccentric cam 55, so that the drive shaft 41 and the ring cam 56 are connected to the eccentric cam 55. It is connected through. In the high-pressure pump 50, when the drive shaft 41 rotates, the eccentric cam 55 rotates eccentrically, and the ring cam 56 displaces following the eccentric cam 55, thereby pushing (or pulling) the plunger 51 in the axial direction. In this way, the two plungers 51 reciprocate between the pumping top dead center and the pumping bottom dead center.

  As described above, on the suction side of the high-pressure pump 50, the suction valve 53 that connects or blocks the pressurizing chamber 52a and the feed pump 40 side is disposed. On the other hand, a discharge valve 54 is also provided on the discharge side of the high-pressure pump 50 to communicate or block the pressurizing chamber 52a and the common rail 16 side. That is, when the pressure in the pressurizing chamber 52a is lowered by the lowering of the plunger 51, the discharge valve 54 is closed and the suction valve 53 is opened. As a result, fuel is supplied from the feed pump 40 into the pressurizing chamber 52 a via the suction adjustment valve 60. Conversely, when the pressure in the pressurizing chamber 52a rises due to the rise of the plunger 51, the suction valve 53 is now closed. When the pressure in the pressurizing chamber 52a reaches a predetermined pressure, the discharge valve 54 is opened, and the high-pressure fuel pressurized in the pressurizing chamber 52a is supplied toward the common rail 16.

  The control system according to the present embodiment is equipped with the fuel supply device including the fuel pump 14. In addition to the above sensors, the vehicle (not shown) is further provided with various sensors for vehicle control. For example, as shown in FIG. 1, on the outer peripheral side of the crankshaft 24, a crank angle sensor 24a that outputs a crank angle signal at every predetermined crank angle (for example, at a cycle of 30 ° CA) is connected to the crankshaft 24 (engine output shaft). ) Rotation angle position, rotation speed (engine rotation speed), and the like. In addition, an accelerator sensor 26 that outputs an electric signal corresponding to the state (displacement amount) of the pedal is provided in the accelerator pedal in order to detect an operation amount (accelerator opening degree) of the accelerator pedal by the driver. . A vehicle speed sensor 28 that outputs a rotation signal related to the axle is provided on the axle connected to the drive wheels (tires) of the vehicle in order to detect the traveling speed of the vehicle.

  In such a system, the ECU 30 functions as the fuel injection pressure control device of the present embodiment and performs engine control mainly as an electronic control unit. The ECU 30 includes a known microcomputer (not shown), grasps the operating state of the target engine and user requirements based on the detection signals of the various sensors, and accordingly various actuators such as the injector 20 Is operated to perform various controls related to the engine in an optimum manner according to the situation at that time. The microcomputer mounted on the ECU 30 basically includes a CPU (basic processing device) that performs various calculations, and a RAM (main memory that temporarily stores data and calculation results during the calculation) ( Random Access Memory (ROM), ROM as program memory (read only storage device), EEPROM (electrically rewritable non-volatile memory) as data storage memory and backup RAM (backup power source such as in-vehicle battery) RAM), signal processing devices such as A / D converters and clock generation circuits, various arithmetic devices such as input / output ports for inputting / outputting signals to / from the outside, storage devices, signal processing devices, It is comprised by the communication apparatus etc. The ROM stores various programs and control maps related to engine control including a program related to the fuel injection pressure control, and the data storage memory (for example, EEPROM) includes design data of the target engine. Various control data to be stored are stored in advance.

  By the way, the apparatus of the present embodiment also has a relationship between the discharge characteristics of the fuel pump defined as the control law on the control map and the actual discharge characteristics under stable operating conditions during idling operation, as described in Patent Document 1. The amount of deviation is learned, and the control map is corrected to compensate for the learned deviation amount, that is, the map error including the deviation caused by the individual difference (machine difference) described above. However, in the apparatus of the present embodiment, during the fuel injection pressure control, a plurality of types of learning correction processing (temporary learning correction processing, main learning correction processing) are performed in order from the ones with the weakest execution conditions, and each time the intake adjustment valve 60 ( The control map (control law) relating to the fuel injection pressure control is corrected based on the learned value of the supply current amount (pressure parameter) with respect to FIG. In addition, for these learning correction processes, as the learning progress level (the progress level from the unlearned stage until all of the learning correction process is performed) increases, the learning correction process (the temporary learning correction process in the temporary learning stage, In the main learning stage, the variable range of the parameter (feedback control gain) related to the correction amount of the main learning correction process is set to a narrower range. Next, the fuel injection pressure control will be described in detail with reference to FIGS. 3, 6, 7, and 9 are basically executed at predetermined crank angles or at predetermined time intervals by executing a program stored in the ROM by the ECU 30. It is executed sequentially. Further, the values of various parameters used in the processes of these drawings are stored as needed in a storage device such as a RAM or EEPROM mounted in the ECU 30, and updated as necessary.

  FIG. 3 is a flowchart showing a processing procedure of pump control corresponding to the main part of the fuel injection pressure control. First, the pump control according to the present embodiment, that is, the processing related to the control of the fuel pump 14 (FIG. 2) will be described with reference mainly to FIG. Since this pump control is based on general control, a detailed description is omitted here, and the scope of the description is limited to a simple description.

  As shown in FIG. 3, in this series of processing, first, in step S11, the engine speed (NE) is calculated based on the output of the crank angle sensor 24a, and the accelerator is based on the output of the accelerator sensor 26. The pedal operation amount (accelerator opening) is calculated.

  Next, in step S12, the target common rail pressure PP is acquired (calculated) based on the engine rotational speed and the accelerator pedal operation amount acquired in step S11. Specifically, for example, a predetermined map (for example, stored in a ROM or a mathematical expression) in which an appropriate value (optimum value) of the target common rail pressure PP is written for each engine rotation speed and each accelerator pedal operation amount by an experiment or the like is used. Get.

  In step S13, the actual common rail pressure NP is acquired (calculated) based on the output of the fuel pressure sensor 22. In the subsequent step S14, based on the actual common rail pressure NP and the target common rail pressure PP acquired in step S12, a pressure deviation DP (= PP−NP) as a difference between the two is calculated.

  Next, in step S15, the required discharge amount PQ of the fuel pump 14 is calculated based on the pressure deviation DP acquired in step S14 and the pressure leak amount (e.g., estimated from the target injection amount). In the following step S16, a drive amount necessary for causing the pump 14 to discharge fuel corresponding to the required discharge amount PQ, that is, a drive current amount PI of the intake adjustment valve 60 is calculated. Specifically, for example, a control map (I-Q map) as indicated by a solid line Q0 in FIG. 4, that is, an appropriate value (equivalent value) of the drive current amount PI is written for each required discharge amount PQ by, for example, experiments in advance. It is acquired using a predetermined map (for example, stored in a ROM or the like, may be a mathematical expression). In a further subsequent step S17, a current corresponding to the drive current amount PI is supplied to the intake adjustment valve 60, whereby the drive amount of the fuel pump 14 is controlled so that the required discharge amount PQ is satisfied. To do.

  In the present embodiment, the series of processes in FIG. 3 are repeatedly executed, so that the pressure in the common rail 16 (corresponding to the fuel injection pressure) is sequentially feedback-controlled (PID control) to the target common rail pressure PP. Moreover, in the present embodiment, the convergence of the control is enhanced by setting a value for increasing the correction amount to the gain related to the feedback control of the fuel injection pressure as the pressure deviation DP increases. FIG. 5 is a graph showing a gain correction amount (feedback correction amount) used for normal fuel injection pressure control in this embodiment.

  As shown in FIG. 5, in this embodiment, a plurality of types of gains G11 to G14 are used properly according to the situation. These gains G11 to G14 all increase the correction amount as the pressure deviation DP increases, but the sensitivity to the pressure deviation DP (corresponding to the slope of the graph) is different for each gain. In the present embodiment, due to such a difference in characteristics, the use gain is switched based on the threshold values TH1 and TH2 in FIG. 5 in detail according to the magnitude of the pressure deviation DP. That is, for example, in the non-idle operation state, when the pressure deviation DP is large, the gain G12 (DP> TH1) or the gain G13 (DP <TH2) having high sensitivity to the pressure deviation DP is used, and the pressure deviation DP is small (TH2 For ≦ DP ≦ TH1, a gain G14 having a small sensitivity to the pressure deviation DP is used. On the other hand, in the idling operation state, the gain G11 having a smaller sensitivity is used.

  In this embodiment, in addition to the pump control shown in FIG. 3, the control map (IQ map, solid line shown in FIG. 4) is used to compensate for the map error including the deviation caused by the individual difference (machine difference). The process (learning process) for learning the map error of Q0) and the correction process for compensating for the learned map error are performed stepwise, such as a temporary learning correction process and then a main learning correction process. . In addition, at this time, the gain integral term (parameter indicating the strength of the integral operation) related to the feedback control of the fuel injection pressure, more specifically, the gain integral term (I term) related to the variable control of the drive current amount PI of the intake adjustment valve 60. ) Is provided so as to define the boundary of the variable range, and the position of the pressure guard is changed so that the variable range of the gain integral term becomes narrower as the learning progress degree becomes larger. Next, the processing contents of the learning process and the correction process (learning correction process) and the pressure guard setting process will be described in detail with reference to FIG. 4 and FIGS.

  FIG. 6 is a flowchart showing a processing procedure for determining success or failure of the learning execution condition for the learning process related to each learning correction process, that is, the temporary learning process and the main learning process. Hereinafter, a series of processes related to the learning execution condition determination will be described with reference to FIG.

  As shown in FIG. 6, in this series of processing, first, in steps S21 and S22, it is determined whether or not the conditions of the provisional learning correction processing and the main learning correction processing are satisfied, and provisional learning correction is performed. If both the execution condition related to the process (provisional learning execution condition) and the execution condition related to the main learning correction process (main learning execution condition) are satisfied, the learning flag (initial value = “0”) is determined in step S232. ")", If only the provisional learning execution condition is satisfied, "1" is set in the learning flag in step S231. If both are not satisfied, the learning flag is set in step S233. "0" is set for each. Here, the provisional learning execution condition is, for example, that the engine is in an idling operation state. More specifically, for example, the accelerator pedal operation amount is substantially “0” (detected by the accelerator sensor 26), the vehicle is in a stopped state (detected by the vehicle speed sensor 28), and the engine speed is within a predetermined range. The engine is in an idling operation state when all (or partly) conditions such as being within the range (detected by the crank angle sensor 24a) are satisfied. On the other hand, the present learning execution condition is based on the premise that the provisional learning execution condition is satisfied, and further is that the fuel injection pressure is in a stable state. More specifically, for example, it is in a warm-up state (for example, determination based on the engine coolant temperature), fuel temperature (detected by the fuel temperature sensor 43a), rail pressure (detected by the fuel pressure sensor 22), target fuel It is assumed that the fuel injection pressure is stable when the injection amount (calculated value) is within a predetermined range, and when all (or partly) such conditions are satisfied. As described above, in the present embodiment, the execution condition of the main learning correction process is set as a stricter condition (a condition that is difficult to be established) than the execution condition of the temporary learning correction process. That is, these learning correction processes are performed in the order of a temporary learning correction process and then a main learning correction process.

  In the present embodiment, the success or failure of the execution condition related to each learning correction process is thus determined. Then, along with the learning execution condition determination in FIG. 6, the series of processing shown in FIG. 7 is also repeatedly executed at predetermined processing intervals, so that whether or not the learning processing is executed is controlled based on the value of the learning flag. Yes. Next, with reference to FIG. 4 and FIG. 8 in addition to FIG. 7, the execution mode of the learning process and the contents of the process are described in detail.

  As shown in FIG. 7, in this series of processes, first, in step S31, it is determined whether or not the value of the learning flag satisfies the execution condition. Only when the learning flag is set to either “1” or “2” by the process 6, the predetermined learning process is executed in the subsequent step S 32. As shown in FIG. 7, in the present embodiment, the same processing is performed regardless of whether the learning flag is “1” or “2”. However, it is also possible to assign different processes for these two cases.

  Here, the learning process executed in step S32 is performed, for example, in a manner as shown in FIG. In the present embodiment, as the main map error, the drive current amount PI (horizontal axis) of the intake adjustment valve 60 and the required discharge amount PQ of the fuel pump 14 with respect to the regular map fit value indicated by the solid line Q0 in FIG. Learning is performed on the assumption that the relationship with the (vertical axis) is offset as it is in parallel with the direction of the current amount PI (horizontal axis) and errors as shown by the broken lines Q1 and Q2 occur.

  In this learning process, for example, the deviation of the current amount PI between the solid line Q0 and the broken line Q1 (or broken line Q2) in FIG. 4 is calculated and used as a learning value. Specifically, for example, by feedback control (PID control), the current amount PI is gradually changed so that the actual discharge amount (for example, converted from the rail pressure) approaches the required discharge amount PQ indicated by the solid line Q0, The amount of change (e.g., calculated as an integral value) from before the change until the discharge amounts match is used as the learning value. The learning value obtained in this way is stored in a nonvolatile manner in, for example, an EEPROM or a backup RAM. By doing so, even if the ECU 30 is disconnected once and restarted when the engine is stopped, the data stored therein remains without being erased. In this learning process, a gain (PID constant) related to feedback control is prepared separately from a gain used for normal fuel injection pressure control, more specifically than a gain for normal control (FIG. 5). A learning gain for increasing the correction amount (for example, a gain having a higher sensitivity to the pressure deviation DP than the gain G11 in FIG. 5) is used.

  Thereafter, the learning value stored by this learning process is used as a correction coefficient. That is, in step S16 shown in FIG. 3, the learning value is used as a correction coefficient so as to compensate for the deviation from the solid line Q0. For example, in FIG. 4, in the current amount I0 at idling, the error of the discharge amount “PQ1−PQ0” in the case of the broken line Q1, and the error of the discharge amount “PQ2−PQ0” in the case of the broken line Q2 Each compensation is made by the processing in step S16.

  As described above, in the present embodiment, the learning process as described above is performed for each of the temporary learning correction process and the main learning correction process. However, the effects of these learning correction processes differ depending on the execution conditions. 8 (a) to 8 (c) show learning from an unlearned stage (FIG. 8 (a)) to a provisional learning stage (FIG. 8 (b)) and a main learning stage (FIG. 8 (c)). In other words, it is a graph showing the state of map errors that decrease as the learning progress degree increases.

  As shown in FIGS. 8A to 8C, the map error Dt gradually decreases as learning progresses. In the present embodiment, the learning progress degree proceeds in one direction from the unlearned stage to the main learning stage in principle. Therefore, once the learning is completed up to the main learning stage, the learning is not performed again except when the pressure deviation DP is abnormally large. In addition, when the pressure deviation DP is abnormally large, fail-safe processing (for example, lighting of a warning lamp) other than re-execution of learning may be performed.

  As described above, the control deviation (error range) becomes smaller as the learning progress degree becomes larger (learning progresses). Along with this, the adjustment range of the parameter value to be learned (the drive current amount PI of the suction adjustment valve 60) also becomes narrower. In the present embodiment, in accordance with the transition of the error range, and hence the adjustment range of the current amount PI, the gain related to feedback control that determines the correction amount in each learning stage, in particular, the boundary of the variable range of the gain integral term. The position of the pressure guard to be determined is variably set. FIG. 9 is a flowchart showing the processing procedure of the pressure guard setting process.

  As shown in FIG. 9, in this process, different values of pressure guards are variably set according to the value of the learning flag. Specifically, when “1” is set in the learning flag, it is determined in step S41 that “1” is set in the learning flag, and in the subsequent step S431, the temporary guard for the pressure guard is determined. Value is set. On the other hand, when “2” is set in the learning flag, it is determined in step S42 that “2” is set in the learning flag, and in the subsequent step S432, the learning guard value is set to the pressure guard. Is set. On the other hand, if a value other than “1” and “2” (for example, “0”) is set in the learning flag, an unlearned guard value is set in the pressure guard in the subsequent step S433. These guard values are provided for the gain integral term (I term) related to the variable control (feedback control) of the drive current amount PI of the suction regulating valve 60, and the guard range (regulation) based on these guard values. The range becomes narrower as the learning progress degree becomes larger (unlearned> temporary learning> main learning).

  Next, referring to FIG. 10, the setting mode of the pressure guard will be described in more detail. Here, a device that corrects the control map (I-Q map) in one learning process, represented by the device described in Patent Document 1, is used as a comparative example. The mode is shown in FIG. 10 (a), the setting mode of the present embodiment is shown in FIG. 10 (b). 10A and 10B, the pressure guard value when the pressure deviation is the same for the unlearned stage, the provisional learning stage, and the main learning stage is the reference value G0 (center of variation). The position of the pressure guard for each of these learning stages is shown.

  As shown in FIG. 10 (a), in the comparative example, the pressure guards G1a and G1b are in the untrained state, and the pressure guards G3a and G3b are in the gain integral terms related to the feedback control when learning (during the actual learning). The upper and lower limits of the variable range are set. When learning progresses, these two pairs of pressure guards are switched from pressure guards G1a and G1b to pressure guards G3a and G3b having a narrower guard range at timing t2.

  On the other hand, in the apparatus of this embodiment, the pressure guards G1a and G1b when not learning, the pressure guards G2a and G2b during temporary learning, and the pressure guards G3a and G3b during main learning, respectively, are gain integral terms related to the feedback control. The upper and lower limits of the variable range are set. These three pairs of pressure guards are set symmetrically with respect to the reference value G0 (variation center). In particular, the unguarded guard range and thus the pressure guards G1a and G1b are set based on variations in pressure parameters estimated in advance through experiments or the like. For example, it is set on the assumption that the drive current amount PI of the suction adjustment valve 60 varies from the center of variation by about “± 450 mA” (an example of a value confirmed by the inventors' experiment).

  As learning progresses, these three pairs of pressure guards are first guarded from pressure guards G1a and G1b to pressure guards G2a and G2b having a narrower guard range at timing t1, and then guarded from pressure guards G2a and G2b at timing t2. The pressure guards G3a and G3b are switched to a narrow range. That is, in the present embodiment, as the learning progress level (the progress level from the unlearned stage to the time when all of the learning correction process is performed) increases, the learning correction process (temporary learning correction process, main learning in the temporary learning stage) is increased. If it is a stage, the variable range of the parameter (gain integral term) relating to the correction amount of this learning correction process) is set to a narrower range, more specifically, as narrow as possible within the range that can be learned for each learning stage. By doing so, inconveniences such as engine stall caused by a sudden change in fuel pressure, which is a concern before the completion of the learning correction process described above, are suppressed. FIG. 11 shows the timing when the engine stall (so-called engine stall) is suppressed by setting the pressure guards G2a and G2b, taking as an example the state when the engine is started for the first time after assembly of the engine (after vehicle mounting). Shown as a chart. In this case as well, the same comparative example as in FIG. 10 is used, the operation mode of the comparative example is shown in FIG. 11A, the setting mode of this embodiment is shown in FIG. I do.

  As shown in FIG. 11A, in the comparative example (corresponding to FIG. 10A), the accelerator operation as shown by the solid line L0a in FIG. When an accelerator operation (acceleration racing) is performed such that it is greatly depressed (instantaneously), each parameter relating to the fuel injection pressure control, and the engine speed, that is, the gain integral term, the actual common rail pressure NP, the target common rail pressure PP , And the engine rotation speed respectively change in a manner shown by a two-dot chain line L1a, a solid line L3a, a one-dot chain line L4a, and a broken line L2a in FIG. That is, in this example, due to the accelerator racing, the gain integral term is overaccumulated to the negative pressure side, and the engine rotation speed is rapidly reduced with the overaccumulation of the gain integral term, and the engine stalls due to poor injection. .

  On the other hand, in the apparatus of this embodiment, as shown in FIG. 11B, when an accelerator operation (accelerator racing) as shown by a solid line L0b in FIG. 11B-1 is performed. , Parameters related to fuel injection pressure control, and engine rotation speed, that is, the gain integral term, actual common rail pressure NP, target common rail pressure PP, and engine rotation speed are indicated by a two-dot chain line in FIG. It changes in the manner shown by L1b, solid line L3b, alternate long and short dash line L4b, and broken line L2b. That is, in the apparatus of the present embodiment, over-integration of the gain integral term (two-dot chain line L1b) is restricted (guarded) by setting the pressure guards G2a and G2b. As a result, a decrease in engine rotation speed due to a sudden change (abrupt decrease) in fuel pressure, and thus the occurrence of the engine stall described above, are suppressed.

  As described above, according to the fuel injection pressure control device according to the present embodiment, the following excellent effects can be obtained.

  (1) For a fuel supply system (common rail fuel injection system) that injects and supplies fuel to the target engine, the fuel injection pressure when the injection is supplied is controlled through variable control of a predetermined pressure parameter. As such a fuel injection pressure control device (ECU 30 for engine control), the pressure parameter (driving adjustment valve 60 drive current amount PI) used for the fuel injection pressure control is targeted, and a plurality of execution conditions with different severity are dealt with. The plurality of types of learning correction processing (temporary learning correction processing and main learning correction processing) are performed in order from those with the milder conditions (see FIG. 6), and each time the fuel injection pressure control is performed based on the learning value of the current amount PI. 4 is provided with a program (learning correction processing execution means, step S32 in FIG. 7) for correcting a predetermined control law (I-Q map, FIG. 4). Furthermore, the amount of correction of the learning correction process increases as the learning progress degree from the unlearned stage until all of the learning correction processes are performed (more progress) for the current amount PI to be learned. The program is provided with a program (correction range variable means, FIG. 9) for setting the variable range of the learning correction parameter (gain integral term) related to the feedback correction amount to a narrower range. Specifically, in step S32 in FIG. 7, the main learning is performed before the execution of the main learning correction process performed based on the establishment of a predetermined main learning condition (condition in which the engine is idling and the fuel injection pressure is stable). Temporary learning correction processing performed based on the establishment of predetermined temporary learning conditions (conditions in the idling operation state) that are looser than the conditions is executed. In the process of FIG. 9, when the learning progress degree is in the provisional learning stage for the current amount PI, the variable integration range of the gain integral term related to the correction amount of the provisional learning correction process is not learned. When the learning progress is in the main learning stage, the variable range of the gain integral term related to the correction amount of the main learning correction process is set to a range narrower than that in the temporary learning stage. (See FIG. 10). By adopting such a configuration, in step S32 of FIG. 7, that is, by the temporary learning correction process and the main learning correction process, the control deviation (error of a predetermined control law) related to the fuel injection pressure is reduced stepwise. It becomes possible to make it. In addition, as the processing in FIG. 9, as the degree of learning progress increases, the variable integration range of the gain integral term is set to a narrower range, so that the excessive correction due to the overintegration of the gain integral term can be accurately regulated. As a result, it is possible to suppress the occurrence of inconvenience such as engine stall due to the aforementioned sudden change in fuel pressure.

  (2) The pressure parameter (current amount PI) to be learned is configured to include a program (regulation strengthening means, FIG. 9) that strengthens restrictions on changes in the pressure parameters as the degree of learning progress increases. Specifically, in the processing of FIG. 9, a pressure guard that defines the boundary of the variable range for the variable range of the gain integral term is provided at the upper and lower limits of the target variable range, or has already been provided. By changing the position of the pressure guard, the variable range of the gain integral term is made narrower as the degree of learning progress increases. With such a configuration, the setting of the variable range of the gain integral term corresponding to the degree of learning progress is suitably performed.

  (3) A program (feedback control) that feedback-controls the drive current amount PI of the intake adjustment valve 60 based on a pressure deviation DP that is a deviation between a target value (calculated value) of fuel injection pressure and a measured value (sensor detected value). Means, FIG. 3). By doing so, it becomes possible to accurately control the fuel injection pressure based on the control law (I-Q map, FIG. 4) corrected by the learned value of the current amount PI.

  (4) The pressure guard is provided for the variable range of the gain integral term that is particularly important in pressure control. With such a configuration, the over-accumulation described above is suppressed by the pressure guard, and the fuel injection pressure control is suitably performed.

  (5) As a gain related to the feedback control of the current amount PI (and thus the fuel injection pressure), a gain that increases the correction amount as the pressure deviation increases is employed (see FIG. 5). By doing so, it is possible to improve the convergence of feedback control (to shorten the time until convergence).

  (6) The learning gain is a learning gain prepared separately from the gain used for normal fuel injection pressure control. By doing so, it is possible to accurately perform the learning correction process while minimizing the influence on the normal fuel injection pressure control using a gain different from the normal control.

  (7) The learning gain is a gain that makes the correction amount larger than the gain used for normal fuel injection pressure control (for example, a gain with high sensitivity to pressure deviation). In this way, the variable range set through the processing of FIG. 9 suppresses the occurrence of inconvenience such as engine stall due to the sudden change in the fuel pressure as described above, and feedback control of the current amount PI (and thus the fuel injection pressure). It is possible to improve convergence and shorten learning time.

  (8) The learning target is a parameter related to the discharge amount of the fuel pump 14. Specifically, learning is performed on a parameter related to the drive amount of the intake adjustment valve 60, that is, the drive current amount PI of the intake adjustment valve 60. With such a configuration, the fuel injection pressure can be controlled with high controllability.

  (9) The execution condition of the temporary learning correction process (provisional learning condition) includes that the engine is in an idling operation state as one of the conditions, and the execution condition of the main learning correction process (main learning condition) is In addition to the provisional learning condition being satisfied, the fuel injection pressure is further included (see FIG. 6). Under such execution conditions, the provisional learning correction process and the main learning correction process can reduce the error of the control law to be corrected stepwise with high accuracy.

  (10) A program for measuring the pressure in the common rail (rail pressure measuring means, step S13 in FIG. 3), and for controlling the fuel injection pressure, the measured pressure value (actual common rail pressure NP) is set to the target value (target common rail pressure PP). And a program (injection pressure control means, steps S14 to S17 in FIG. 3) for performing the control so as to be close to. According to such a configuration as the fuel injection pressure control device, the common rail fuel injection system can inject and supply high-pressure fuel whose injection pressure is controlled with high accuracy to the target engine. . In automobile engines and the like, high-pressure fuel injection is a technology that greatly contributes to emission improvement. In order to realize an environmentally friendly clean diesel vehicle, this configuration is very significant.

  The above embodiment may be modified as follows.

  The type of gain that is the target of pressure guard setting is not limited to the integral term (I term), and is arbitrary, for example, any one of PID constants (proportional term, integral term, derivative term), or any combination Can be used. Incidentally, among these three, the proportional term is suitable for the target after the integral term.

  In the above embodiment, the learning target pressure parameter (the pressure itself or another parameter acting on the pressure) is set to one (current amount PI). However, a learning process may be performed for a plurality of pressure parameters. .

  In the above-described embodiment, only the parallel component (current direction) of the IQ map (FIG. 4) is learned and corrected. However, the slope component is applied in a manner similar to the apparatus described in Patent Document 1. This may also be learned and corrected.

  In the above embodiment, the case where the fuel injection control is performed based on one control map is illustrated as a simpler configuration example. However, for example, a plurality of control maps associated with different engine operating conditions, injection conditions, and the like are used. It may be. For these control maps, if the above-described learning, correction, and pressure guard setting are performed separately for each of the engine operating state (for example, engine speed) and injection conditions (for example, rail pressure), the above-described map is described for each map. Errors due to individual differences are compensated, and more precise fuel injection control becomes possible.

  The setting mode of the pressure guard is not limited to the method of setting different values according to the learning progress degree for one map as shown in FIG. For example, a plurality of maps in which different guard values are set in advance may be used, and these maps may be switched according to the degree of learning progress.

  In the above-described embodiment, the case where the learning correction process is performed in two stages of the main learning correction process and the provisional learning correction process is illustrated as a simpler configuration example, but the learning correction process is performed in three or more stages. Also good.

  Regardless of the setting of the pressure guard with respect to the variable range of the gain, a program for strengthening the restriction on the change of the pressure parameter as the learning progress degree of the pressure parameter (current amount PI) to be learned (regulation strengthening means) If it is the structure provided with, the effect similar to or equivalent to the effect of said (1) will be acquired at least.

  Specifically, for example, a pressure guard is provided for the variable range of the pressure parameter (for example, the current amount PI) to be learned, and the pressure according to the degree of learning progress in a mode similar to the mode shown in FIG. By changing the position of the guard (narrowing with progress), the regulation regarding the change of the pressure parameter may be strengthened as the degree of learning progress increases.

  Alternatively, in the above-described embodiment, the gain corresponding to the change amount (pressure change rate) per unit time of the pressure parameter (current amount PI) may be reduced (less likely to change) as the learning progress degree increases. Good.

  In short, the greater the learning progress, the more difficult the pressure parameter to be learned changes, or the condition and range in which the parameter cannot be changed increases. Thus, the occurrence of inconvenience such as engine stall due to the aforementioned sudden change in fuel pressure can be suppressed.

  The type of engine to be controlled (including a spark ignition gasoline engine and the like) and the system configuration can be changed as appropriate according to the application.

  In the embodiment and the modification, it is assumed that various kinds of software (programs) are used. However, similar functions may be realized by hardware such as a dedicated circuit.

1 is a configuration diagram showing an outline of an engine control system to which the fuel injection pressure control device according to an embodiment of the present invention is applied. FIG. The block diagram which shows the detailed structure of a fuel pump. The flowchart which shows the process sequence of the pump control which concerns on this embodiment. The graph which shows an example of the control map used for the pump control. The graph which shows an example of the gain for normal control. The flowchart which shows the process sequence of the process which concerns on the success or failure determination of learning execution conditions. The flowchart which shows the process sequence of the process which concerns on learning execution. (A)-(c) is a graph which shows the change aspect of the map error accompanying learning progress. The flowchart which shows the process sequence of a pressure guard setting process. (A) And (b) is a figure which shows the setting aspect of a pressure guard. (A) And (b) is a timing chart which shows the operation | movement aspect about the comparative example and the apparatus of this embodiment, respectively.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Fuel tank, 14 ... Fuel pump, 16 ... Common rail, 20 ... Injector, 22 ... Fuel pressure sensor, 30 ... ECU (electronic control unit), 40 ... Feed pump, 43a ... Fuel temperature sensor, 50 ... High pressure pump, 60 ... Suction adjustment valve (SCV).

Claims (18)

Regarding a fuel supply system for injecting and supplying fuel to a target engine, in a fuel injection pressure control device for controlling a fuel injection pressure when the injection supply is performed through variable control of a predetermined pressure parameter,
Targeting at least one pressure parameter used for the fuel injection pressure control, a plurality of types of learning correction processing corresponding to a plurality of types of execution conditions having different severities are performed in order from the one with the milder condition, and each time Learning correction processing execution means for correcting a predetermined control law related to the fuel injection pressure control based on a learned value of a pressure parameter;
For at least one of the parameters to be learned, the learning correction increases as the degree of learning progress from the unlearned stage until all of the plurality of types of learning correction processing are performed by the learning correction processing execution unit is increased. Correction range variable means for setting the variable range of the learning correction parameter related to the correction amount of the processing to a narrower range;
A fuel injection pressure control device comprising: The learning correction processing execution means performs provisional learning performed on the basis of establishment of a predetermined provisional learning condition that is looser than the main learning conditions before execution of the main learning correction process performed on the basis of establishment of a predetermined main learning condition. To execute the correction process,
When the correction range variable means is in a provisional learning stage in which the degree of learning progresses to the provisional learning correction process for a predetermined pressure parameter among the pressure parameters to be learned, When the variable range of the learning correction parameter related to the correction amount of the temporary learning correction process is set to a range narrower than the unlearned stage, and the learning progress degree is in the main learning stage further advanced to the main learning correction process The fuel injection pressure control device according to claim 1, wherein the variable range of the learning correction parameter related to the correction amount of the main learning correction process is further set to a range narrower than the temporary learning stage.   The correction range variable means is provided by newly providing a pressure guard that defines the boundary of the variable range for the variable range of the learning correction parameter, at least one of the upper limit and the lower limit of the target variable range, or has already been The fuel injection pressure control according to claim 1 or 2, wherein the variable range of the learning correction parameter is made narrower as the learning progress degree becomes larger by changing the position of the provided pressure guard. apparatus.   Feedback control means for feedback-controlling at least one of the pressure parameters to be learned based on a pressure deviation which is a deviation between a target value and a measured value of the fuel injection pressure; The fuel injection pressure control apparatus according to any one of claims 1 to 3, wherein the fuel injection pressure control apparatus is a gain related to the feedback control.   The fuel injection pressure control apparatus according to claim 4, wherein the learning correction parameter is a gain integral term indicating the strength of the integral operation.   The fuel injection pressure control apparatus according to claim 4 or 5, wherein the gain related to the feedback control increases the correction amount as the pressure deviation increases.   The fuel injection pressure control apparatus according to any one of claims 4 to 6, wherein the learning correction parameter is a learning gain prepared separately from a gain used for normal fuel injection pressure control.   8. The fuel injection pressure control apparatus according to claim 7, wherein the learning gain is a gain that makes a correction amount larger than a gain used for normal fuel injection pressure control. The fuel supply system includes a fuel pump that pumps a predetermined fuel, and a fuel injection valve that injects and supplies the fuel pumped by the fuel pump to the engine.
The fuel injection pressure control apparatus according to any one of claims 1 to 8, wherein one of the pressure parameters to be learned is a parameter related to a discharge amount of the fuel pump. The fuel supply system includes an intake adjustment valve that varies a fuel intake amount to the fuel pump according to a drive amount,
The fuel injection pressure control apparatus according to claim 9, wherein the parameter related to the discharge amount of the fuel pump is a parameter related to the drive amount of the suction adjustment valve. The first learning condition related to the first learning correction process that is one of the plurality of types of learning correction processes includes that the engine is in an idling operation state as one of the conditions,
In addition to the fact that the second learning condition related to the second learning correction process, which is a learning correction process having a learning progress degree higher than that of the first learning correction process, is satisfied, The fuel injection pressure control device according to any one of claims 1 to 10, comprising one or more conditions relating to fuel injection pressure. The fuel supply system is a common rail fuel injection system,
Rail pressure measuring means for measuring the pressure in the common rail;
As the control of the fuel injection pressure, an injection pressure control unit that performs control to bring the pressure measurement value by the rail pressure measurement unit close to the target value;
The fuel injection pressure control device according to any one of claims 1 to 11. Regarding a fuel supply system for injecting and supplying fuel to a target engine, in a fuel injection pressure control device for controlling a fuel injection pressure when the injection supply is performed through variable control of a predetermined pressure parameter,
Targeting at least one pressure parameter used for the fuel injection pressure control, a plurality of types of learning correction processing corresponding to a plurality of types of execution conditions having different severities are performed in order from the one with the milder condition, and each time Learning correction processing execution means for correcting a predetermined control law related to the fuel injection pressure control based on a learned value of a pressure parameter;
For at least one of the parameters to be learned, the pressure parameter increases as the degree of learning progress from the unlearned stage until all of the plurality of types of learning correction processing is performed by the learning correction processing execution unit increases. Regulations strengthening means to strengthen regulations on changes in
A fuel injection pressure control device comprising: The learning correction processing execution means performs provisional learning performed on the basis of establishment of a predetermined provisional learning condition that is looser than the main learning conditions before execution of the main learning correction process performed on the basis of establishment of a predetermined main learning condition. To execute the correction process,
The restriction strengthening means targets a predetermined pressure parameter among the pressure parameters to be learned, and when the degree of learning progress is in a provisional learning stage where the learning progressed to the provisional learning correction process, When the restriction on the change of the pressure parameter is strengthened more than the unlearned stage, and the learning progress degree is further in the main learning stage that has advanced to the main learning correction process, the restriction on the change of the pressure parameter is further set in the temporary learning. The fuel injection pressure control device according to claim 13, wherein the fuel injection pressure control device is stronger than the step.   The restriction strengthening means determines at least one of a variable range of the target pressure parameter, a pressure change rate corresponding to a change amount per unit time of the parameter, and a variable range of the pressure change rate as the learning progress degree increases. The fuel injection pressure control device according to claim 13 or 14, wherein the fuel injection pressure control device is smaller.   The restriction strengthening means targets at least one of a variable range of the target pressure parameter and a variable range of the pressure change rate, and defines a pressure guard that defines a boundary of the variable range, at least an upper limit and a lower limit of the target variable range. The variable range of interest is made narrower as the learning progress degree increases by newly providing on one side or changing the position of a pressure guard already provided. 15. The fuel injection pressure control device according to 15.   Feedback control means for feedback-controlling at least one of the pressure parameters to be learned based on a pressure deviation that is a deviation between a target value and a measured value of the fuel injection pressure; The fuel injection pressure control apparatus according to claim 16, wherein the fuel injection pressure control apparatus is a gain related to the feedback control.   The fuel injection pressure control apparatus according to claim 17, wherein the pressure change rate is a gain integral term indicating an intensity of an integral operation.

Images