JP2019039380A - Control device of engine - Google Patents

Abstract

To detect both of SI knock and CI knock in a time of partial compression ignition combustion.SOLUTION: A control device of an engine includes a detection portion for detecting a parameter relating to noise generated by combustion in a cylinder, and a calculation portion for specifying a SI knock index value having a correlation to SI knock in which an unburnt gas outside of a region in which an air fuel mixture is SI burned, is quickly burnt by abnormal local self-ignition, and a CI knock index value having a correlation to CI knock in which noise of frequency lower than that in occurrence of SI knock generates during CI combustion on the basis of a detection value by the detection portion. The calculation portion specifies the SI knock index value on the basis of the value detected by the detection portion in both of the SI combustion period in which SI combustion occurs and the CI combustion period in which CI combustion occurs, and specifies the CI knock index value on the basis of the value detected by the detection portion during at least the CI combustion period.SELECTED DRAWING: Figure 15

Description

  The present invention relates to an engine control apparatus capable of performing partial compression ignition combustion in which a part of an air-fuel mixture in a cylinder is forcibly subjected to SI combustion by spark ignition and then the remaining air-fuel mixture in the cylinder is subjected to CI combustion by self-ignition. About.

  In recent years, attention has been paid to HCCI combustion in which gasoline fuel mixed with air is burned by self-ignition in a sufficiently compressed cylinder. This HCCI combustion is based on the ignition timing (mixing) of the mixture due to external factors such as temperature. There is also a problem that control during transient operation in which the load changes suddenly is difficult.

  Therefore, it has been proposed that not all of the air-fuel mixture is combusted by self-ignition, but a part of the air-fuel mixture is combusted by spark ignition using an ignition plug. That is, a part of the air-fuel mixture is forcibly burned by flame propagation (SI combustion) triggered by spark ignition, and the remaining air-fuel mixture is burned by self-ignition (CI combustion). Hereinafter, such combustion is referred to as partial compression ignition combustion.

  However, in the partial compression ignition combustion, since both SI combustion and CI combustion are performed in the cylinder, there is a concern that knocking (a phenomenon causing unpleasant noise or vibration) due to each combustion may occur. That is, during operation by partial compression ignition combustion, there is a possibility that SI knock, which is knocking due to SI combustion, and CI knock, which is knocking due to CI combustion, respectively, and both these SI / CI knocks occur. It is necessary to control combustion appropriately so that there is no.

  Here, the thing of the following patent document 1 is known as an example of the technique which detects knocking. In Patent Document 1, a knock detection window is set so as to include a specific period in which vibration due to knocking is noticeable, and whether or not knocking has occurred is determined based on engine vibration data detected in the knock detection window. Determined. Specifically, a peak value of a vibration level at a specific frequency is extracted from vibration data in the knock detection window through predetermined digital signal processing, and the extracted peak value and a predetermined threshold (knock determination threshold) are obtained. The presence or absence of knocking is determined by comparison.

Japanese Patent No. 5899972

  However, since the technique of the above-mentioned Patent Document 1 is intended for a normal spark ignition engine, it is only necessary to monitor the SI knock, and in addition to this, it is not necessary to monitor the CI knock. . For example, Patent Document 1 describes an example in which a filter is used as digital signal processing for analyzing vibration data, and a vibration level of a natural frequency (that is, a frequency corresponding to SI knock) is extracted by the filter. When such a method is applied to an engine in which partial compression ignition combustion is performed, the SI knock can be monitored, but the CI knock cannot be monitored.

  The present invention has been made in view of the circumstances as described above, and provides an engine control device capable of reliably detecting both SI knock and CI knock that are concerned during partial compression ignition combustion. With the goal.

  As a result of earnest research to solve the above problems, the present inventor has found that the frequency of noise differs between SI knock and CI knock. Specifically, the present inventors have found that the noise when CI knock occurs has a lower frequency than the noise when SI knock occurs. This is due to the difference in the cause of the occurrence of SI knock and CI knock, and is a common phenomenon regardless of the engine material and structure. Further, the present inventor has found that although CI knock occurs only during the period of CI combustion, SI knock can occur not only during the SI combustion period but also during the subsequent CI combustion period. And based on the above knowledge, the following invention was completed.

  That is, according to the present invention, partial compression ignition combustion in which part of the air-fuel mixture in the cylinder is forcibly SI-combusted by spark ignition and then the remaining air-fuel mixture in the cylinder is CI-combusted by self-ignition is performed at least in part. An engine control apparatus executed in a region, wherein a detection unit that detects a parameter related to noise generated by combustion in the cylinder, and unburned gas outside the region where the air-fuel mixture is SI-combusted are abnormal local self. The SI knock index value correlated with the SI knock that burns rapidly upon ignition, and the CI knock index value correlated with the CI knock generated during the CI combustion when noise having a frequency lower than that of the SI knock is generated. A calculation unit that identifies the detection unit based on a detection value by the unit, and the calculation unit is configured to perform a preceding operation in both an SI combustion period in which SI combustion occurs and a CI combustion period in which CI combustion occurs. The SI knock index value is specified based on a value detected by a detection unit, and the CI knock index value is specified based on a value detected by the detection unit at least during the CI combustion period. (Claim 1).

  According to the present invention, at the time of execution of partial compression ignition combustion in which SI combustion and CI combustion are mixed, that is, during an operation in which both SI knock caused by SI combustion and CI knock caused by CI combustion are concerned. Since the SI knock index value correlated with the SI knock and the CI knock index value correlated with the CI knock are specified based on the detection values (parameters related to noise) by the detection unit, these SI knock index values. Both SI knock and CI knock can be reliably detected by the determination using the CI knock index value. In particular, in the present invention, since the detected value in both the SI combustion period and the CI combustion period is taken into account in specifying the SI knock index value, whether SI knock occurs in either the SI combustion period or the CI combustion period. , Never miss this. On the other hand, when the CI knock index value is specified, at least the detection value during the CI combustion period is considered, but since CI knock occurs only during the CI combustion period, only the detection value during the CI combustion period is considered. Even if the detection value during the SI combustion period is ignored, the CI knock will not be missed. In this way, according to the present invention, it is possible to reliably detect both SI knock and CI knock without overlooking when performing partial compression ignition combustion.

  Preferably, the calculation unit specifies the intensity of a predetermined first frequency component based on values detected by the detection unit in both the SI combustion period and the CI combustion period as the SI knock index value, As the CI knock index value, the intensity of the second frequency component having a frequency lower than that of the first frequency component is specified based on at least the value detected by the detection unit during the CI combustion period (claim 2).

  According to this configuration, the degree of SI knock and CI knock with different generation causes can be accurately grasped as the intensity of different frequency components corresponding to each knock.

  In the above configuration, more preferably, the calculation unit calculates the intensity of the first frequency component by performing Fourier analysis on the waveform of the detection value detected by the detection unit in both the SI combustion period and the CI combustion period. In addition, the intensity of the second frequency component is calculated by performing Fourier analysis on the waveform of the detection value detected by the detection unit at least during the CI combustion period.

  According to this configuration, the intensity of each frequency component can be accurately calculated using Fourier analysis.

  In the above configuration, more preferably, the calculation unit applies a window function for removing ignition noise caused by the spark ignition to the waveform of the detection value detected by the detection unit, and this window function. The waveform after application is subjected to Fourier analysis.

  According to this configuration, it is possible to avoid extraction of extra frequency components due to ignition noise, so that the intensity of each frequency component can be calculated with higher accuracy.

  Preferably, the calculation unit includes a predetermined period including a combustion center of gravity which is a time point that overlaps at least a part of each of the SI combustion period and the CI combustion period and is combusted by 50% mass of fuel supplied to the cylinder. The SI knock index value is specified based on the value detected by the detection unit over a period of time, and the entire period of the predetermined period overlaps at least a part of the CI combustion period. The CI knock index value is specified based on the value detected by the detection unit (Claim 5).

  According to this configuration, the SI knock index value can be appropriately calculated based on the value detected by the detection unit in the period spanning both the preceding SI combustion period and the subsequent CI combustion period, and at least the subsequent CI combustion period. It is possible to appropriately calculate the CI knock index value based on the value detected by the detection unit in the period overlapping with.

  Further, the predetermined period may be a finite and series of periods overlapping both the SI combustion period and the CI combustion period, and the calculation unit is a value detected by the detection unit over the predetermined period. Both the SI knock index value and the CI knock index value may be specified based on the above (claim 6).

  According to this configuration, it is possible to reliably detect the SI knock and the CI knock while reducing the burden on the calculation of the SI knock index value and the CI knock index value.

  Preferably, the control device further includes a combustion control unit that adjusts at least the timing of the spark ignition based on the SI knock index value and the CI knock index value specified by the calculation unit.

  According to this configuration, regardless of whether SI knock or CI knock occurs, it can be immediately converged by adjusting the ignition timing.

  As described above, according to the engine control apparatus of the present invention, it is possible to reliably detect both SI knock and CI knock that are a concern during partial compression ignition combustion.

1 is a system diagram schematically showing an overall configuration of an engine according to an embodiment of the present invention. It is the figure which showed together sectional drawing of an engine main body, and the top view of a piston. FIG. 2 is a schematic plan view showing the structure of a cylinder and an intake / exhaust system in the vicinity thereof. It is a block diagram which shows the control system of an engine. It is the map figure which divided the operation area | region of the engine with the difference in the combustion form. It is a time chart for demonstrating roughly the combustion control performed in each operation area | region of an engine. It is a graph which shows the waveform of the heat release rate at the time of SPCCI combustion (partial compression ignition combustion). It is a flowchart which shows the specific procedure of the control performed at the time of SPCCI combustion. It is a subroutine which shows the specific procedure of control of FIG.8 S7. It is a figure for demonstrating the method to determine ignition timing in FIG.8 S3. It is a graph which shows the waveform of the cylinder pressure detected by the cylinder pressure sensor. It is a graph which shows the window function applied with respect to the pressure waveform of FIG. It is a graph which shows the waveform of the cylinder pressure obtained after applying the window function of FIG. It is a graph which shows the result of having carried out the Fourier analysis of the pressure waveform of FIG. It is a graph which shows the result of having processed the frequency spectrum of FIG. 14 into 1/3 octave band. FIG. 8 is a diagram corresponding to FIG. 7 for describing various methods of defining the SI rate.

(1) Overall Configuration of Engine FIGS. 1 and 2 are views showing a preferred embodiment of an engine to which the control device of the present invention is applied. The engine shown in this figure is a four-cycle gasoline direct injection engine mounted on a vehicle as a driving power source. The engine body 1, an intake passage 30 through which intake air introduced into the engine body 1 circulates, An exhaust passage 40 through which exhaust gas discharged from the engine body 1 flows and an EGR device 50 that recirculates part of the exhaust gas flowing through the exhaust passage 40 to the intake passage 30 are provided.

  The engine body 1 is inserted into a cylinder block 3 in which a cylinder 2 is formed, a cylinder head 4 attached to the upper surface of the cylinder block 3 so as to close the cylinder 2 from above, and a reciprocating slide in the cylinder 2 And a piston 5 which is made. The engine main body 1 is typically of a multi-cylinder type having a plurality of (for example, four) cylinders, but here, for the sake of simplification, the description will be focused on only one cylinder 2.

  A combustion chamber 6 is defined above the piston 5, and fuel mainly composed of gasoline is supplied to the combustion chamber 6 by injection from an injector 15 described later. The supplied fuel burns while being mixed with air in the combustion chamber 6, and the piston 5 pushed down by the expansion force due to the combustion reciprocates in the vertical direction. In addition, the fuel injected into the combustion chamber 6 should just contain gasoline as a main component, for example, in addition to gasoline, it may contain subcomponents, such as bioethanol.

  Below the piston 5, a crankshaft 7 that is an output shaft of the engine body 1 is provided. The crankshaft 7 is connected to the piston 5 via a connecting rod 8 and is rotationally driven around the central axis according to the reciprocating motion (vertical motion) of the piston 5.

  The geometric compression ratio of the cylinder 2, that is, the ratio between the volume of the combustion chamber 6 when the piston 5 is at the top dead center and the volume of the combustion chamber when the piston 5 is at the bottom dead center is the SPCCI combustion (described later) A value suitable for partial compression ignition combustion) is set to 13 or more and 30 or less. More specifically, the geometric compression ratio of the cylinder 2 is set to 14 or more and 17 or less in the case of a regular specification using a gasoline fuel having an octane number of about 91, and a high-octane specification using a gasoline fuel having an octane number of about 96. In some cases, it is preferably set to 15 or more and 18 or less.

  The cylinder block 3 is provided with a crank angle sensor SN1 that detects the rotation angle (crank angle) of the crankshaft 7 and the rotation speed (engine rotation speed) of the crankshaft 7.

  The cylinder head 4 is provided with an intake port 9 and an exhaust port 10 that open to the combustion chamber 6, an intake valve 11 that opens and closes the intake port 9, and an exhaust valve 12 that opens and closes the exhaust port 10. The valve type of the engine of this embodiment is a four-valve type of 2 intake valves × 2 exhaust valves as shown in FIG. That is, the intake port 9 has a first intake port 9A and a second intake port 9B, and the exhaust port 10 has a first exhaust port 10A and a second exhaust port 10B. A total of two intake valves 11 are provided so as to open and close the first intake port 9A and the second intake port 9B, respectively, and the exhaust valve 12 opens and closes the first exhaust port 10A and the second exhaust port 10B, respectively. A total of two are provided.

  As shown in FIG. 3, a swirl valve 18 that can be opened and closed is provided in the second intake port 9B. The swirl valve 18 is provided only in the second intake port 9B, and is not provided in the first intake port 9A. When such a swirl valve 18 is driven in the closing direction, the ratio of the intake air flowing into the combustion chamber 6 from the first intake port 9A where the swirl valve 18 is not provided increases, and thus the swirl valve 18 turns around the cylinder axis. The swirl flow (swirl flow) can be strengthened. Conversely, the swirl flow can be weakened by driving the swirl valve 18 in the opening direction.

  The intake valve 11 and the exhaust valve 12 are driven to open and close in conjunction with the rotation of the crankshaft 7 by valve mechanisms 13 and 14 including a pair of camshafts and the like disposed in the cylinder head 4.

  The valve operating mechanism 13 for the intake valve 11 incorporates an intake VVT 13a that can change at least the opening timing of the intake valve 11. Similarly, the valve mechanism 14 for the exhaust valve 12 incorporates an exhaust VVT 14a that can change at least the closing timing of the exhaust valve 12. By controlling the intake VVT 13a and the exhaust VVT 14a, in this embodiment, it is possible to adjust the valve overlap period in which both the intake valve 11 and the exhaust valve 12 are opened across the exhaust top dead center. By adjusting the valve overlap period, the amount of burned gas (internal EGR gas) remaining in the combustion chamber 6 can be adjusted. The intake VVT 13a (exhaust VVT 14a) may be a variable mechanism that changes only the closing timing (opening timing) while fixing the opening timing (closing timing) of the intake valve 11 (exhaust valve 12). A phase-type variable mechanism that simultaneously changes the opening timing and closing timing of the intake valve 11 (exhaust valve 12) may be used.

  The cylinder head 4 ignites a mixture of an injector 15 that injects fuel (mainly gasoline) into the combustion chamber 6, and fuel that is injected from the injector 15 into the combustion chamber 6 and air that is introduced into the combustion chamber 6. A spark plug 16 is provided. The cylinder head 4 is further provided with an in-cylinder pressure sensor SN2 that detects the pressure in the combustion chamber 6 (hereinafter also referred to as in-cylinder pressure). The in-cylinder pressure sensor SN2 corresponds to a “detection unit” in the claims.

  As shown in FIG. 2, a cavity 20 is formed on the crown surface of the piston 5, in which a relatively wide region including the central portion is recessed on the opposite side (downward) from the cylinder head 4. A substantially conical bulge 20a is formed at the center of the cavity 20 so as to bulge relatively upward, and both sides in the radial direction sandwiching the bulge 20a are recesses each having a bowl-like cross section. . In other words, the cavity 20 is a donut-shaped concave portion in plan view formed so as to surround the raised portion 20a. A region on the outer side in the radial direction from the cavity 20 in the crown surface of the piston 5 is a squish portion 21 formed of an annular flat surface.

  The injector 15 is a multi-hole type injector having a plurality of injection holes at the tip thereof, and can inject fuel radially from the plurality of injection holes (F in FIG. Represents the spray of fuel injected from the hole). The injector 15 is provided so that the tip thereof faces the central portion (the raised portion 20 a) of the crown surface of the piston 5.

  The spark plug 16 is disposed at a position somewhat shifted to the intake side with respect to the injector 15. The tip (electrode part) of the spark plug 16 is set at a position overlapping the cavity 20 in plan view.

  As shown in FIG. 1, the intake passage 30 is connected to one side surface of the cylinder head 4 so as to communicate with the intake port 9. Air (fresh air) taken from the upstream end of the intake passage 30 is introduced into the combustion chamber 6 through the intake passage 30 and the intake port 9.

  In the intake passage 30, in order from the upstream side thereof, an air cleaner 31 that removes foreign matters in the intake air, an openable / closable throttle valve 32 that adjusts the flow rate of intake air, a supercharger 33 that sends out compressed air, An intercooler 35 that cools the intake air compressed by the feeder 33 and a surge tank 36 are provided.

  Each part of the intake passage 30 includes an air flow sensor SN3 that detects the flow rate of intake air, first and second intake air temperature sensors SN4 and SN6 that detect the temperature of intake air, and first and second intake air that detect the pressure of intake air. Barometric pressure sensors SN5 and SN7 are provided. The air flow sensor SN3 and the first intake air temperature sensor SN4 are provided in a portion of the intake passage 30 between the air cleaner 31 and the throttle valve 32, and detect the flow rate and temperature of the intake air passing through the portion. The first intake pressure sensor SN5 is provided in a portion of the intake passage 30 between the throttle valve 32 and the supercharger 33 (downstream from a connection port of an EGR passage 51 described later), and the intake air pressure passing through the portion is reduced. Detect pressure. The second intake air temperature sensor SN6 is provided in a portion of the intake passage 30 between the supercharger 33 and the intercooler 35, and detects the temperature of intake air that passes through the portion. The second intake pressure sensor SN7 is provided in the surge tank 36 and detects the pressure of intake air in the surge tank 36.

  The supercharger 33 is a mechanical supercharger (supercharger) mechanically linked to the engine body 1. Although the specific form of the supercharger 33 is not particularly limited, for example, any of the known superchargers such as a Rishorum type, a roots type, or a centrifugal type can be used as the supercharger 33.

  Between the supercharger 33 and the engine main body 1, an electromagnetic clutch 34 capable of electrically switching between fastening and releasing is interposed. When the electromagnetic clutch 34 is engaged, driving force is transmitted from the engine body 1 to the supercharger 33, and supercharging by the supercharger 33 is performed. On the other hand, when the electromagnetic clutch 34 is released, the transmission of the driving force is interrupted and the supercharging by the supercharger 33 is stopped.

  The intake passage 30 is provided with a bypass passage 38 for bypassing the supercharger 33. The bypass passage 38 connects the surge tank 36 and an EGR passage 51 described later to each other. A bypass valve 39 that can be opened and closed is provided in the bypass passage 38.

  The exhaust passage 40 is connected to the other side of the cylinder head 4 so as to communicate with the exhaust port 10. Burned gas (exhaust gas) generated in the combustion chamber 6 is discharged to the outside through the exhaust port 10 and the exhaust passage 40.

  A catalytic converter 41 is provided in the exhaust passage 40. The catalytic converter 41 includes a three-way catalyst 41a for purifying harmful components (HC, CO, NOx) contained in the exhaust gas flowing through the exhaust passage 40, and particulate matter (PM) contained in the exhaust gas. And a GPF (gasoline particulate filter) 41b for collecting the gas. Note that another catalytic converter incorporating an appropriate catalyst such as a three-way catalyst or a NOx catalyst may be added downstream of the catalytic converter 41.

  The EGR device 50 includes an EGR passage 51 that connects the exhaust passage 40 and the intake passage 30, and an EGR cooler 52 and an EGR valve 53 that are provided in the EGR passage 51. The EGR passage 51 connects a portion of the exhaust passage 40 downstream of the catalytic converter 41 and a portion of the intake passage 30 between the throttle valve 32 and the supercharger 33. The EGR cooler 52 cools the exhaust gas (external EGR gas) recirculated from the exhaust passage 40 to the intake passage 30 through the EGR passage 51 by heat exchange. The EGR valve 53 is provided so as to be openable and closable in the EGR passage 51 on the downstream side (closer to the intake passage 30) than the EGR cooler 52, and adjusts the flow rate of the exhaust gas flowing through the EGR passage 51.

  The EGR passage 51 is provided with a differential pressure sensor SN8 for detecting a difference between the upstream pressure and the downstream pressure of the EGR valve 53.

(2) Control System FIG. 4 is a block diagram showing an engine control system. The ECU 100 shown in the figure is a microprocessor for comprehensively controlling the engine, and includes a known CPU, ROM, RAM, and the like.

  The ECU 100 receives detection signals from various sensors. For example, the ECU 100 includes the crank angle sensor SN1, the in-cylinder pressure sensor SN2, the air flow sensor SN3, the first / second intake temperature sensors SN4, SN6, the first / second intake pressure sensors SN5, SN7, and the differential pressure sensor SN8. Information detected by these sensors (ie, crank angle, engine speed, in-cylinder pressure, intake air flow rate, intake air temperature, intake air pressure, differential pressure before and after the EGR valve 53, etc.) is detected by the ECU 100. Are input sequentially.

  Further, the vehicle is provided with an accelerator sensor SN9 that detects an opening degree of an accelerator pedal operated by a driver driving the vehicle, and a detection signal from the accelerator sensor SN9 is also input to the ECU 100.

  The ECU 100 controls each part of the engine while executing various determinations and calculations based on the input signals from the sensors. That is, the ECU 100 is electrically connected to the intake VVT 13a, the exhaust VVT 14a, the injector 15, the spark plug 16, the throttle valve 32, the electromagnetic clutch 34, the bypass valve 39, the EGR valve 53, and the like. Based on this, a control signal is output to each of these devices.

  The ECU 100 as described above corresponds to a “calculation unit” and a “combustion control unit” in the claims.

(3) Control According to Operating State FIG. 5 is a map for explaining the difference in control according to the engine speed / load. As shown in the figure, the engine operating region is roughly divided into four operating regions A1 to A4 depending on the combustion mode. Assuming that the first operation region A1, the second operation region A2, the third operation region A3, and the fourth operation region A4, respectively, the first operation region A1 is a low speed / low load region in which both the rotational speed and the load are low. The third operation region A3 is a low speed / high load region where the rotation speed is low and the load is high, the fourth operation region A4 is a high speed region where the rotation speed is high, and the second operation region A2 is the first operation region. The remaining regions excluding the third and fourth operation regions A1, A3, and A4 (in other words, the region combining the low speed / medium load region and the medium speed region). Hereinafter, the combustion mode selected in each operation region will be described in order.

(A) First operation region In the first operation region A1 of low speed and low load, partial compression ignition combustion (hereinafter referred to as SPCCI combustion) in which SI combustion and CI combustion are mixed is executed. Specifically, in the first operation region A1, SPCCI combustion is performed in a state where supercharging by the supercharger 33 is stopped (a state of natural intake). Note that “SPCCI” in SPCCI combustion is an abbreviation of “Spark Controlled Compression Ignition”.

  Here, SI combustion is a mode in which the air-fuel mixture is ignited by the spark plug 16 and the air-fuel mixture is forcibly burned by flame propagation that expands the combustion region from the ignition point to the surroundings. This is a form in which the air-fuel mixture is combusted by self-ignition in an environment where the temperature of the piston 5 is increased and the pressure is increased. The SPCCI combustion, which is a mixture of SI combustion and CI combustion, SI burns a part of the air-fuel mixture in the combustion chamber 6 by spark ignition performed in an environment just before the air-fuel mixture self-ignites. This is a combustion mode in which the remaining air-fuel mixture in the combustion chamber 6 is subjected to CI combustion by self-ignition after combustion (by further increase in temperature and pressure accompanying SI combustion).

  In SPCCI combustion, heat generation during SI combustion is gentler than heat generation during CI combustion. For example, in the waveform of the heat generation rate when SPCCI combustion is performed, the rising slope is relatively small as shown in FIG. Further, the pressure fluctuation (dp / dθ) in the combustion chamber 6 is also gentler during SI combustion than during CI combustion. In other words, the waveform of the heat generation rate at the time of SPCCI combustion has a first heat generation rate portion formed by SI combustion with a relatively small rising gradient and a relatively large rising gradient formed by CI combustion. The second heat generation part is formed to be continuous in this order.

  When the temperature and pressure in the combustion chamber 6 are increased by the SI combustion, the unburned mixture is self-ignited and CI combustion is started. As illustrated in FIG. 6 or FIG. 7 described later, the slope of the heat generation rate waveform changes from small to large at the timing of this self-ignition (that is, the timing at which CI combustion starts). That is, the waveform of the heat generation rate in SPCCI combustion has an inflection point (X in FIG. 7) that appears at the timing when CI combustion starts.

  After the start of CI combustion, SI combustion and CI combustion are performed in parallel. Since CI combustion generates more heat than SI combustion, the heat generation rate is relatively large. However, since CI combustion is performed after the compression top dead center, the slope of the heat generation rate waveform does not become excessive. That is, when the compression top dead center is passed, the motoring pressure is lowered due to the lowering of the piston 5, and as a result, the increase in the heat generation rate is suppressed, so that it is avoided that dp / dθ during CI combustion becomes excessive. The Thus, in SPCCI combustion, due to the nature that CI combustion is performed after SI combustion, dp / dθ, which is an indicator of combustion noise, is unlikely to be excessive, and simple CI combustion (when all fuels are subjected to CI combustion) ), Combustion noise can be suppressed.

  As the CI combustion ends, the SPCCI combustion also ends. Since CI combustion has a higher combustion speed than SI combustion, the combustion end timing can be advanced compared to simple SI combustion (when all fuels are subjected to SI combustion). In other words, in SPCCI combustion, the combustion end timing can be brought close to the compression top dead center in the expansion stroke. Thereby, in SPCCI combustion, fuel consumption performance can be improved compared with simple SI combustion.

  In order to realize the SPCCI combustion as described above, in the first operation region A1, each part of the engine is controlled by the ECU 100 as follows.

  The injector 15 injects all or most of the fuel to be injected during one cycle during the compression stroke. For example, at the operation point P1 included in the first operation region A1, the injector 15 injects fuel in two parts from the middle stage to the latter stage of the compression stroke as shown in the chart (a) of FIG.

  The spark plug 16 ignites the air-fuel mixture in the vicinity of the compression top dead center. For example, at the operating point P1, the spark plug 16 ignites the air-fuel mixture at a timing slightly ahead of the compression top dead center. Then, SPCCI combustion is triggered by this ignition, a part of the air-fuel mixture in the combustion chamber 6 is combusted by flame propagation (SI combustion), and then the remaining air-fuel mixture is combusted by self-ignition (CI combustion). .

  The supercharger 33 is turned off. That is, the electromagnetic clutch 34 is released, the connection between the supercharger 33 and the engine body 1 is released, and the bypass valve 39 is fully opened, whereby the supercharging by the supercharger 33 is stopped.

  The intake VVT 13a and the exhaust VVT 14a are valve timings of the intake valve 11 and the exhaust valve 12 for performing internal EGR, that is, valves that both the intake and exhaust valves 11 and 12 are opened across the exhaust top dead center. The timing is set so that the overlap period is sufficiently formed. Thereby, internal EGR which makes burned gas remain in the combustion chamber 6 is implement | achieved, and the temperature (initial temperature before compression) of the combustion chamber 6 is raised.

  The throttle valve 32 is fully opened.

  The opening degree of the EGR valve 53 is controlled so that the air-fuel ratio (A / F), which is the weight ratio between the air (fresh air) in the combustion chamber 6 and the fuel, becomes a predetermined target air-fuel ratio. . Specifically, the target air-fuel ratio in the first operation region A1 is leaner (λ> 1) than the stoichiometric air-fuel ratio on the lower load side than the load line L shown in FIG. On the side, the stoichiometric air-fuel ratio or the vicinity thereof (λ≈1) is set. Note that λ is an excess air ratio, λ = 1 when the air-fuel ratio is the stoichiometric air-fuel ratio (14.7), and λ> 1 when the air-fuel ratio is leaner than the stoichiometric air-fuel ratio. The EGR valve 53 adjusts the amount of exhaust gas (external EGR gas) recirculated through the EGR passage 51 so that the amount of air (fresh air) corresponding to the target air-fuel ratio described above is introduced into the combustion chamber 6. In other words, the EGR valve 53 uses the amount of air equivalent to the target air-fuel ratio from the total amount of gas introduced into the combustion chamber 6 when the throttle valve 32 is fully open, and the burned fuel that is left in the combustion chamber 6 by the internal EGR. The flow rate in the EGR passage 51 is adjusted so that the amount of gas excluding the amount of gas is recirculated from the EGR passage 51 to the combustion chamber 6 as external EGR gas. In the first operation region A1, the air-fuel ratio (A / F) is set to the stoichiometric air-fuel ratio or leaner as described above, and EGR gas (external EGR gas and internal EGR gas) is introduced into the combustion chamber 6. Therefore, the gas air-fuel ratio (G / F), which is the weight ratio of the total gas in the combustion chamber 6 to the fuel, becomes lean in any of the first operation region A1.

  The swirl valve 18 is fully closed or closed to a low opening degree close to full closing. As a result, all or most of the intake air introduced into the combustion chamber 6 becomes intake air from the first intake port 9A (the intake port on the side where the swirl valve 18 is not provided). A strong swirl flow is formed. This swirl flow grows during the intake stroke and remains halfway through the compression stroke. For this reason, for example, when fuel is injected during the compression stroke as in the operation point P1 described above, stratification of the fuel is realized by the action of the swirl flow. That is, when fuel is injected after the middle stage of the compression stroke in a situation where a swirl flow exists, the injected fuel is collected at the center of the combustion chamber 6 where the swirl flow is relatively weak. As a result, a difference in concentration occurs in which the fuel concentration in the central portion of the combustion chamber 6 is higher than that in the outer region (outer peripheral portion), thereby realizing fuel stratification. For example, the air-fuel ratio in the central portion of the combustion chamber 6 is set to 20 or more and 30 or less, and the air-fuel ratio in the outer peripheral portion of the combustion chamber 6 is set to 35 or more.

(B) Second operation region In the second operation region A2 (region combining the low-speed / medium load region and the medium-speed region), control for performing the SPCCI combustion of the air-fuel mixture while supercharging by the supercharger 33 is executed. The Specifically, in order to realize such SPCCI combustion with supercharging, each part of the engine is controlled by the ECU 100 as follows in the second operation region A2.

  The injector 15 injects a part of the fuel to be injected during one cycle during the intake stroke and injects the remaining fuel during the compression stroke. For example, at the operation point P2 included in the second operation region A2, the injector 15 performs the first fuel injection during the intake stroke, as shown in the chart (b) of FIG. 6, during the intake stroke. At the same time, the second fuel injection for injecting a smaller amount of fuel than the first fuel injection is performed during the compression stroke. Further, at the operation point P3 at a higher load and higher rotation side than the operation point P2, the injector 15 injects fuel over a series of periods from the intake stroke to the compression stroke as shown in the chart (c) of FIG. .

  The spark plug 16 ignites the air-fuel mixture in the vicinity of the compression top dead center. For example, the spark plug 16 ignites the air-fuel mixture at a timing slightly ahead of the compression top dead center at the operating point P2 (chart (b) in FIG. 6), and from the compression top dead center at the operating point P3. The air-fuel mixture is ignited at a slightly retarded timing (chart (c) in FIG. 6). Then, SPCCI combustion is triggered by this ignition, a part of the air-fuel mixture in the combustion chamber 6 is combusted by flame propagation (SI combustion), and then the remaining air-fuel mixture is combusted by self-ignition (CI combustion). .

  The supercharger 33 is turned on. That is, supercharging by the supercharger 33 is performed by engaging the electromagnetic clutch 34 and connecting the supercharger 33 and the engine body 1. At this time, the bypass valve is set so that the pressure (supercharging pressure) in the surge tank 36 detected by the second intake pressure sensor SN7 matches the target pressure predetermined for each operating condition (rotational speed / load). The opening of 39 is controlled. For example, as the opening degree of the bypass valve 39 increases, the flow rate of the intake air that flows backward to the upstream side of the supercharger 33 through the bypass passage 38 increases, and as a result, the pressure of the intake air introduced into the surge tank 36, that is, the supercharging pressure, becomes lower. Become. The bypass valve 39 controls the supercharging pressure to the target pressure by adjusting the reverse flow rate of the intake air in this way.

  The intake valve 11 and the exhaust VVT 13a and the exhaust VVT 14a are subjected to internal EGR only at a part of the low load side of the second operation region A2 (in other words, the internal EGR is stopped at the high load side). The valve timing of the valve 12 is controlled.

  The throttle valve 32 is fully opened.

  The opening degree of the EGR valve 53 is controlled so that the air-fuel ratio (A / F) in the combustion chamber 6 becomes a predetermined target air-fuel ratio. Specifically, the target air-fuel ratio in the second operation region A2 is leaner (λ> 1) on the lower load side than the load line L (FIG. 5), or the stoichiometric air-fuel ratio on the higher load side than the load line L or Its vicinity (λ≈1) is set. The EGR valve 53 adjusts the amount of exhaust gas (external EGR gas) recirculated through the EGR passage 51 so that the amount of air (fresh air) corresponding to the target air-fuel ratio is introduced into the combustion chamber 6. For example, the recirculation amount of the exhaust gas is adjusted so as to decrease as the load increases, and is almost zero in the vicinity of the maximum load of the engine. In other words, the gas air-fuel ratio (G / F) in the combustion chamber 6 is all lean except in the vicinity of the maximum engine load.

  The swirl valve 18 is fully closed, or is opened to an appropriate intermediate opening degree other than full close / full open. Specifically, the swirl valve 18 is fully closed in a part on the low load side of the second operation region A2, and is set to an intermediate opening degree in the remaining region on the high load side. In addition, the opening degree of the swirl valve 18 in the latter area | region is enlarged, so that load is high.

(C) Third operation region In the third operation region A3 of low speed and high load, control for performing SI combustion of the air-fuel mixture while supercharging by the supercharger 33 is executed. Specifically, in order to realize such SI combustion with supercharging, each part of the engine is controlled by the ECU 100 as follows in the third operation region A3.

  The injector 15 injects a part of the fuel to be injected during one cycle during the intake stroke and injects the remaining fuel during the compression stroke. For example, at the operation point P4 included in the third operation region A3, the injector 15 performs the first fuel injection during the intake stroke, as shown in the chart (d) of FIG. 6, during the intake stroke. At the same time, the second fuel injection for injecting a smaller amount of fuel than the first fuel injection is executed in the latter half of the compression stroke (just before the compression top dead center).

  The spark plug 16 ignites the air-fuel mixture at a relatively late timing, for example, when about 5 to 20 ° CA has elapsed from the compression top dead center. Then, the SI combustion is started by this ignition, and all of the air-fuel mixture in the combustion chamber 6 is combusted by flame propagation. The reason why the ignition timing in the third operation region A3 is retarded as described above is to prevent abnormal combustion such as knocking and pre-ignition. However, in the third operation region A3, the second fuel injection is set at a rather late timing in the latter half of the compression stroke (immediately before the compression top dead center), so the ignition timing is retarded as described above. Even so, the combustion speed (flame propagation speed) after ignition is relatively high. That is, since the period from the second fuel injection to ignition is sufficiently short, the flow (turbulent energy) in the combustion chamber 6 at the time of ignition becomes relatively strong, and the combustion speed after ignition using the flow is relatively high. Can speed up. Thereby, high thermal efficiency can be maintained while preventing abnormal combustion.

  The supercharger 33 is turned on. That is, supercharging by the supercharger 33 is performed by engaging the electromagnetic clutch 34 and connecting the supercharger 33 and the engine body 1. At this time, the opening degree of the bypass valve 39 is controlled so that the pressure in the surge tank 36 (supercharging pressure) matches the target pressure.

  The throttle valve 32 is fully opened.

  The opening degree of the EGR valve 53 is controlled so that the air-fuel ratio (A / F) in the combustion chamber 6 becomes the stoichiometric air-fuel ratio or slightly richer. On the other hand, the gas air-fuel ratio (G / F) in the combustion chamber 6 is lean except for the vicinity of the maximum engine load.

  The opening degree of the swirl valve 18 is set to a predetermined intermediate opening degree (for example, 50%) or a value close thereto.

(D) Fourth operation region In the fourth operation region A4 on the higher speed side than the first to third operation regions A1 to A3, relatively orthodox SI combustion is performed. In order to realize this SI combustion, in the fourth operation region A4, each part of the engine is controlled by the ECU 100 as follows.

  The injector 15 injects at least a predetermined period that overlaps with the intake stroke. For example, at the operation point P5 included in the fourth operation region A4, the injector 15 injects fuel over a series of periods from the intake stroke to the compression stroke, as shown in the chart (e) of FIG. Since the operating point P5 is a condition of a considerably high speed and a high load, the amount of fuel to be injected in one cycle is large in the first place, and the crank angle period required to inject the required amount of fuel is prolonged. To do. This is why the fuel injection period at the operation point P5 is longer than any of the other operation points (P1 to P4) described above.

  The spark plug 16 ignites the air-fuel mixture in the vicinity of the compression top dead center. For example, at the operating point P5, the spark plug 16 ignites the air-fuel mixture at a timing slightly ahead of the compression top dead center. Then, the SI combustion is started by this ignition, and all of the air-fuel mixture in the combustion chamber 6 is combusted by flame propagation.

  The supercharger 33 is turned on. That is, supercharging by the supercharger 33 is performed by engaging the electromagnetic clutch 34 and connecting the supercharger 33 and the engine body 1. At this time, the opening degree of the bypass valve 39 is controlled so that the pressure in the surge tank 36 (supercharging pressure) matches the target pressure.

  The throttle valve 32 is fully opened.

  The opening degree of the EGR valve 53 is controlled so that the air-fuel ratio (A / F) in the combustion chamber 6 becomes the stoichiometric air-fuel ratio or slightly richer. On the other hand, the gas air-fuel ratio (G / F) in the combustion chamber 6 is lean except for the vicinity of the maximum engine load.

  The swirl valve 18 is fully opened. Thereby, not only the first intake port 9A but also the second intake port 9B is completely opened, and the charging efficiency of the engine is increased.

(4) About SI rate As mentioned above, in this embodiment, SPCCI combustion which mixed SI combustion and CI combustion is performed in the 1st operation field A1 and the 2nd operation field A2, but in this SPCCI combustion, It is important to control the ratio of SI combustion and CI combustion according to operating conditions.

  Here, in the present embodiment, the SI ratio that is the ratio of the heat generation amount by SI combustion to the total heat generation amount by SPCCI combustion (SI combustion and CI combustion) is used as the ratio. FIG. 7 is a diagram for explaining the SI rate, and shows a change in the heat generation rate (J / deg) depending on the crank angle when the SPCCI combustion occurs. The inflection point X in the waveform of FIG. 7 is an inflection point that appears when the combustion mode is switched from SI combustion to CI combustion, and the crank angle θci corresponding to this inflection point X is defined as the start timing of CI combustion. can do. Then, the area Q1 of the waveform of the heat generation rate located on the advance side with respect to θci (CI combustion start timing) is defined as the heat generation amount by SI combustion, and the waveform of the heat generation rate located on the retard side with respect to θci. Is defined as the heat generation rate by CI combustion. Thereby, the above-mentioned SI rate defined by (heat generation amount by SI combustion) / (heat generation amount by SPCCI combustion) can be expressed by Q1 / (Q1 + Q2) using the areas Q1 and Q2. . That is, in this embodiment, the SI rate = Q1 / (Q1 + Q2).

  In the case of CI combustion, the air-fuel mixture is simultaneously and frequently burned by self-ignition, so that the heat generation rate is likely to be higher than that of SI combustion by flame propagation, and large noise is likely to be generated. For this reason, it is desirable that the SI rate (= Q1 / (Q1 + Q2)) in SPCCI combustion generally increases as the load increases. This is because when the load is high, the amount of fuel injected is large and the total amount of heat generated in the combustion chamber 6 is large, so the SI rate is reduced (that is, the rate of CI combustion is increased). This is because a large noise is generated. On the contrary, since CI combustion is excellent in terms of thermal efficiency, it is preferable to perform CI combustion with as much fuel as possible unless noise becomes a problem. For this reason, it is desirable that the SI rate in SPCCI combustion is generally smaller as the load is lower (that is, the rate of CI combustion is increased). From this point of view, in this embodiment, the target SI rate (target SI rate) is the engine so that the SI rate increases as the load increases (in other words, the SI rate decreases as the load decreases). The target values of the control amounts such as the ignition timing, the fuel injection amount / injection timing, and the in-cylinder state amount are respectively determined so as to realize this target SI rate. Yes. Note that the in-cylinder state quantity referred to here is, for example, the temperature in the combustion chamber 6 or the EGR rate. The EGR rate includes the external EGR rate that is the ratio of the external EGR gas (exhaust gas recirculated to the combustion chamber 6 through the EGR passage 51) to the total gas in the combustion chamber 6, and the internal EGR to the total gas in the combustion chamber 6. The internal EGR rate which is the ratio of gas (burned gas remaining in the combustion chamber 6) is included.

  For example, the more the ignition timing is advanced, the more fuel is burned by SI combustion, and the SI rate increases. Further, as the fuel injection timing is advanced, more fuel is burned by CI combustion, and the SI rate is lowered. Alternatively, the higher the temperature of the combustion chamber 6, the more fuel is burned by CI combustion, and the SI rate is lowered.

  Based on the above trends, in this embodiment, the target values of the ignition timing, the fuel injection amount / injection timing, and the in-cylinder state quantity (temperature, EGR rate, etc.) realize the target SI rate described above. It is predetermined for each operating condition so as to be a possible combination. During operation by SPCCI combustion (that is, during operation in the first and second operation regions A1 and A2), the ECU 100 determines the injector 15, spark plug 16, EGR valve 53, intake / exhaust based on the target values of these control amounts. The VVT 13a, 14a, etc. are controlled. For example, the ignition plug 16 is controlled based on the target value of the ignition timing, and the injector 15 is controlled based on the target value of the fuel injection amount / injection timing. Further, the EGR valve 53 and the intake / exhaust VVTs 13a and 14a are controlled based on the target values of the temperature of the combustion chamber 6 and the EGR rate, and the exhaust gas (external EGR gas) recirculation amount through the EGR passage 51 and the internal EGR. Adjust the residual amount of burnt gas (internal EGR gas).

(5) Knocking avoidance control Here, in SPCCI combustion in which SI combustion and CI combustion are mixed, knocking due to each of SI combustion and CI combustion may occur. When knocking due to SI combustion is SI knock, and knocking due to CI combustion is CI knock, SI knock is an abnormal local self-ignition (normal CI) in which the unburned gas outside the region where the air-fuel mixture is SI burned. CI knock is a phenomenon of rapid combustion due to local auto-ignition that is distinctly different from combustion. ) Is a phenomenon that resonates. The SI knock appears as a large noise having a frequency of about 6.3 kHz due to air column vibrations occurring in the combustion chamber 6 due to local self-ignition. On the other hand, CI knock appears as a loud noise having a frequency of about 1 to 4 kHz (more strictly, a plurality of frequencies included in the range) due to resonance of the main components of the engine. Thus, SI knock and CI knock appear as noises of different frequencies due to different causes, and the frequency of noise at the time of CI knock is lower than the frequency of noise at the time of SI knock. .

  Since both the SI knock and the CI knock as described above are detected as noise that is annoying to the passengers in the passenger compartment, it is necessary to control the SPCCI combustion so that the SI knock and the CI knock do not occur at least continuously. is there. Therefore, in the present embodiment, the SI knock index value correlated with the SI knock and the CI knock index value correlated with the CI knock are identified based on the detected value of the in-cylinder pressure sensor SN2, and each identified knock index. SPCCI combustion is controlled based on the value. In the present embodiment, the SI knock index value is a value representing noise around 6.3 kHz that increases with the occurrence of SI knock, and the CI knock index value increases with the occurrence of CI knock. This value is representative of noise in the vicinity of 1 to 4 kHz. Each of these knock index values is calculated by Fourier transforming the waveform detected by the in-cylinder pressure sensor SN2, as will be described in detail later.

  FIG. 8 is a flowchart showing a specific procedure of control performed by ECU 100 during SPCCI combustion, that is, during operation in first operation region A1 and second operation region A2 shown in FIG. When the control shown in this flowchart is started, the ECU 100 specifies the engine rotational speed detected by the crank angle sensor SN1, the detected value (accelerator opening) of the accelerator sensor SN9, the detected value (intake flow rate) of the airflow sensor SN3, and the like. Based on the engine load to be performed, the fuel injection amount and injection timing from the injector 15 are determined (step S1). As described in (4) above, in this embodiment, the target SI rate is determined in advance according to the engine operating conditions, and the fuel injection amount / injection timing for realizing the target SI rate. Is predetermined for each engine operating condition. The fuel injection amount / injection timing determined in step S1 is the injection amount / injection timing for realizing the target SI rate.

  Next, ECU 100 determines whether or not a predetermined specific crank angle has arrived based on the detection value of crank angle sensor SN1 (step S2). The specific crank angle is predetermined as a timing for determining the ignition timing by the spark plug 16, and is set to, for example, about 60 ° CA before compression top dead center.

  When it is determined YES in step S2 and it is confirmed that the specific crank angle has arrived, the ECU 100 determines the ignition timing by the spark plug (step S3). Here, in this embodiment, since the target SI rate and the target value of the ignition timing for realizing the target SI rate are determined in advance for each engine operating condition, the ignition timing is based on this target value. Can be determined. However, the target value can be used only under the condition that the in-cylinder state quantity (temperature, EGR rate, etc.) is achieved as desired and neither SI knock nor CI knock occurs. In other words, it is necessary to shift (correct) the ignition timing from the target value depending on the in-cylinder state quantity and the presence / absence of SI / CI knock. In step S3, the ignition timing is corrected from such a viewpoint.

  FIG. 10 is a diagram for explaining a method of determining the ignition timing in step S3. As shown in the figure, in step S3, the in-cylinder state quantity at the specific crank angle point (based on the detected values of the second intake temperature sensor SN6, the second intake pressure sensor SN7, the differential pressure sensor SN8, etc.) Temperature, EGR rate, etc.) are predicted. Then, the amount of deviation between the predicted in-cylinder state quantity and the target value of the in-cylinder state quantity determined for each operating condition is examined, and based on this deviation quantity, how much the ignition timing is corrected from the target value. It is decided whether to do it. Hereinafter, the ignition timing determined based on the target value of the ignition timing and the in-cylinder state quantity will be referred to as a standard ignition timing.

  Further, when at least one of SI knock and CI knock occurs (that is, when YES is determined in any of steps S8 to S10 described later), the ignition timing is retarded (retarded) from the standard ignition timing. It is necessary to let As the retard correction amount at this time, the retard correction amount determined in steps S11 to S13 described later is employed. Then, a timing obtained by retarding the standard ignition timing by the retard correction amount is determined as the final ignition timing. When neither SI knock nor CI knock occurs, the standard ignition timing is adopted as it is as the ignition timing (final ignition timing).

  When the ignition timing is determined as described above, ECU 100 determines the fuel injection amount / injection timing determined in step S1, the ignition timing determined in step S3, and the specific crank angle point. Based on the in-cylinder state quantity and the engine rotational speed, the combustion center of gravity at the time when the fuel (50% mass) corresponding to half the mass of the fuel injected into the combustion chamber 6 during one cycle burns is calculated. Prediction is made (step S4). The prediction process in step S4 is performed based on a prediction model prepared in advance.

  Next, the ECU 100 determines the predetermined period including the combustion center of gravity predicted in step S7 as a period for detecting the in-cylinder pressure for SI / CI knock index value calculation in step S10 described later (step S5). A period for detecting the in-cylinder pressure (hereinafter referred to as a detection period) is determined to be a finite and series of periods including the combustion center of gravity. The detection period may be a period set based on time, or may be a period set based on the crank angle. When the crank angle is used as a reference, for example, the period from the crank angle advanced by 40 ° CA from the combustion center of gravity to the crank angle delayed by 40 ° CA can be determined as the detection period (see FIG. 11 described later). reference).

  The detection period determined to include the combustion center of gravity as described above includes a period in which the previous stage SI combustion occurs (hereinafter referred to as an SI combustion period) and a period in which the subsequent stage CI combustion occurs (hereinafter, referred to as “SI combustion period”). This is a period spanning both the CI combustion period). In other words, the detection period in this embodiment is a finite and series of periods that overlaps at least a part of each of the SI combustion period and the CI combustion period and includes the combustion center of gravity.

  Next, the ECU 100 causes the spark plug 16 to ignite at the ignition timing determined in step S3, and causes the air-fuel mixture to undergo SPCCI combustion using this ignition (step S6).

  Next, ECU 100 calculates an SI knock index value and a CI knock index value based on the in-cylinder pressure waveform detected by in-cylinder pressure sensor SN2 during the detection period determined in step S5 (step S7). .

  FIG. 9 is a subroutine showing a specific procedure of the control in step S7 for calculating the SI / CI knock index value. When the control shown in this subroutine starts, the ECU 100 reads the in-cylinder pressure waveform detected by the in-cylinder pressure sensor SN2 during the detection period (step S21).

  FIG. 11 is a graph showing an example of the in-cylinder pressure waveform read in step S21. The crank angle displayed on the horizontal axis of the graph is the crank angle (deg. ATDC) when the compression top dead center is 0 ° CA. In this example, the combustion center of gravity is predicted to be in the vicinity of ATDC 20 ° CA, and a period (a period from ATDC-20 ° CA to ATDC 60 ° CA) including 40 ° CA before and after the combustion center of gravity is set as the detection period. Yes. When the detection period is set in such a range, all or most of the SI combustion period and the CI combustion period are included in the detection period (in the case of FIG. 11, the entire SI / CI combustion period is the detection period). Included). Note that the noise waveform appearing just before the compression top dead center in the graph of FIG. 11 is ignition noise caused by ignition by the ignition plug 16.

  Next, the ECU 100 applies a predetermined window function to the in-cylinder pressure waveform read in step S21 to remove ignition noise (step S22).

  FIG. 12 is a graph showing the window function (hereinafter referred to as ignition noise removal window function) used in step S22. In the graph of FIG. 12, the ignition noise elimination window function is indicated by a solid line, while the window function generally used in Fourier analysis (so-called Hanning window function) is indicated by a one-dot chain line. As is apparent from the comparison between the two, the ignition noise elimination window function is modified so that the function value becomes zero over a predetermined period before and after the ignition timing, unlike the Hanning window function. By applying such an ignition noise elimination window function to the pressure waveform of FIG. 11, a pressure waveform from which ignition noise has been removed is obtained as shown by a solid line in FIG. In addition, the waveform of the dashed-dotted line of FIG. 13 is a pressure waveform at the time of applying the Hanning window function mentioned above, and it turns out that ignition noise remains in this case.

  Next, the ECU 100 performs Fourier analysis on the pressure waveform indicated by the solid line in FIG. 13, that is, the waveform obtained by applying the ignition noise removal window function to the detected waveform of the in-cylinder pressure, and obtains the amplitude for each frequency component (step) S23). FIG. 14 is a graph showing a frequency spectrum obtained by this Fourier analysis. As shown in this graph, the analysis result in step S23, that is, the spectrum (solid line) obtained by Fourier analysis of the waveform after application of the ignition noise removal function is the spectrum (dashed line) when the Hanning window function is applied. In comparison, it can be seen that the amplitude of each frequency component is greatly different. That is, according to the method of this embodiment in which the Fourier analysis is performed after the ignition noise elimination function is applied, an unnecessary frequency component due to the ignition noise is not mixed, and the original frequency component in the in-cylinder pressure waveform is accurately obtained. Can be extracted.

  Next, the ECU 100 subjects the result (frequency spectrum) of the Fourier analysis in step S23 to 1/3 octave banding (step S24). The 1/3 octave banding process is a process of dividing each octave region of the frequency spectrum (region from a certain frequency to twice that frequency) into three, and calculating the in-cylinder pressure level (CPL) of each divided band. That is. Accordingly, as shown in FIG. 15, for example, 1 kHz, 1.25 kHz, 1.6 kHz, 2 kHz, 2.5 kHz, 3.15 kHz, 4 kHz, 5 kHz, 6.3 kHz, and 8.0 kHz are center frequencies. The in-cylinder pressure level of each band is specified.

  Next, the ECU 100 calculates an SI knock index value and a CI knock index value based on the result of the 1/3 octave banding process in step S24 (FIG. 15) (steps S25 and S26). Specifically, ECU 100 calculates an in-cylinder pressure level indicated by a plot “♦” in FIG. 15, that is, an in-cylinder pressure level in a band having a center frequency of 6.3 kHz as an SI knock index value (step S25). Further, the ECU 100 sets the in-cylinder pressure level indicated by the plot “◇” in FIG. 15, that is, each band whose center frequencies are 1 kHz, 1.25 kHz, 1.6 kHz, 2 kHz, 2.5 kHz, 3.15 kHz, and 4 kHz, respectively. The average value of the in-cylinder pressure level is calculated as the CI knock index value (step S26). As described above, when SI knock occurs, noise around 6.3 kHz increases, and when CI knock occurs, noise around 1 to 4 kHz increases. Therefore, the in-cylinder pressure level in the 6.3 kHz band is at the time of SI knock occurrence. Particularly, the average value of the in-cylinder pressure level in the 1 to 4 kHz (1 kHz, 1.25 kHz,... 4 kHz) band should be particularly high when the CI knock occurs. Therefore, in the present embodiment, the former is calculated as the SI knock index value, and the latter is calculated as the CI knock index value.

  Returning to FIG. 8, ECU 100 determines whether or not the SI knock index value calculated in step S25 is larger than a predetermined threshold value α (step S8). The threshold value α used here is a threshold value for determining the presence or absence of an SI knock. If the SI knock index value is larger than the threshold value α, an SI knock occurs. If the SI knock index value is equal to or smaller than the threshold value α, an SI knock occurs. It means that you have not. The threshold value α is variable according to engine operating conditions (rotational speed / load), and is set in advance so as to increase regardless of the engine rotational speed or the load. That is, the threshold value α is lowest when the engine speed and load are low and low and low, and is highest when the engine speed and load are high and high. Is set. This is because even under low rotation and low load conditions, it is easy to detect even small noise (in other words, high noise and high load conditions are difficult to detect even with high noise).

  When it is determined YES in step S8 and it is confirmed that SI knock has occurred (SI knock index value> α), ECU 100 determines the CI knock index value calculated in step S26 in advance. It is determined whether it is larger than the threshold value β (step S9). The threshold value β used here is a threshold value for determining the presence or absence of CI knock. If the CI knock index value is larger than the threshold value β, a CI knock occurs. If the CI knock index value is equal to or smaller than the threshold value β, a CI knock occurs. It means that you have not. As with the threshold value α described above, the threshold value β is set in advance so as to increase regardless of either the engine speed or the load.

  If it is determined NO in step S9 and it is confirmed that CI knock has not occurred (CI knock index value ≦ β), that is, CI knock has not occurred but SI knock has occurred. If confirmed, the ECU 100 determines a retard correction amount for the ignition timing necessary for avoiding SI knock (step S11). Specifically, the ECU 100 obtains a difference (SI knock index value−α) between the SI knock index value calculated in step S25 and the above-described threshold value α, and is proportional to the difference (larger as the difference is larger). Determine the amount of retard correction. This is because the greater the difference between the SI knock index value and the threshold value α (that is, the stronger the SI knock), the greater the ignition retard required to avoid SI knock.

  On the other hand, if it is determined YES in step S9 and it is confirmed that CI knock has occurred (CI knock index value> β), that is, both SI knock and CI knock have occurred. If confirmed, the ECU 100 determines a retard correction amount for the ignition timing necessary to avoid both of these SI / CI knocks (step S12). Specifically, ECU 100 obtains a difference (SI knock index value−α) between the SI knock index value calculated in step S25 and the above-described threshold value α, and the CI knock index value calculated in step S26. A difference (CI knock index value−β) from the above-described threshold value β is obtained, and a retard correction amount is determined based on each obtained difference. For example, a first retard correction amount that is proportional to the difference between the SI knock index value and the threshold value α is obtained, and a second retard correction amount that is proportional to the difference between the CI knock index value and the threshold value β is obtained. The larger of the correction amounts is determined as the final retard correction amount.

  Next, the control when it is determined NO in step S8, that is, when SI knock has not occurred will be described. In this case, the ECU 100 determines whether or not the CI knock index value is larger than the threshold value β in order to confirm the presence or absence of the CI knock (step S10).

  If it is determined YES in step S10 and it is confirmed that CI knock has occurred (CI knock index value> β), that is, SI knock has not occurred but CI knock has occurred. If confirmed, the ECU 100 determines a retard correction amount for the ignition timing necessary for avoiding CI knock (step S13). The retard correction amount at this time is determined to be proportional to the difference between the CI knock index value and the threshold value β (CI knock index value−β).

  On the other hand, if it is determined NO in step S10 and it is confirmed that CI knock has not occurred (CI knock index value ≦ β), that is, it is confirmed that neither SI knock nor CI knock has occurred. If so, the ECU 100 determines whether or not the currently employed ignition timing is retarded from the standard ignition timing described above, that is, whether or not the control for retarding the ignition timing to avoid knocking is ongoing. Is determined (step S14).

  If it is determined as YES in step S14 and it is confirmed that the ignition retard is being performed, the ECU 100 decreases the retard correction amount so that the ignition timing approaches the standard ignition timing (step S15). At this time, it is conceivable to reduce the retard correction amount to zero at once (that is, immediately return the ignition timing to the standard ignition timing). However, in order to avoid the occurrence of SI knock or CI knock again after the decrease, the retard correction amount may be reduced several times. It is preferable to gradually reduce the retard correction amount in the following manner.

  After the control in any of steps S11 to S15 described above, the ECU 100 calculates the combustion center of gravity based on the in-cylinder pressure waveform during the detection period, and calculates a prediction model for the combustion center of gravity based on the calculated combustion center of gravity. Correction is made (step S16). That is, the ECU 100 predicts the amount of heat generated by combustion (heat generation amount) for each crank angle based on the waveform of the in-cylinder pressure detected by the in-cylinder pressure sensor SN2 during the detection period, and for each crank angle. Based on the data of the heat generation amount, the combustion center of gravity at the time when 50% of the fuel combusts is calculated. And based on the deviation | shift amount of this calculated combustion gravity center and the combustion gravity center estimated by said step S4, the prediction model for estimating a combustion gravity center is corrected. This correction of the prediction model leads to an improvement in accuracy when the combustion center of gravity is predicted under the same conditions after the next time (suppression of the deviation amount between the predicted value and the actual value).

(6) Operational Effects As described above, in the present embodiment, at the time of execution of SPCCI combustion in which SI combustion and CI combustion are mixed, that is, between SI knock caused by SI combustion and CI knock caused by CI combustion. During operation where both sides are concerned, the unburned gas outside the SI combustion region burns rapidly due to abnormal local self-ignition (thus generating noise near 6.3 kHz). And the CI knock index value correlated with the CI knock that causes engine parts to resonate due to the CI combustion (which generates noise in the vicinity of 1 to 4 kHz) spans both the SI combustion period and the CI combustion period. Since it is specified based on the in-cylinder pressure detected by the in-cylinder pressure sensor SN2 over a predetermined detection period (FIG. 11), the SI knock index value and the CI knock index value are determined. Both the SI knock and the CI knock can be reliably detected by the determination used. When at least one of the SI knock and the CI knock is detected, a necessary measure is taken to converge the detected knock (in the above embodiment, the ignition timing is retarded), whereby the SI knock and the CI knock are detected. Can be avoided, and the quietness of the engine can be ensured.

  In the above embodiment, the waveform of the in-cylinder pressure detected during the detection period is Fourier-analyzed, and the intensity of the frequency component correlated with SI knock, that is, the in-cylinder pressure level in the band centered on 6.3 kHz. Is calculated as the SI knock index value, and the intensity of the frequency component correlated with the CI knock, that is, the average value of the in-cylinder pressure levels in a plurality of bands having the center frequency of 1 to 4 kHz is calculated as the CI knock index value. Therefore, the degree of SI knock and CI knock with different generation causes can be accurately grasped as the intensity of different frequency components corresponding to each knock.

  In the above embodiment, a window function (ignition noise removal window function) for removing ignition noise is applied to the in-cylinder pressure detection waveform (FIG. 11) during the detection period. Since the waveform is Fourier-analyzed, it is possible to avoid extraction of extra frequency components due to ignition noise, and it is possible to accurately calculate the intensity of each frequency component described above.

  In the above embodiment, the detection period is determined to be a finite and series of periods including at least the combustion center of gravity at which 50% of the fuel supplied to the cylinder burns. Since the SI knock index value and the CI knock index value are specified based on the in-cylinder pressure detected by the internal pressure sensor SN2, the SI knock index value and the CI knock index value are determined based on the detected pressure in the period spanning both the previous SI combustion period and the subsequent CI combustion period. / CI knock index value can be calculated appropriately. Further, since the in-cylinder pressure detected during a period unrelated to SI knock and CI knock (period other than combustion) can be excluded from the analysis target for calculating the SI / CI knock index value, the SI / CI knock index value It is possible to reduce the burden required for the calculation.

(7) Modification In the above embodiment, the SI knock index value and the CI knock index value are specified based on the in-cylinder pressure detected by the in-cylinder pressure sensor SN2 (detection unit). The parameter that needs to be detected in order to be identified may be a parameter related to noise caused by combustion, and the in-cylinder pressure is just one example. For example, vibration (vibration acceleration) of the engine body 1 or noise generated from the engine body 1 may be detected as the parameter. That is, as the detection unit in the present invention, a vibration sensor that detects vibration of the engine body, a noise sensor that detects noise, and the like can be used in addition to the in-cylinder pressure sensor.

  In the above embodiment, a finite and series of periods sandwiching the combustion center of gravity is set as the detection period, and both the SI knock index value and the CI knock index value are determined based on the in-cylinder pressure data detected during the detection period. Although calculated, data of periods different from each other may be used in calculating each knock index value. For example, since CI knock occurs only during the CI combustion period, in calculating the CI knock index value, only the cylinder pressure data during the CI combustion period is considered, and the cylinder pressure detected during the SI combustion period is considered. The data of may be ignored. That is, in-cylinder pressure data necessary for calculating the CI knock index value may be data at least during the period overlapping with the CI combustion period, and detection data during the SI combustion period is not essential. On the other hand, since SI knock may occur not only in the SI combustion period but also in the CI combustion period, in calculating the SI knock index value, data in a period overlapping with both the SI combustion period and the CI combustion period is necessary. . However, it is not necessary to be a series of periods extending over both periods, and may be periods separated from each other overlapping at least a part of each of the SI combustion period and the CI combustion period.

  In the above embodiment, the average of in-cylinder pressure levels in each band of 1 kHz, 1.25 kHz, 1.6 kHz, 2 kHz, 2.5 kHz, 3.15 kHz, and 4 kHz obtained by Fourier analysis of the in-cylinder pressure detection waveform. Although the value is calculated as the CI knock index value, the maximum value of the in-cylinder pressure level of each band may be calculated as the CI knock index value instead. In addition, CI knock is a resonance phenomenon with engine parts, and the resonance frequency is naturally limited to several frequencies. Therefore, the CI knock index is obtained by using only the in-cylinder pressure level in the limited band closest to each resonance frequency. A value may be calculated. In other words, the in-cylinder pressure level in a part of the above bands that is far from the resonance frequency may be ignored.

  In the above embodiment, the supercharger 33 mechanically driven by the engine body 1 is provided in the intake passage 30. However, instead of such a mechanical supercharger 33 (supercharger), it is driven by an electric motor. An electric supercharger or a turbocharger driven by the energy of exhaust gas may be provided.

  In the above embodiment, the donut-shaped cavity 20 in plan view that surrounds the conical raised portion 20a is provided on the crown surface of the piston 5, but the concave portion of the cavity 20 that faces the spark plug 16, that is, the raised portion. The concave portion of the portion located on the intake side from 20a may be formed to be smaller than the concave portion on the opposite side (exhaust side). In this way, when the fuel is injected from the injector 15 in the later stage of the compression stroke, the fuel spray can be moved to the vicinity of the electrode of the spark plug 16 more quickly.

  In the above embodiment, when performing the internal EGR that causes the burned gas to remain in the combustion chamber 6, a valve overlap period is formed in which both the intake and exhaust valves 11 and 12 are opened across the exhaust top dead center. The intake / exhaust VVTs 13a, 14a are controlled as described above, but on the contrary, by forming a so-called negative overlap period in which both the intake / exhaust valves 11, 12 are closed across the exhaust top dead center. It is also possible to perform EGR.

  In the above embodiment, a control target value such as ignition timing is determined in advance so that the target SI rate determined for each operating condition is realized, and when knocking is detected (SI / CI knock index value is a threshold value α). Feedback control based on the SI / CI knock index value is performed so that the ignition timing is corrected in the case of exceeding (β), in addition to such control, further based on the SI rate Feedback control may be performed. That is, the SI rate in each combustion (SPCCI combustion) is calculated from the detection waveform by the in-cylinder pressure sensor SN2 each time, and when the calculated SI rate does not match the target SI rate, ignition is performed in a direction in which the deviation amount becomes smaller. The timing, the fuel injection amount / injection timing, and the like may be corrected.

  Moreover, when calculating the SI rate of each combustion as described above, various specific methods for calculating this SI rate are conceivable.

  For example, the heat generation rate at each crank angle is calculated from the detection waveform by the in-cylinder pressure sensor SN2, and the areas Q1 and Q2 shown in FIG. 7 are calculated based on the calculated heat generation rate data (waveform). May be. In this case, as already described, the SI rate = Q1 / (Q1 + Q2) can be calculated. Alternatively, the SI rate = Q1 / Q2 may be used.

  Alternatively, the SI rate may be calculated using Δθ1 and Δθ2 shown in FIG. That is, the crank angle period of SI combustion (crank angle period on the advance side from the inflection point X) is Δθ1, and the crank angle period of CI combustion (crank angle period on the retard side from the inflection point X) is Δθ2. In this case, the SI rate = Δθ1 / (Δθ1 + Δθ2) or the SI rate = Δθ1 / Δθ2.

  Alternatively, SI rate = ΔH1 / (ΔH1 + ΔH2) or SI rate = ΔH1 / ΔH2, where the peak of the heat generation rate of SI combustion is ΔH1 and the peak of the heat generation rate of CI combustion is ΔH2.

DESCRIPTION OF SYMBOLS 1 Engine body 2 Cylinder 16 Spark plug 100 ECU (calculation part, combustion control part)
SN2 In-cylinder pressure sensor (detector)

Claims (7)

An engine in which partial compression ignition combustion is performed in at least a part of the operation region in which a part of the air-fuel mixture in the cylinder is forcibly subjected to SI combustion by spark ignition and then the remaining air-fuel mixture in the cylinder is subjected to CI combustion by self-ignition A control device of
A detection unit for detecting a parameter related to noise generated by combustion in the cylinder;
The SI knock index value correlated with the SI knock in which the unburned gas outside the region where the air-fuel mixture is SI burned is rapidly burned due to abnormal local self-ignition, and the noise that is lower in frequency than the occurrence of the SI knock is the CI combustion. A CI knock index value correlated with a CI knock occurring therein, based on a detection value by the detection unit;
The calculation unit specifies the SI knock index value based on values detected by the detection unit in both the SI combustion period in which SI combustion occurs and the CI combustion period in which CI combustion occurs, and at least the CI An engine control device characterized by identifying the CI knock index value based on a value detected by the detection unit during a combustion period. The engine control device according to claim 1,
The calculation unit specifies an intensity of a predetermined first frequency component as the SI knock index value based on values detected by the detection unit in both the SI combustion period and the CI combustion period, and the CI knock An engine control device characterized in that, as an index value, the intensity of a second frequency component having a frequency lower than that of the first frequency component is specified based on at least a value detected by the detection unit during the CI combustion period. . The engine control device according to claim 2,
The calculation unit calculates the intensity of the first frequency component by performing Fourier analysis on the waveform of the detection value detected by the detection unit in both the SI combustion period and the CI combustion period, and at least the CI combustion period. An engine control device characterized in that the intensity of the second frequency component is calculated by performing Fourier analysis on a waveform of a detection value detected by the detection unit. The engine control device according to claim 3,
The calculation unit applies a window function for removing ignition noise caused by the spark ignition to the waveform of the detection value detected by the detection unit, and performs a Fourier analysis on the waveform after the window function is applied. An engine control device characterized by that. In the engine control device according to any one of claims 1 to 4,
The calculation unit overlaps at least a part of each of the SI combustion period and the CI combustion period, and covers a predetermined period including a combustion center of gravity at which 50% of the fuel supplied to the cylinder is burned. The SI knock index value is specified based on the value detected by the detection unit, and the detection unit is configured to cover all or a part of the predetermined period and overlap with at least a part of the CI combustion period. An engine control device that identifies the CI knock index value based on a detected value. The engine control apparatus according to claim 5,
The predetermined period is a finite and series of periods overlapping with both the SI combustion period and the CI combustion period,
The engine control device, wherein the calculation unit specifies both the SI knock index value and the CI knock index value based on values detected by the detection unit over the predetermined period. The engine control apparatus according to any one of claims 1 to 6,
An engine control apparatus, further comprising a combustion control unit that adjusts at least a timing of the spark ignition based on the SI knock index value and the CI knock index value specified by the arithmetic unit.