Ignition timing control device

Abstract

To suppress degradation of performance of an engine after fuel replenishment.SOLUTION: An ECM (Engine Control Module) 17 corrects a basic ignition timing determined according to an engine operation state on the basis of a feedback correction value updated according to an occurrence state of knocking and a multipoint learning value updated according to the feedback correction value, and determines a control target value of ignition timing. The multipoint learning values are individually determined to each of the plurality of learning points distributed in a multipoint learning region. When a fuel is possibly replenished, the ECM 17 enlarges a range of a region to update the multipoint learning value in the multipoint learning region than usual.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an ignition timing control device that controls the ignition timing of an engine. [Background technology]

[0002] In engine ignition timing control, feedback correction of ignition timing depending on whether knocking occurs and learning of the steady-state deviation of the feedback correction value may be performed. Patent Document 1 describes an ignition timing control device that learns the steady-state deviation for each multi-point learning region divided by engine speed and engine load. In the ignition timing control device of Patent Document 1, when updating a learned value in a central learning region, which is the multi-point learning region currently in operation, the learned value in a peripheral learning region, which is a multi-point learning region near the central learning region, is also updated. Furthermore, in this ignition timing control device, when updating the learned value in the peripheral learning region, the farther the region is from the central learning region, the smaller the learning reflection rate. The learning reflection rate represents the ratio of the update amount of the learned value in the peripheral learning region to the update amount of the learned value in the central learning region. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-209886 Summary of the Invention [Problem to be solved by the invention]

[0004] The properties of the fuel used in the engine may change due to refueling. When the knocking resistance of the fuel changes, the learned value is updated from the value corresponding to the properties before the change to the value corresponding to the properties after the change. On the other hand, there may be cases where an infrequently used operating range and a frequently used operating range exist within the multi-point learning range. In such a case, in a multi-point learning range that exists in an infrequently used operating range, even if the properties of the fuel change, the learned value will not be updated to the value corresponding to the changed properties for a while afterwards. If the engine is operated in such an operating range where the learned value has not been updated, the ignition timing may be controlled at a time that is different from the appropriate time, and the engine may not be able to perform at its original performance. [Means for solving the problem]

[0005] The ignition timing control device that solves the above problems is: An ignition timing control device that sets a control target value of ignition timing by correcting a basic ignition timing set based on an engine operating state using a feedback correction value that is updated in accordance with the occurrence of knocking, a basic learning value that is updated based on the feedback correction value, and a multi-point learning value that is updated based on the feedback correction value, wherein the basic learning value is individually prepared for each of a plurality of basic learning regions that are partitioned according to engine speed in a coordinate system consisting of two axes of engine speed and engine load, and the multi-point learning value is prepared in a multi-point learning region that is set in a low-load region of the basic learning region that is located on the lowest rotation side. The ignition timing control device performs peripheral learning to update, based on the feedback correction value, each of the multi-point learning values ​​of the learning points within the multi-point learning region that are located at a normalized distance from an operating point where the engine is operating on the coordinate system that is less than a certain value, based on the feedback correction value, and if there is a possibility that fuel has been refilled into the fuel tank of the engine, the ignition timing control device updates all of the multi-point learning values ​​based on the feedback correction value for a period until the occurrence of knocking is confirmed.

[0006] When fuel is replenished, the knocking resistance of the fuel used in the engine may change. If there is a delay in updating the multi-point learning value to a value corresponding to the change in knocking resistance, the engine may not be able to perform at its full potential during that delay. In the above-described ignition timing control device, when there is a possibility that fuel has been replenished, All Multi-point learning values will be updated Therefore, the multi-point learned values ​​at each learning point can be quickly updated to values ​​that correspond to changes in knocking resistance due to refueling. Therefore, the above-described ignition timing control device has the effect of suppressing deterioration in engine performance after refueling. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a diagram schematically illustrating a configuration of an embodiment of an ignition timing control device; [Figure 2] FIG. 4 is a diagram showing a setting mode of a learning region of the control device. [Figure 3] FIG. 10 is a diagram showing a setting mode of a multi-point learning area. [Figure 4] 4 is a flowchart of a multi-point learning routine executed by the control device. [Figure 5] 10 is a graph showing how weighting factors are set in a multi-point learning routine. DETAILED DESCRIPTION OF THE INVENTION

[0008] An embodiment of the ignition timing control device will be described in detail below with reference to FIGS. <Configuration of ignition timing control device> First, the configuration of the ignition timing control device of this embodiment will be described with reference to FIG. 1. An engine 10 shown in FIG. 1 is mounted on a vehicle. The engine 10 has a combustion chamber 11 in which a mixture of intake air and fuel is combusted. An intake passage 12 and an exhaust passage 13 are connected to the combustion chamber 11. The intake passage 12 is an introduction passage for intake air used for combustion in the combustion chamber 11. The exhaust passage 13 is an exhaust passage for exhaust gas generated by combustion in the combustion chamber 11. The engine 10 also has a throttle valve 14 that adjusts the flow rate of intake air in the intake passage 12, and an injector 15 that injects fuel supplied from a fuel tank 15A into the intake air. The engine 10 also has an ignition device 16 that ignites the mixture introduced into the combustion chamber 11 by spark discharge. The engine 10 also has an EGR (Exhaust Gas Recirculation) system 27 that recirculates a portion of the exhaust gas into the intake air. It should be noted that there are two types of vehicle fuel available on the market: normal fuel and premium fuel, which has higher knock resistance than normal fuel. However, it is recommended to use premium fuel for this engine 10.

[0009] The engine 10 is controlled by an ECM (Engine Control Module) 17. The ECM 17 includes an execution device 18 that executes various processes for engine control, and a storage device 19 that stores programs and data for engine control. The ECM 17 starts up when an ignition switch 26 is turned on. The ECM 17 stops operating when the ignition switch 26 is turned off.

[0010] The ECM 17 is connected to various sensors installed in various parts of the vehicle. The sensors connected to the ECM 17 include an air flow meter 20, a crank angle sensor 21, a knock sensor 22, and an accelerator pedal sensor 23. The air flow meter 20 detects the intake air flow rate GA in the intake passage 12. The crank angle sensor 21 detects the rotation angle of a crankshaft 24, which is the output shaft of the engine 10. The knock sensor 22 detects whether knocking occurs in the combustion chamber 11. The accelerator pedal sensor 23 detects the accelerator pedal depression amount ACC, which is the amount of depression of an accelerator pedal 25 by the driver. The ECM 17 determines the operating state of the engine 10 based on the detection results of these sensors. The operating state of the engine 10 determined by the ECM 17 includes an engine speed NE and an engine load KL. The engine speed NE represents the rotation speed of the crankshaft 24. The engine load KL represents the intake air filling rate of the combustion chamber 11. The ECM 17 calculates the engine speed NE based on the detection result of the crank angle sensor 21. The ECM 17 also calculates the engine load KL based on the calculation result of the engine speed NE, the detection result of the air flow meter 20, etc.

[0011] The ECM 17 determines control command values ​​such as the throttle opening, fuel injection amount, ignition timing, and EGR rate based on the determined operating state of the engine 10. The throttle opening represents the opening rate of the throttle valve 14, the fuel injection amount represents the amount of fuel injected by the injector 15, and the ignition timing represents the timing at which the ignition device 16 ignites the air-fuel mixture. The EGR rate represents the proportion of EGR gas in the total amount of gas introduced into the combustion chamber 11. The ECM 17 controls the engine by outputting command signals to the throttle valve 14, the injector 15, the ignition device 16, and the EGR system 27 based on the determined control command values. In this embodiment, the ECM 17, which controls the ignition timing as part of engine control, corresponds to an ignition timing control device.

[0012] The engine 10 is mounted on a vehicle whose driving performance can be changed by switching the driving mode. A driving mode selector switch 28 is installed at the driver's seat of the vehicle. The driving modes selectable by operating the selector switch 28 include an economy mode and a power mode. A signal indicating the operating position of the selector switch 28 is input to the ECM 17. The ECM 17 then changes the control content of the engine 10 so that fuel economy performance is improved when the economy mode is selected and output performance is improved when the power mode is selected. For example, when the economy mode is selected, the ECM 17 controls the engine to increase the EGR rate compared to when the power mode is selected and to make the air-fuel ratio leaner.

[0013] <Ignition timing control overview> Next, an overview of the ignition timing control of the engine 10 executed by the ECM 17 will be described. In the ignition timing control, a final ignition timing AOP value, which is an ignition timing command value for the ignition device 16, is calculated based on the operating state of the engine 10. The ignition timing of the ignition device 16 is controlled to be more advanced as the value of this final ignition timing AOP increases. The ECM 17 calculates the final ignition timing AOP in the following procedure.

[0014] When calculating the final ignition timing AOP, the ECM 17 first calculates the MBT ignition timing and the knock limit ignition timing based on the engine load KL and engine speed NE. The MBT ignition timing is the ignition timing at which engine torque is maximized at the current engine speed NE and engine load KL. The knock limit ignition timing represents the following timing: The likelihood of knocking varies depending on environmental conditions such as air temperature, humidity, atmospheric pressure, and coolant temperature. Assume that the engine 10 is operated at the current engine speed NE and engine load KL under environmental conditions that are most likely to cause knocking. The knock limit ignition timing is the most advanced timing within the range of ignition timings at which knocking does not occur.

[0015] Next, the ECM 17 calculates the more retarded of the two timings, the MBT ignition timing and the knock limit ignition timing, as the basic ignition timing ABSE. The ECM 17 then calculates the most retarded ignition timing AKMF, which is the timing retarded from the basic ignition timing ABSE by the maximum retard amount R. The maximum retard amount R is a constant whose value is determined through experiments or the like. The most retarded ignition timing AKMF calculated in this manner represents the following timing: Assume that the engine 10 is operated at the current engine speed NE and engine load KL under environmental conditions that are most likely to cause knocking. The most retarded ignition timing AKMF is the timing most advanced within the range of ignition timings that will prevent knocking.

[0016] The ECM 17 then corrects the most retarded ignition timing AKMF using the feedback correction value AKCS and the knock learning value AGT to calculate the final ignition timing AOP. Specifically, the ECM 17 adds the knock learning value AGT to the most retarded ignition timing AKMF and subtracts the feedback correction value AKCS from the added value to calculate the final ignition timing AOP.

[0017] The feedback correction value AKCS is an ignition timing correction value that increases or decreases depending on whether knocking has occurred. The ECM 17 determines whether knocking has occurred based on a detection signal from the knock sensor 22. If the ECM 17 determines that knocking has not occurred, it gradually decreases the feedback correction value AKCS. On the other hand, if the ECM 17 determines that knocking has occurred, it gradually increases the feedback correction value AKCS.

[0018] The knock learning value AGT is a learning value for compensating for the steady-state deviation of the feedback correction value AKCS from the reference value "0." The ECM 17 calculates the knock learning value AGT by adding the basic learning value AG and the multi-point learning value AGDP (AGT←AG+AGDP).

[0019] The basic learning value AG is prepared for each of a plurality of basic learning regions RA defined according to the engine speed NE. The ECM 17 learns the basic learning value AG by gradually changing the feedback correction value AKCS in each basic learning region RA and storing the result as the basic learning value AG for the corresponding learning region.

[0020] An example of how basic learning areas RA are set is shown in Figure 2. In Figure 2, three areas, "RA1", "RA2", and "RA3", are set as basic learning areas RA. <Multi-point learning of ignition timing> Next, the multi-point learning of ignition timing will be explained. Deposits adhering inside the combustion chamber 11 affect the occurrence of knocking. The degree to which deposits affect the occurrence of knocking can vary greatly depending on the operating conditions, even within the same basic learning region RA. The multi-point learning region RB, which is the learning region for the multi-point learning value AGDP, is set to an operating region where the influence of such deposits is significant. Specifically, as shown in Figure 2, the multi-point learning region RB is set to the low-load region within the basic learning region RA1, which is located at the lowest rotation speed.

[0021] As shown in FIG. 3, a large number of learning points P are arranged within the multi-point learning region RB. The learning points P are arranged at equal intervals in both the direction of change of the engine speed NE and the direction of change of the engine load KL. "ΔN" in the figure indicates the interval between the learning points P in the direction of change of the engine speed NE. Also, "ΔK" in the figure indicates the interval between the learning points P in the direction of change of the engine load KL. A multi-point learning value AGDP is prepared for each learning point P. In this embodiment, these multiple learning points P correspond to multiple learning positions distributed within the multi-point learning region RB.

[0022] When the engine 10 is operating outside the multipoint learning region RB, the ECM 17 sets the value of the multipoint learned value AGDP to "0." On the other hand, when the engine 10 is operating within the multipoint learning region RB, the ECM 17 sets the value of the multipoint learned value AGDP as follows: That is, when the engine 10 is operating at an operating point on the learning point P, the ECM 17 uses the value of the multipoint learned value AGDP at the learning point P as is as the value of the multipoint learned value AGDP. On the other hand, when the engine 10 is operating at an operating point deviated from the learning point P, the ECM 17 calculates the value of the multipoint learned value AGDP at the operating point during operation by linear interpolation or the like, based on the multipoint learned values ​​AGDP at a plurality of learning points P neighboring the operating point during operation.

[0023] <Learning multi-point learning values> Next, learning of the multipoint learned value AGDP will be described with reference to Figures 4 and 5. When the engine 10 is operating within the multipoint learning region RB, the ECM 17 learns the multipoint learned value AGDP without learning the basic learned value AG. On the other hand, when the engine 10 is operating outside the multipoint learning region RB, the ECM 17 learns the basic learned value AG without learning the multipoint learned value AGDP.

[0024] 4 is a flowchart of a multi-point learning routine executed by the ECM 17 to learn the multi-point learned value AGDP. The ECM 17 repeatedly executes this routine at predetermined control intervals while the engine 10 is running.

[0025] When this routine starts, the ECM 17 first determines in step S100 whether an update condition for the multipoint learned value AGDP is satisfied. The requirements for the update condition to be satisfied include that the engine 10 is operating within the multipoint learning region RB. Other requirements for the update condition to be satisfied include, for example, that the operating state of the engine 10 is stable and that the warm-up of the engine 10 is complete. If the update condition is satisfied (YES), the ECM 17 proceeds to step S110. If the update condition is not satisfied (NO), the ECM 17 ends the processing of this routine for the current control cycle.

[0026] In step S110, the ECM 17 calculates a gradual change processing value AKSM of the feedback correction value AKCS. In this embodiment, the moving average value of the feedback correction value AKCS is used as the gradual change processing value.

[0027] Next, in step S120, the ECM 17 determines whether the expansion flag F is set. The expansion flag F is set when the vehicle starts a trip and when the driving mode is switched from the economy mode to the power mode. If the expansion flag F is set (YES), the ECM 17 proceeds to step S130. If the expansion flag F is not set (NO), the ECM 17 proceeds to step S150.

[0028] If the process proceeds to step S130, the ECM 17 determines in step S130 whether knocking has occurred during the period from the previous control cycle to the current control cycle. If knocking has occurred (YES), the ECM 17 clears the enlargement flag F in step S140 and then proceeds to step S150. On the other hand, if knocking has not occurred (NO), the ECM 17 proceeds to step S160.

[0029] If the process proceeds to step S150, the ECM 17 uses the normal map MAP1 to calculate the value of the weighting factor KA for each learning point P. On the other hand, if the process proceeds to step S160, the ECM 17 uses the enlarged map MAP2 to calculate the value of the weighting factor KA for each learning point P. After calculating the weighting factor KA in step S150 or step S160, the ECM 17 proceeds to step S170.

[0030] In step S170, the ECM 17 calculates the update amount of the multipoint learned value AGDP for each learning point P. The update amount for each learning point P is calculated by multiplying the weighting coefficient KA and the gradual-change processing value AKSM for that learning point P. Then, in the following step S180, the ECM 17 updates the value of the multipoint learned value AGDP for each learning point P in accordance with the update amount calculated in step S170, and then ends the processing of this routine for the current control cycle.

[0031] Next, with reference to FIG. 5, the manner in which the weighting coefficient KA is calculated in steps S150 and S160 of the multi-point learning routine will be described. The weighting coefficient KA for each learning point P is calculated based on the normalized distance D between the learning point P and the operating point where the engine 10 is operating. The normalized distance D is calculated as a value that satisfies the relationship of mathematical formula (1). In mathematical formula (1), "NX" represents the engine speed NE at the operating point where the engine 10 is operating, and "KX" represents the engine load KL at the same operating point. Furthermore, "NP" represents the engine speed NE at the learning point P, and "KP" represents the engine load KL at the learning point P. The normal map MAP1 and the expanded map MAP2 are configured as maps that use the normalized distance D between the learning point P for which the weighting coefficient KA is calculated and the operating point where the engine is operating as an argument, and return the weighting coefficient KA for that learning point P as a return value.

[0032]

number

[0033] 5 shows the relationship between the normalized distance D and the weighting factor KA in each of the normal map MAP1 and the enlarged map MAP2. The normal map MAP1 is configured so that the weighting factor KA has a value of "1" when the normalized distance D is "0" and has a value of "0" when the normalized distance D is "4" or greater. The normal map MAP1 is configured so that the value of the weighting factor KA gradually decreases as the normalized distance D increases, from a value of "1" when the normalized distance D is "0" to a value of "0" when the normalized distance D is "4", within the range of the normalized distance D from "0" to "4". On the other hand, the enlarged map MAP2 is configured so that the value of the weighting factor KA is "1" regardless of the normalized distance D.

[0034] <Effects of the embodiment> The ECM 17 corrects the basic ignition timing ABSE, which is set based on the operating state of the engine 10, with a feedback correction value AKCS and a knock learning value AGT to set the final ignition timing AOP, which is a control target value for the ignition timing. The feedback correction value AKCS is a correction value for the ignition timing that is updated depending on the occurrence of knocking. The knock learning value AGT is a learned value for the ignition timing that is updated based on the feedback correction value AKCS. In this embodiment, the knock learning value AGT is composed of a basic learning value AG and a multi-point learning value AGDP. Of these, the multi-point learning value AGDP is individually set for each of multiple learning points P distributed within a multi-point learning region RB, which is set within the operating region of the engine 10.

[0035] The ECM 17 updates the multi-point learned value AGDP in the multi-point learning routine of Fig. 4. In the multi-point learning routine, the ECM 17 calculates the update amount of the multi-point learned value AGDP for each learning point P by multiplying the gradual-change processing value AKSM of the feedback correction value AKCS by the weighting coefficient KA calculated for each learning point P. The ECM 17 calculates the weighting coefficient KA for each learning point P using the normal map MAP1 if the expansion flag F is not set, and using the expansion map MAP2 if the expansion flag F is set.

[0036] In the normal map MAP1, the weighting coefficient KA is calculated to be "0" for a learning point P that is located within a range where the normalized distance D from the operating point when the engine 10 is operating is equal to or greater than "4." When the weighting coefficient KA is "0," the update amount of the multi-point learning value AGDP is also "0." Therefore, when the expansion flag F is not set, the multi-point learning value AGDP is updated only for learning points P that are located within a range where the normalized distance D from the operating point when the engine 10 is operating is less than "4." In the following description, the simultaneous update of the multi-point learning value AGDP for each of multiple learning points P that are located within a predetermined range from the operating point when the engine 10 is operating is referred to as peripheral learning. Furthermore, the range of the operating region of the engine 10 where the multi-point learning value AGDP is updated through peripheral learning is referred to as the peripheral learning region. When the expansion flag F is not set, the peripheral learning region is the range where the normalized distance D from the operating point when the engine 10 is operating is less than "4."

[0037] On the other hand, in the enlarged map MAP2, the weighting coefficients KA of all learning points P within the multi-point learning region RB are set to "1" regardless of the normalized distance D. Therefore, when the enlargement flag F is set, the values ​​of the multi-point learning values ​​AGDP of all learning points P within the multi-point learning region RB are updated. In this way, when the enlargement flag F is set, the peripheral learning region is enlarged more than when the enlargement flag F is not set.

[0038] The expansion flag F is set when the vehicle starts a trip and when the driving mode is switched. When the vehicle starts a trip, there is a possibility that fuel tank 15A has been refilled with fuel during the soak period prior to the start of the trip. Consider a case where, after regular fuel has been used for a certain period of time, the fuel in fuel tank 15A is replaced with premium fuel through refueling. In this case, the multi-point learning value AGDP at each learning point P is a value corresponding to regular fuel at the time of fuel replacement. After the fuel is replaced, the multi-point learning value AGDP at each learning point P is updated to a value corresponding to premium fuel.

[0039] The multi-point learning region RB includes a driving region that is frequently used and a driving region that is infrequently used. In this embodiment, the region O indicated by hatching in FIG. 3 is the driving region that is frequently used. The driving regions other than region O in the multi-point learning region RB are driving regions that are infrequently used.

[0040] Consider a case where the multipoint learned value AGDP is updated without expanding the peripheral learning region after the fuel change. In this case, the multipoint learned value AGDP for the learning point P located inside and around region O is updated to a value corresponding to premium fuel in a relatively short period of time. However, the multipoint learned value AGDP for the learning point P located in a region where the fuel is used less frequently may not be updated to a value corresponding to premium fuel for some time after the fuel change. Note that knocking is less likely to occur when premium fuel is used than when regular fuel is used. Therefore, if the operating point of the engine 10 shifts to an operating region where the multipoint learned value AGDP has not yet been updated to a value corresponding to premium fuel, the final ignition timing AOP will be set to a timing that is more retarded than the timing at which it could be advanced. As a result, the engine 10 will not be able to achieve its intended fuel economy and output performance.

[0041] In contrast, in this embodiment, the ECM 17 expands the peripheral learning region to cover the entire multi-point learning region RB after the start of a trip. Therefore, the multi-point learning value AGDP at the learning point P, which is in an operating region that is used less frequently, is also updated to a value corresponding to premium fuel within a relatively short period of time. Therefore, the ignition timing control device of this embodiment is effective in suppressing performance degradation of the engine 10 after a fuel change.

[0042] If the multipoint learned value AGDP is updated until knocking occurs, it will have been updated to a value corresponding to premium fuel. Therefore, when knocking is confirmed, the ECM 17 returns the expanded peripheral learning region to the normal range. That is, in this embodiment, confirmation of knocking is set as the condition for canceling the expansion of the peripheral learning region. Thereafter, the multipoint learned value AGDP is individually updated for each learning point P so that the value is appropriate for that learning point P.

[0043] Note that if the multi-point learned value AGDP is not updated to a value corresponding to regular fuel after switching from premium fuel to regular fuel, the frequency of knocking increases. The ignition timing control device of this embodiment also has the effect of suppressing such an increase in the frequency of knocking after switching from premium fuel to regular fuel.

[0044] The torque of the engine 10 increases as the ignition timing approaches the MBT ignition timing. Therefore, in the power mode, the control of the engine 10 is set so that the ignition timing approaches the MBT ignition timing more closely than in the economy mode without causing knocking. Therefore, when the driving mode is switched from the economy mode to the power mode, the engine experiences a similar state to when switching from regular fuel to premium fuel. Therefore, even in this case, if the peripheral learning region is not expanded, the engine may not be able to achieve output performance equivalent to that of the power mode for a certain period of time after the driving mode is switched. In response to this, the ECM 17 sets the expansion flag F to expand the peripheral learning region even when the driving mode is switched from the economy mode to the power mode. Therefore, the ignition timing control device of this embodiment also has the effect of suppressing a decrease in the output performance of the engine 10 after the driving mode is switched from the economy mode to the power mode.

[0045] <Other embodiments> This embodiment can be modified as follows: This embodiment and the following modifications can be combined and implemented within the scope of technical compatibility.

[0046] In the above embodiment, the condition for canceling the expansion of the peripheral learning area was that knocking was confirmed. However, this condition may be changed. For example, the condition for canceling the expansion may be that the multipoint learned value AGDP has been updated a predetermined number of times since the first occurrence of knocking was confirmed after the expansion flag F was set. Alternatively, the condition for canceling the expansion may be that a predetermined amount of time has elapsed since the expansion flag F was set.

[0047] In the above embodiment, when the surrounding learning area is expanded, the weighting coefficients KA of all learning points P are uniformly set to "1." Even when the surrounding learning area is expanded, the weighting coefficients KA of each learning point P may be changed according to the normalized distance D from the operating point during driving. Furthermore, when the expansion flag F is not set, the weighting coefficients KA of all learning points P in the surrounding learning area may be uniformly set to "1."

[0048] In the above embodiment, when the expansion flag F is set, the peripheral learning area is expanded to cover the entire multi-point learning area RB. However, the peripheral learning area when the expansion flag F is set may be narrower than the entire multi-point learning area RB, as long as the area is wider than when the expansion flag F is not set. For example, the peripheral learning area when the expansion flag F is set may be a range in which the normalized distance D from the operating point during driving is a default value greater than "4."

[0049] In the above embodiment, the peripheral learning area was expanded at the start of a trip based on the determination that fuel may have been refueled. However, this determination may be made in other ways. For example, if there is a means for checking whether the fuel filler cap was opened or closed during the vehicle's soak period, the peripheral learning area may be expanded based on the history of the fuel filler cap being opened or closed. Furthermore, the peripheral learning area may be expanded based on the fact that the amount of fuel remaining in fuel tank 15A has increased since the end of the previous trip.

[0050] The multipoint learned value AGDP can be updated in two directions. That is, an update in a direction in which the final ignition timing AOP changes to the advance side while maintaining the basic ignition timing ABSE and the feedback correction value AKCS, and an update in a direction in which the final ignition timing AOP changes to the retard side. Here, the former is referred to as an update in the advance direction of the multipoint learned value AGDP, and the latter is referred to as an update in the retard direction of the multipoint learned value AGDP. Of these, it is the update in the advance direction that causes a delay in updating the multipoint learned value AGDP to deteriorate fuel economy and output performance. Therefore, the expansion of the peripheral learning region in the above embodiment may be applied only to the update in the advance direction, not to the update in the retard direction.

[0051] The surrounding learning area may be expanded only when there is a possibility that fuel has been replenished, rather than when the driving mode is changed. The multi-point learning region RB may be divided into multiple learning regions, and a multi-point learning value AGDP may be set for each of these learning regions. In this case, each learning region corresponds to a plurality of learning positions distributed in the multi-point learning region RB.

[0052] The entire operating range of the engine 10 may be set as the multi-point learning range RB. Alternatively, the basic learning range RA and the basic learning value AG may not be set, and the multi-point learning value AGDP may be used as the knock learning value AGT. [Explanation of symbols]

[0053] 10...Engine 11...Combustion chamber 12...Intake passage 13...Exhaust passage 14...Throttle valve 15...Injector 15A...Fuel tank 16…Ignition device 17...ECM 18...Execution device 19...Storage device 20...Air flow meter 21...Crank angle sensor 22...Knock sensor 23...Accelerator pedal sensor 24...Crankshaft 25...Accelerator pedal 26...Ignition switch 27...EGR system 28...Selector switch

Claims

1. 1. An ignition timing control device that sets a control target value of ignition timing by correcting a basic ignition timing set based on an operating state of an engine using a feedback correction value that is updated in accordance with an occurrence state of knocking, a basic learned value that is updated based on the feedback correction value, and a multi-point learned value that is updated based on the feedback correction value, the basic learning values ​​are individually prepared for each of a plurality of basic learning regions partitioned according to the engine speed in a coordinate system having two axes of the engine speed and the engine load, the multi-point learning values ​​are individually set for each of a plurality of learning points that are distributed at equal intervals in the direction of change of the engine speed and the direction of change of the engine load within a multi-point learning region that is set in a region on the low load side of the basic learning region that is located on the lowest rotation speed side, the ignition timing control device performs peripheral learning to update, based on the feedback correction value, each of the multi-point learning values ​​of the learning points, among the plurality of learning points within the multi-point learning region, whose normalized distance from an operating point where the engine is operating on the coordinate system is less than a certain value; If there is a possibility that fuel has been refilled into the fuel tank of the engine, all of the multipoint learned values ​​are updated based on the feedback correction value until occurrence of knocking is confirmed. Ignition timing control device.

2. The engine is mounted on a vehicle whose driving performance can be changed in response to switching of a driving mode, and the control content is changed in response to switching of the driving mode, The ignition timing control device updates all of the multi-point learned values ​​based on the feedback correction value until occurrence of knocking is confirmed even when the running mode is switched.

2. The ignition timing control device according to claim 1.

Images