JPH1136942A - Fuel supply system of internal combustion engine - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the fluctuation of a fuel injection amount, and to improve the stability of the combustion immediately after the starting, by learning an ascending pattern of the fuel pressure immediately after the start of an engine, and determining a fuel injection time corresponding to the ascending pattern. SOLUTION: A fuel supply pump 5 driven in synchronization with the rotation of an engine 6, increases the discharge pressure from a feed pump 2 to supply the same to each fuel injection valve 8. The discharge pressure from the fuel supply pump 5 is regulated by a high pressure regulator 7. On this occasion, the learning of the ascending pattern of the fuel pressure to be supplied to the fuel injection valve 8 by the fuel supply pump 5 is performed every start of the engine. The ascending pattern to the rated pressure is changed by a fuel temperature, a cranking engine speed or the like, and the learning value of the ascending pattern of the fuel pressure is memorized corresponding to such operation conditions. The fuel pressure at that time is estimated corresponding to the fuel injection period after the starting, by using the learning pattern, and the basic fuel injection pulse width is corrected on the basis of the estimated fuel pressure.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a fuel supply device used for a direct fuel injection type spark ignition internal combustion engine or the like.

[0002]

2. Description of the Related Art As a spark ignition internal combustion engine, fuel is directly injected into a combustion chamber, and the fuel injection timing is made different between an intake stroke and a compression stroke depending on an operation state, so that the mixture in the combustion chamber is homogenized or homogeneous. 2. Description of the Related Art An in-cylinder direct fuel injection type internal combustion engine is known which is formed in a stratified form, and in particular, enables operation with an ultra-lean mixture during stratified charge combustion (see, for example, JP-A-5-79337).

In such an internal combustion engine, the fuel injection pressure is higher than that of a premixing type in which fuel is injected into an intake port because fuel is directly injected into a combustion chamber during a compression stroke. Relatively high. Therefore, the discharge pressure required for the fuel supply pump increases. And
In such a fuel supply pump, there is a period in which the discharge pressure is low until the engine enters rated operation, such as when the engine is started, but the time required to reach this rated pressure becomes longer due to the relatively high rated pressure. Take it.

[0004]

Further, this time delay until the discharge pressure of the fuel supply pump reaches the rated discharge pressure adversely affects the operability particularly when combustion is unstable, such as during cold start.

[0005] Therefore, the fuel injection amount at the time of starting the engine greatly varies not only with the fuel injection pulse width but also with the fuel pressure, and unless the pulse width is determined in consideration of the fuel pressure fluctuation, the fuel injection amount is reduced. Control cannot be performed exactly as intended, resulting in poor operability and exhaust performance immediately after starting.

Conventionally, according to Japanese Utility Model Application Laid-Open No. 6-58158, a correction value obtained by weighted averaging of the fuel supply pressure is obtained in order to prevent the fluctuation of the fuel injection amount to the internal combustion engine due to the pulsation of the fuel supply pressure. There has been proposed a method in which the discharge amount of the fuel pump is controlled based on the fuel supply pressure based on the fuel supply pressure at the time of starting due to a control response delay (time lag). Is not always reflected correctly, and in particular, correction of the fuel pressure only during a steady state cannot cope with fluctuations in the fuel injection pressure immediately after the engine is started.

The present invention has been proposed in order to solve such a problem. The fuel injection amount is determined by learning a fuel pressure increase pattern immediately after the engine is started and determining a fuel injection time in accordance with the learned pattern. And to improve the stability of combustion immediately after starting.

[0008]

According to a first aspect of the present invention, there is provided a means for detecting an operating state, a means for calculating a fuel injection amount in accordance with an operating state, and a fuel injection valve for obtaining the calculated fuel injection amount. A fuel supply device for an internal combustion engine having a means for controlling the injection time of the engine, a means for detecting start of the engine, a means for detecting fuel supply pressure, and an increase characteristic of the fuel pressure with respect to an elapsed time after engine start. There are provided learning means for storing and learning until the fuel pressure reaches a predetermined value, and correction means for correcting the fuel injection time until the fuel pressure reaches the predetermined value based on the learning value.

In a second aspect based on the first aspect, the correction means obtains a learned value of the fuel pressure at each fuel injection timing and corrects the fuel injection time based on the learned value.

In a third aspect based on the second aspect, the correcting means corrects the fuel injection pulse width by comparing the learned value of the fuel pressure with the fuel pressure during steady operation.

In a fourth aspect based on the first to third aspects, the learning means updates a learning value every time the engine is started.

In a fifth aspect based on the first to fourth aspects, the learning means selects and updates a learning value according to a fuel temperature and a battery voltage.

In a sixth aspect based on the first to fifth aspects, the learning means stores an average value of the maximum value and the minimum value of the pulsation of the fuel pressure based on the relationship between the elapsed time from the start and the start time.

According to a seventh aspect of the present invention, in the first to sixth aspects, means for learning an average value of a pulsation width of fuel pressure in a predetermined steady-state operating state of the engine, and compares the learning result with a previous learning value. Means for judging deterioration of the fuel supply pump when there is a change equal to or more than a predetermined value, and correcting the fuel injection time corrected based on the learned value of the fuel pressure rise characteristic at the time of starting the engine when judging deterioration. Correction means.

In an eighth aspect based on the seventh aspect, the learning value of the pulsation width is selected and updated in accordance with a driving state.

In a ninth aspect based on the seventh or eighth aspect, the learning value when the pulsation width changes by a predetermined value or more is not updated.

[0017]

According to the first aspect of the present invention, since the characteristic of increasing the fuel pressure after the engine is started is learned, the predicted value of the fuel pressure at each fuel injection timing after the start is calculated based on the learned value. Then, the fuel injection time (pulse width) is calculated based on the relationship with the fuel pressure so that the required fuel injection amount is obtained at that time. Therefore, even when the fuel pressure is lower than the fuel pressure in the steady state after warm-up, such as immediately after the start of the engine, the fuel injection time is corrected accordingly, and the target fuel injection amount is accurately obtained. As a result, the combustion characteristics immediately after the start are stabilized, and the warm-up can be promoted, and the exhaust gas composition and the fuel efficiency can be improved.

According to the second and third aspects of the present invention, at each fuel injection timing, the predicted value of the learned fuel pressure is compared with the fuel pressure at the time of steady operation. The fuel injection time is corrected so that the injection time becomes relatively long.

In the fourth aspect, the learning value is updated each time the engine is started, so that the discharge characteristics of the fuel supply pump can be accurately grasped based on the latest data.

In the fifth aspect, the learning value is selected and updated in accordance with the fuel temperature and the battery voltage. Therefore, even when the fuel pressure rising pattern fluctuates due to the fuel temperature and the battery voltage, the pressure rising characteristics are correctly reflected. Thus, an accurate predicted value of the fuel pressure can be obtained.

In the sixth aspect, since the pressure rise pattern is obtained from the relationship between the average value of the maximum value and the minimum value of the fuel pressure pulsation and the elapsed time, the pressure rise characteristic can be accurately calculated.

In the seventh invention, the deterioration of the fuel injection pump is judged, for example, when the pulsation width is sharply smoothed by learning the pulsation width of the fuel pressure in the steady state of the engine and comparing it with the previous value. I do. When the deterioration progresses stepwise, the fuel rise pattern after startup also fluctuates, and the rise pattern becomes more gradual than before the deterioration due to fuel leakage due to pump deterioration. When predicting fuel pressure based on
The actual fuel pressure becomes lower than the predicted value, and the fuel injection amount decreases. However, the deterioration of the fuel supply pump is determined, and the fuel injection amount after the engine is started is corrected based on the correction value according to the degree of deterioration, so that the error in the fuel injection amount is eliminated and the combustion stability after the start is improved. Can be maintained.

In the eighth aspect, the learning value of the pulsation width is selected and updated in accordance with the operation state such as the intake air amount and the number of revolutions, so that the pulsation width is learned and updated with high accuracy in accordance with the operation state. be able to.

In the ninth aspect, when the pulsation width changes by a predetermined value or more, the learning value at that time is not updated, so that the learning error can be reduced.

[0025]

Preferred embodiments of the present invention will be described below with reference to the accompanying drawings.

In FIG. 1, 1 is a fuel tank, 2 is a feed pump for pressurizing the fuel at a low pressure, and is driven to rotate by a motor 3. Reference numeral 4 denotes a low pressure regulator for regulating the discharge pressure from the feed pump 2 to a substantially constant low pressure value. Reference numeral 5 denotes a fuel supply pump that is driven to rotate in synchronization with the rotation of the engine 6, and further increases the discharge pressure from the feed pump 2. Reference numeral 7 denotes a high-pressure regulator for regulating the discharge pressure from the fuel supply pump 5 to a substantially constant high-pressure value.

The fuel discharged from the fuel supply pump 5 is supplied from a fuel injection valve 8 provided in each cylinder to the engine 6, and the fuel injection amount and the injection timing of the fuel injection valve 8 are provided as shown in FIG. Is controlled according to the injection signal from the controller 10.

Therefore, the controller 10 includes an air flow meter 11, a crank angle sensor 12, a throttle opening sensor 13, a coolant temperature sensor 14, and a fuel pressure sensor 1.
5. A signal from the fuel temperature sensor 16 or the like is input, and a voltage value from the battery 17 is also input.

The controller 10 forms a combustible mixture layer only in the vicinity of the ignition plug 9 by injecting fuel in the latter half of the compression stroke, such as when the engine is under a low load, and has an air-fuel ratio A / F = 40.
Stable stratified combustion is performed even with an ultra-lean mixture that exceeds the limit, and at high engine load, etc., fuel is injected during the intake stroke to premix fuel and air, and a uniform A mixed gas layer is formed and high output is exhibited by homogeneous combustion at the stoichiometric air-fuel ratio.

The fuel supply pump 5 is used for starting the engine.
While the discharge pressure rises to the rated pressure, the fuel pressure at that time is obtained according to the learned value of the pressure rise characteristic during this time, and the fuel injection time is calculated so that the required fuel injection amount is obtained. In the process of increasing the fuel pressure, even when the fuel pressure is lower than the rated pressure after warm-up, the fuel injection time is corrected according to the fuel pressure so that the target fuel injection amount can be accurately obtained. To control.

For this reason, the contents of the control executed by the controller 10 will be described with reference to the flowcharts of FIGS.

First, FIG. 3 is a flowchart for learning the rising characteristic (fuel pressure rise pattern) of the fuel discharge pressure of the fuel supply pump 5 at the time of starting the engine.

In step 1, it is confirmed that the starter switch has been turned on. In step 2, the fuel pressure value shown in FIG. 7 is initialized.

In step 3, the fuel pressure b and the elapsed time B from the start of cranking are read. In step 4, in order to judge fuel pressure pulsation, if the measured fuel pressure value is increasing, it is determined whether the fuel pressure value has turned to the decreasing side, or if the measured value is decreasing. If this is the case, it is determined whether or not it has increased.
When it is reversed, the process proceeds to step 5. In step 5, the number of reversals is set to n = n + 1, and the average value Pn of the fuel pressure during the period a to b until the peak of the fuel pressure is reversed is obtained as Pn = (a + b) / 2. Is calculated as Tn = (A + B) / 2.

Then, in step 6, b is replaced with a.
A is replaced with B, and the routine proceeds to step 7, where it is determined whether the fuel pressure has reached a preset rated fuel pressure, and the above operation is repeated until the fuel pressure reaches the set fuel pressure (steps 3 to 7).

Thus, the fuel pressure rise pattern after engine start is learned, that is, the average pressure in each cycle of fuel pressure pulsation starting from fuel pressure 0, and the elapsed time from the start of starting when the average value is obtained. Are respectively stored.

FIG. 4 shows how the learning value thus obtained is updated.
5 is a flowchart illustrating an operation for storing.

In step 1, the battery voltage (related to the cranking speed) and the fuel temperature (related to the viscosity) which are significantly related to the fuel rising pattern at the time of starting the engine are read. In step 2, based on these battery voltage and fuel temperature,
The storage positions of the learning values Pn and Tn are selected. The storage position is determined by a map in which the battery voltage and fuel temperature are plotted on the vertical and horizontal axes, and the fuel discharge pressure characteristics vary depending on the cranking speed at that time and the viscosity of the fuel. When the fuel temperature is low and the viscosity is high, the rising pattern also becomes gentle.

In step 3, these Pn and Tn
Is stored in the predetermined storage location, and the process is terminated.

FIG. 5 is a flowchart showing an operation for utilizing the learning result at the time of the next engine start.

In step 1, the battery voltage and the fuel temperature are read, and in step 2, the storage position of the learning value is selected based on the battery voltage and the fuel temperature at that time. In step 3, the stored learning values Pn and Tn are read out, and the process ends.

Next, FIG. 6 is a flowchart for correcting the fuel injection amount (injection time) after the engine is started according to the fuel rising pattern learned in this way.

After confirming that the starter switch has been turned on in step 1, it is determined in step 2 whether the fuel pressure is a predetermined fuel pressure at the pump rating. If the fuel pressure has reached the predetermined fuel pressure, there is no need for correction. The process ends.

On the other hand, if the fuel pressure has not reached the predetermined value, the routine proceeds to step 3. Step 3 to Step 5
Then, the time from when the starter switch is turned on to when the next fuel is injected is calculated. Therefore, in step 3, it is determined whether or not the crank angle has reached the top dead center (TPO). If the crank angle has passed through the top dead center, the engine speed Ne, the injection timing IT,
The TPO passage time Tt is read, and the time Tx up to the fuel injection timing is calculated as follows.

Tx = (720-IT) × 60 / (Ne ×
360) + Tt As shown in FIG. 8, the fuel injection is performed first after the starter switch is turned on because a predetermined crank angle, that is, fuel injection, is greater than that after the first TPO after passing through the TPO. It is just before the time IT, and this is obtained as described above.

In step 6, a learning value at the time of fuel injection (injection timing) Tx is calculated. Where x,
y represents the fuel pressure and X and Y represent the time at each fuel pressure. Therefore, the learning value is selected such that Tx is between X and Y.

Then, in step 7, the fuel pressure (pulsation average pressure) Px at the time of fuel injection is predicted as follows (see FIG. 9).

Px = (y−x) × (Tx−X) / (Y−X) + X That is, the average fuel pressure Px at the intermediate Tx between the time X and the time Y is predicted. When Px is determined, in step 8, the fuel injection pulse width Ti is corrected as follows.

Ti = Ti × (P0 / Px) ^1/2 where P0 is the fuel pressure when the discharge pressure of the fuel supply pump 5 reaches a stable predetermined value after the warm-up of the engine is completed. In the comparison, the fuel injection pulse width is modified. Therefore, as Px is smaller than P0, the fuel injection pulse width becomes larger, and even when the fuel pressure is lower than the predetermined value, the fuel injection amount accurately matches the target value.

The configuration is as described above. Next, the operation will be described.

The discharge pressure of the fuel supply pump 5 does not immediately reach the rated pressure during normal operation, such as when the engine is started at a low temperature, and increases with a certain time delay. For this reason, when the fuel pressure is lower than a predetermined value in a steady state, even if the fuel injection pulse width is the same, the fuel injection amount is reduced, and combustion stability is likely to be impaired.

However, in the present invention, every time the engine is started, learning of a pattern of increasing the pressure of the fuel supplied to the fuel injection valve 8 by the fuel supply pump 5 is performed. The rising pattern up to the rated pressure changes depending on the fuel temperature, the cranking speed, and the like at that time, and a learned value of the fuel pressure rising pattern corresponding to such an operating condition is stored.

Since this learning pattern is stored as the elapsed time after the start of the start and the fuel pressure, if the fuel injection timing (timing) after the start is determined, the fuel pressure at that time is predicted.

For this reason, if the fuel pressure at the fuel injection timing can be predicted in each cylinder, the fuel injection pulse width at that time is the basic fuel injection pulse width, that is, the pulse width when the fuel pressure reaches the rated pressure. Is corrected based on the predicted fuel pressure to prevent the fuel injection amount from decreasing below the target value.

Since the fuel pressure increase pattern is selected based on the fuel temperature and the battery voltage, even when the pressure increase characteristic changes depending on the fuel temperature and the battery voltage, an appropriate increase pattern is selected in accordance with each of them. Therefore, the fuel pressure at each fuel injection timing can be accurately predicted, and therefore, under all operating conditions, the control accuracy of the fuel injection amount immediately after the start of the engine is increased, and the operability and the exhaust composition can be improved.

Next, another embodiment will be described with reference to the flowcharts of FIGS. In this embodiment, the deterioration of the pump is determined from the pulsation state of the discharge pressure of the fuel supply pump 5 after the engine is warmed up. Is corrected. When the deterioration of the pump progresses gradually, the pulsation width gradually becomes smaller than that of a new pump, but in this case, there is no problem even if the fuel injection pulse width at the time of starting is corrected as described above. However, when the deterioration of the pump progresses stepwise, leaks and the like suddenly increase, and the fuel rising pattern at the time of starting the engine relatively decreases.

Therefore, when such stepwise deterioration of the fuel injection pump 5 is determined, a correction is made to prevent a change in the fuel injection amount due to a change in the rising pattern.

FIG. 10 is a flow chart for determining whether or not to permit learning of the fuel pressure pulsation width during normal operation after engine warm-up.

In step 1, it is determined whether the fuel temperature, the rate of change of the throttle opening ΔTVO, and the rate of change of the engine speed ΔNe are all within predetermined values. Is permitted, otherwise, learning is disallowed in step 3.

Accordingly, the learning of the pulsation width is not performed under the transient driving condition, and it is possible to remove an error factor such as a variation in the learning result.

FIG. 11 is a flowchart for learning the fuel pressure pulsation. In step 1, a flag FREV = 0 is set, in step 2, the fuel pressure Pf is read, and in step 3, the magnitude of the fuel pressure Pf is inverted. Determine if you did. In other words, as shown in FIG. 15, the fuel pressure pulsation changes from increase to decrease and from decrease to increase each time the maximum value a and the minimum value b appear. Judgment is made from the differential value of.

Steps 2 and 3 are repeated until there is a reversal. When the reversal is made, the process proceeds to step 4, and it is determined whether or not FREV = 1. , A = Pf,
Return to step 2.

Then, in step 4, FREV =
If it is 1, it is determined that the fuel pressure has been inverted twice after the flow starts, and b = Pf is set in step 6.

In step 7, the pulsation width learning value LP is calculated as follows.

LP = LP + G × (│ab│−LP) In this case, the pulsation width LP is weighted for learning, and G as a weighting constant at this time is 0 <G ≦
Let it be 1.

Then, in step 8, it is determined whether or not the number of inversions has reached a predetermined number set in advance. If the number is less than the predetermined number, the process returns to step 1 and the same operation is repeated.

After repeating the predetermined number of times in this way, it is determined that the pulsation width LP has converged to a certain range.
The process ends.

Next, FIG. 12 compares the learned value of the pulsation width performed in this way with the old learned value, and judges whether or not the fuel supply pump 5 has deteriorated stepwise based on the comparison result. It is a flow.

First, in step 1, signals representative of the operating state such as the throttle opening TVO, engine speed Ne, and air-fuel ratio are read, and in step 2, the old learned value LPO is stored based on these operating states. The read position is selected from the map. Then, after reading the old pulsation width learning value LPO in step 3, the comparison with the current pulsation width learning value LP is performed in step 5.

That is, when the difference between the old learning value and the current learning value is larger than the predetermined value DLP, it is determined that the fuel supply pump 5 has deteriorated stepwise, and the routine proceeds to step 5, where FDLP = 1 If the value is equal to or smaller than the predetermined value, the process proceeds to step 6 where FDLP = 0.

As shown in FIG. 16, the fuel supply pump 5
When the pressure is new, the wear of the plunger and the like is small, and the pressure pulsation width of the fuel in each discharge stroke is large. However, when the fuel pressure deteriorates, the pulsation width decreases. Therefore, when there is a large difference between the learning results of the pulsation widths, it can be determined that the fuel supply pump 5 has deteriorated in a stepwise manner. When the deterioration changes in a stepwise manner, the fuel pressure rise pattern at the time of starting the engine becomes slow due to fuel leakage or the like.

FIG. 13 is a flowchart for storing the learning result of the pulsation width.

At step 1, it is determined whether or not the flag FDLP is equal to "0". If not, it is determined that stepwise deterioration has occurred, and the processing is terminated without updating the learning value.
When FDLP = 0, the throttle opening T
Signals representing operating conditions such as VO, engine speed Ne, and air-fuel ratio are read, and a storage position of a learning value is selected in step 3 based on these operating conditions.
Then, the learning value LPO stored at the selected position is replaced with a new learning value LP, and the process ends.

Here, when the pulsation width changes stepwise, the learning value is not updated, and the fuel injection pulse is changed based on the pulsation width changed stepwise as shown in the flow of FIG. Correct the width to prevent fluctuations in the fuel injection amount at startup.

For this reason, in the flow of FIG. 14, it is determined in step 1 whether the flag FDLP = 1, and if so, the flow proceeds to step 2 in which the correction amount Z is obtained by table lookup. The process ends as it is.

In step 2, a correction amount Z corresponding to the difference DLP between the old learning value LPO and the current learning value LP is calculated. In step 3, the fuel injection pulse width Ti obtained according to the fuel pressure predicted value at the time of starting is corrected, and calculated as Ti = Ti × Z.

As described above, when the pulsation width of the fuel supply pump 5 after the warm-up is deteriorated in a stepwise manner, that is, when the pulsation width is sharply smoothened, the deterioration of the pump is determined, and the fuel injection pulse width at the time of starting is reduced. The correction prevents the fuel injection amount from relatively decreasing, and stabilizes the combustion characteristics.

That is, in this embodiment, the fuel pressure pulsation after warm-up is determined, and when the pulsation width becomes significantly smaller than usual, it is determined that the deterioration of the fuel supply pump 5 has progressed in a stepwise manner. I do. When the deterioration progresses stepwise in this way, the fuel rising pattern after the start also fluctuates, and the fuel rising pattern due to pump deterioration causes the rising pattern to become more gradual than before the deterioration. If the fuel pressure is predicted based on the learning pattern, the actual fuel pressure will be lower than the predicted value, and the fuel injection amount will decrease.

However, since the deterioration of the fuel supply pump 5 is judged and the fuel injection amount after the engine is started is corrected based on the correction value corresponding to the deterioration degree, the error in the fuel injection amount is eliminated.
Good combustion stability after starting can be maintained.

When the deterioration of the fuel injection pump 5 is not stepwise, that is, when the fuel injection pump 5 is gradually deteriorated, the fuel pressure rise pattern after the start is learned at each start. Even without correction, the fuel injection amount can be accurately controlled. The correction based on the learning result of the pulsation width after warm-up is performed only when the deterioration is stepwise, and it is considered that some abnormality has occurred rather than deterioration over time. However, such a situation can be dealt with immediately.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a fuel supply system according to an embodiment of the present invention.

FIG. 2 is a schematic configuration diagram of the engine.

FIG. 3 is a flowchart of a learning operation of a fuel rising pressure pattern.

FIG. 4 is a flowchart of a learning value update operation.

FIG. 5 is a flowchart of a learning value selecting operation.

FIG. 6 is a flowchart of a fuel correction operation at the time of starting.

FIG. 7 is an explanatory diagram showing a fuel pressure increase pattern at the time of starting.

FIG. 8 is an explanatory diagram showing fuel injection timing.

FIG. 9 is an explanatory diagram showing a relationship between fuel pressure and elapsed time.

FIG. 10 is a flowchart of a learning permission operation of fuel pressure pulsation according to another embodiment.

FIG. 11 is a flowchart of the learning operation.

FIG. 12 is a flowchart for determining deterioration of the fuel supply pump.

FIG. 13 is a flowchart of a learning value update operation.

FIG. 14 is a flowchart of a fuel injection amount correcting operation.

FIG. 15 is an explanatory diagram showing pressure pulsation of fuel.

FIG. 16 is an explanatory diagram showing the characteristics of the discharge pulsation of the fuel supply pump in comparison with when it is new and when it is deteriorated.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Fuel tank 5 Fuel supply pump 6 Engine 7 High pressure regulator 8 Fuel injection valve 10 Controller 11 Air flow meter 12 Crank angle sensor 13 Throttle opening sensor 14 Cooling water temperature sensor 15 Fuel pressure sensor 16 Fuel temperature sensor 17 Battery

Claims (9)

[Claims] A means for detecting an operating state; a means for calculating a fuel injection amount in accordance with the operating state; and a means for controlling an injection time of a fuel injection valve so as to obtain the calculated fuel injection amount. In the fuel supply device for an internal combustion engine, a means for detecting the start of the engine, a means for detecting the fuel supply pressure, and learning of a characteristic of increasing the fuel pressure with respect to an elapsed time after the engine is started until the fuel pressure reaches a predetermined value. A fuel supply device for an internal combustion engine, comprising: learning means for performing a fuel injection time period until the fuel pressure reaches a predetermined value based on the learning value. 2. The fuel supply system for an internal combustion engine according to claim 1, wherein said correction means obtains a learned value of the fuel pressure at each fuel injection timing, and corrects the fuel injection time based on the learned value. 3. The fuel supply device for an internal combustion engine according to claim 2, wherein said correction means corrects a fuel injection time by comparing a learned value of the fuel pressure with a fuel pressure during a steady operation. 4. The fuel supply device for an internal combustion engine according to claim 1, wherein said learning means updates a learning value every time the engine is started. 5. The fuel supply device for an internal combustion engine according to claim 1, wherein said learning means selects and updates a learning value according to a fuel temperature and a battery voltage. 6. The internal combustion engine according to claim 1, wherein the learning means stores an average value of the maximum value and the minimum value of the pulsation of the fuel pressure based on a relationship between an elapsed time from a start and an internal combustion engine. Engine fuel supply. 7. A means for learning an average value of a pulsation width of fuel pressure in a predetermined steady-state operating state of an engine, and comparing the learning result with a previous learning value and detecting a change in the fuel pressure pulsation when the average value exceeds a predetermined value. 7. A device according to claim 1, further comprising means for judging deterioration of the supply pump, and correcting means for correcting the fuel injection time corrected based on a learned value of the fuel pressure rise characteristic at the time of starting the engine when the deterioration is judged. A fuel supply device for an internal combustion engine according to one of the above aspects. 8. The fuel supply device for an internal combustion engine according to claim 7, wherein the learning value of the pulsation width is selected and updated according to an operating state. 9. The fuel supply device for an internal combustion engine according to claim 7, wherein a learning value when the pulsation width changes by a predetermined value or more is not updated.