JP2013133799A - Engine control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an engine control device estimating a signal period with high accuracy even when the variation characteristic of the signal period of a crank angle signal according to a crank angle varies.SOLUTION: The engine control device uses the signal periods of previous and present crank angle signals as reference signal periods and estimates the periods of the present and next crank angle signals by adjusting the reference signal periods by the angle characteristic factor set according to a crank angle. The engine control device increases or decreases a learning value (S520) based on whether or not a period error between the actual signal period measured actually and an estimated signal period is within a predetermined error range. The engine control device, when the absolute valve of the learning value reaches the majority of the maximum learning number or a learning number becomes equal to or larger than the maximum learning number (S522: Yes, S526: Yes), corrects an adjusting amount to the reference signal periods (S528) if the learning value is larger than a predetermined upper limit or is smaller than a predetermined lower limit.

Description

  The present invention relates to an engine control device that controls an engine based on a crank angle signal generated at a predetermined angular interval as the crankshaft of the engine rotates.

  2. Description of the Related Art Conventionally, it is known to execute control for appropriately operating an engine such as fuel injection control for a fuel injection valve based on a crank angle signal representing a crank angle that is a rotation angle of a crankshaft.

  The crank angle signal is obtained by, for example, installing a crank rotor having teeth on the outer periphery at predetermined angular intervals on the crankshaft, and setting the crank rotor teeth on the outer peripheral side of the crank rotor when the crank rotor rotates together with the crankshaft. Detected by the crank sensor, each time the tooth passes, the crank sensor outputs a pulse signal and is generated. Each pulse of the crank angle signal is generated at a predetermined angular interval such as 10 ° CA (CA: Crank Angle) or 6 ° CA according to the angular interval of teeth provided on the crank rotor.

However, in order to improve fuel economy and drivability, it is desired to perform engine control based on a crank angle that is more accurate than the crank angle represented by the crank angle signal.
Therefore, as disclosed in Patent Document 1, the signal cycle of the crank angle signal up to this time is measured, and the signal cycle of the next crank angle signal is predicted in advance based on the measured signal cycle. A technique for performing engine control at an angular position that is more detailed than an angular interval at which a crank angle signal is generated within a signal period is known.

JP 2001-263150 A

  However, the signal period of the crank angle signal changes according to the crank angle, and the change characteristic thereof changes due to engine differences or aging. Therefore, when the signal period is predicted from the measured signal period based on the change characteristic of the signal period, if the change characteristic of the signal period is changed, an error between the actual actual signal period and the predicted signal period is increased. Problems arise.

  The present invention has been made to solve the above-described problem, and provides an engine control device that predicts a signal cycle with high accuracy even if a change characteristic of a signal cycle of a crank angle signal corresponding to a crank angle changes. For the purpose.

  According to the invention as set forth in claims 1 to 45, the measuring means measures the signal period of the crank angle signal, which is a signal sequence generated at a predetermined angular interval as the crankshaft of the engine rotates, and the period predicting means comprises: Of the signal periods measured by the measuring means, the reference signal period to be adjusted is adjusted to predict a signal period after the reference signal period, and the learning means predicts the actual actual signal period and period measured by the measuring means. Learning a comparison result with a prediction signal cycle which is a signal cycle predicted by the means, correcting an adjustment amount with respect to a reference signal cycle when the cycle prediction means predicts a prediction signal cycle based on the learning result, and an angular position prediction means Finds a crank angle position that is more detailed than the angle interval of the crank angle signal in the range of the predicted signal period after correction by the learning means.

  According to this configuration, even if the change characteristic of the signal period of the crank angle signal that changes according to the crank angle changes due to engine differences or aging, the comparison result between the actual signal period and the predicted signal period is learned. Since the adjustment amount with respect to the reference signal period is corrected, the reference signal period can be adjusted to predict a signal period after the reference signal period with high accuracy.

According to the second aspect of the invention, the period predicting unit determines the adjustment amount based on the adjustment coefficient, and the learning unit corrects the adjustment amount by correcting the adjustment coefficient.
According to this configuration, since the adjustment amount is determined according to the adjustment coefficient, the process for calculating the adjustment amount can be shared.

  According to the third aspect of the present invention, the learning means learns a periodic error that is a difference between the actual signal period and the predicted signal period as a comparison result, and sets the adjustment coefficient so that the periodic error is within a predetermined error range. to correct.

According to this configuration, it is possible to correct the adjustment coefficient and reduce the cyclic error to a predetermined error range.
According to the fourth aspect of the present invention, the learning means learns the period error of the predicted signal period to be learned a plurality of times and corrects the adjustment coefficient.

  According to this configuration, even if the cyclic error varies due to measurement error, etc., the variation of the cyclic error is reduced by learning multiple times, and the adjustment amount is corrected by correcting the adjustment coefficient with high accuracy based on the learning result of the cyclic error. Can be corrected. Thereby, the signal period can be predicted with high accuracy.

  According to the fifth aspect of the invention, the learning means learns at least the number of times that the cyclic error is out of the predetermined error range and the number of times that the cyclic error is within the predetermined error range among the plurality of times of learning the cyclic error. A correction determination value for determining whether or not to correct the adjustment coefficient based on one of the numbers of times is set.

  According to this configuration, since it is only necessary to count the correction determination value to determine whether or not to correct the adjustment coefficient, the number of times that the cyclic error has deviated from the predetermined error range and the cyclic error within the predetermined error range. The correction determination value can be set by counting the number of occurrences with a simple circuit or a small processing load.

  According to the sixth aspect of the present invention, the learning means has at least a periodic error within a predetermined error range among the number of times that the cyclic error has deviated from the predetermined error range and the number of times that the cyclic error was within the predetermined error range. The correction judgment value is set based on the number of deviations, and when the correction judgment value reaches a majority of the plurality of learning times set in advance to learn the periodic error during the periodic error learning, the periodic error learning is performed. Stop and shift to adjustment coefficient correction processing.

According to this configuration, when the correction determination value reaches a majority of the number of learning times before reaching a preset number of learning times, learning of the periodic error is stopped, so that the processing time can be reduced.
According to the seventh aspect of the present invention, the learning means has at least a periodic error within a predetermined error range among the number of times that the cyclic error has deviated from the predetermined error range and the number of times that the cyclic error was within the predetermined error range. A correction determination value is set based on the number of deviations, the adjustment coefficient is increased when the correction determination value is larger than a predetermined upper limit value, and the adjustment coefficient is decreased when the correction determination value is smaller than a predetermined lower limit value.

  According to this configuration, the adjustment coefficient is increased or decreased based on the result of comparing the correction determination value and the upper limit value or the lower limit value set to determine whether the cyclic error is within an appropriate error range. The adjustment amount can be corrected by correcting the adjustment coefficient with high accuracy. Thereby, the signal period can be predicted with high accuracy.

According to the invention described in claim 8, the predetermined upper limit value and the predetermined lower limit value are arbitrarily set in advance.
According to this configuration, for example, the predetermined upper limit value and the predetermined lower limit value can be arbitrarily set appropriately for each engine model. Thus, the adjustment amount can be corrected by appropriately correcting the adjustment coefficient for each engine model.

  According to the invention described in claim 9, at least one of the predetermined upper limit value and the predetermined lower limit value is set in plural for the correction determination value, and the correction amount for correcting the adjustment coefficient is set every set number of times. Has been.

According to this configuration, since the adjustment coefficient can be appropriately corrected according to the numerical value of the correction determination value, the adjustment amount can be corrected with high accuracy. Thereby, the signal period can be predicted with high accuracy.
According to the tenth aspect of the invention, the notifying means notifies the abnormality when the periodic error is outside the predetermined error range within a range in which the learning means can correct the adjustment amount and predict the predicted signal period.

  According to this configuration, an abnormality is notified when an abnormality occurs in which the period error between the actual signal period and the predicted signal period is out of the predetermined error range even if the adjustment amount is corrected due to an abnormality in the measuring unit that measures the signal period. Therefore, appropriate measures can be taken according to the abnormality notification.

  According to the eleventh aspect of the present invention, a common adjustment coefficient is set in an angle section composed of a plurality of crank angles, and the learning means learns a comparison result at each crank angle in the angle section, Based on this, a common adjustment coefficient is corrected for the same angle section.

According to this configuration, since the same adjustment correction coefficient can be used for a plurality of crank angles, it is possible to reduce the circuit or processing load for signal period prediction and correction.
By the way, the engine rotation speed varies according to the crank angle depending on the pressure in the cylinder and the magnitude of the load applied to the engine. Therefore, the signal period also changes according to the crank angle.

Therefore, according to the twelfth aspect of the invention, the adjustment coefficient is set according to at least the crank angle of the crank angle and the engine speed.
According to this configuration, the reference signal period is appropriately adjusted by the adjustment coefficient based on the change characteristic of the signal period that changes according to the crank angle, and a signal period after the reference signal period can be predicted with high accuracy. it can.

According to the thirteenth aspect of the invention, the learning prohibiting means for prohibiting the processing by the learning means when the prohibition signal is inputted from the outside is provided.
According to this configuration, for example, when a signal cycle cannot be properly measured due to an abnormality of a sensor that detects a crank angle signal and a prohibition signal is input from the outside, the signal cycle is predicted based on an incorrect signal cycle. Therefore, it is possible to prevent engine control based on an erroneous prediction signal period.

According to the fourteenth aspect of the present invention, the learning means executes the comparison result learning process and the adjustment amount correction process at a plurality of crank angles.
According to this configuration, the engine control can be executed based on the prediction signal cycle in which the cycle error is reduced by the correction at a plurality of crank angles.

According to the fifteenth aspect of the present invention, the learning means sequentially executes the learning process and the correction process for each crank angle separated by a predetermined angular interval.
According to this configuration, the learning process and the correction process are sequentially executed for each crank angle separated by a predetermined angular interval, so that the error of the predicted signal period can be reduced with respect to the entire change characteristic of the signal period.

According to the invention described in claim 16, the crank angle at which the learning means learns the comparison result is set for each cylinder of the engine.
According to this configuration, the position of the crank angle to be learned can be set for each cylinder, which is effective when the important crank angle to be learned for each cylinder differs.

According to the seventeenth aspect of the present invention, the crank angle at which the learning means learns the comparison result is set to the same position between the cylinders of the engine.
According to this configuration, since learning is performed at the same crank angle position between the cylinders, it is possible to reduce as much as possible the cycle error between the actual signal cycle and the predicted signal cycle at the same crank angle. As a result, variations in engine control between cylinders can be reduced.

  By the way, a predetermined angle section among the angular sections in which the engine rotational speed decelerates in the compression stroke of at least one cylinder of the engine, and a predetermined section among the angular sections in which the engine rotational speed accelerates in the combustion stroke of at least one cylinder of the engine. In the angle section, since the amount of change in the engine speed is large, the amount of change in the signal period is also large.

  For this reason, when the signal period after the reference signal period is predicted by adjusting the reference signal period, the error of the predicted signal period with respect to the actual signal period increases if the change characteristic of the signal period shifts due to secular change or the like.

  Therefore, as in the inventions according to claims 18 to 21, the comparison result is learned in an angle section in which the change amount of the engine rotation speed is large and the change amount of the signal period is large, or the learning frequency is higher than in other angle sections. Or the like, the error of the prediction signal period can be reduced.

According to the twenty-second aspect of the present invention, the processing by the learning means is executed by a circuit. Thereby, the processing load by the software in an engine control apparatus can be reduced.
According to the invention of claim 23, the circuit of claim 22 is an integrated circuit. Thereby, an engine control apparatus can be reduced in size.

According to the invention as set forth in claim 24, the processing by the learning means is executed by software. Thereby, the circuit amount in an engine control apparatus can be reduced.
By the way, in the compression stroke of at least one cylinder, the deceleration amount of the engine rotation speed is increased in a predetermined angle section where the engine rotation speed is reduced, and the increase amount of the signal period of the crank angle signal is larger than before and after this section. .

  Therefore, according to the invention described in claim 25, when the period predicting means predicts the reference signal period in a predetermined angle section among the angle sections in which the engine speed is reduced in the compression stroke of at least one cylinder of the engine, The adjustment amount with respect to the reference signal period is made larger than before and after the predetermined angle section.

According to this configuration, the signal period can be predicted with high accuracy according to the change characteristic of the signal period when the engine rotation speed is reduced in the compression stroke of at least one cylinder of the engine.
According to a twenty-sixth aspect of the present invention, the prediction prohibiting means is a period predicting means when the prediction signal period is smaller than the reference signal period in an angle section where the engine speed is reduced in the compression stroke of at least one cylinder of the engine. The prediction by is prohibited.

  According to this configuration, when an abnormality occurs in which the predicted signal period obtained by adjusting the reference signal period is smaller than the reference signal period in the angle interval in which the engine rotation speed is reduced and the signal period is increased in the compression stroke, Prediction by the prediction means can be prohibited and appropriate measures can be taken.

  In addition, the predetermined angular section before and after the top dead center in the angular section of the crank angle is a section in which the engine speed that has been decelerated toward the top dead center exceeds the top dead center and turns to acceleration. This is a section where the change in rotational speed is small.

  Therefore, according to the invention as set forth in claim 27, when the cycle predicting means predicts the signal cycle of a predetermined angular section before and after the top dead center of at least one cylinder of the engine, immediately before the predetermined angular section. And the absolute value of the adjustment amount with respect to the reference signal period is made smaller than immediately after.

  Thereby, when the signal period of the angle section where the change in the engine rotational speed is small, such as a predetermined angle section before and after the top dead center including at least one cylinder of the engine, is predicted, the adjustment amount can be obtained appropriately. .

  By the way, in the combustion stroke of at least one cylinder, the acceleration amount of the engine rotation speed is reduced in a predetermined angle section in which the engine rotation speed is accelerated, and the decrease amount of the signal period is smaller than the previous time.

  Therefore, according to the invention as set forth in claim 28, when the cycle prediction means predicts the signal cycle in a predetermined angle section among the angle sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine, The absolute value of the adjustment amount with respect to the reference signal period is made smaller.

  According to this configuration, the signal according to the change characteristic of the signal cycle when the engine rotation speed is accelerated in a predetermined angle section among the angle sections in which the engine rotation speed is accelerated in the combustion stroke of at least one cylinder of the engine. The period can be predicted with high accuracy.

  According to a twenty-ninth aspect of the present invention, the prediction prohibiting means has a prediction signal period larger than a reference signal period in a predetermined angle section among the angular sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine. In this case, the prediction by the period predicting means is prohibited.

  According to this configuration, when an abnormality occurs in which the prediction signal period is longer than the reference signal period in an angle interval in which the engine rotation speed is accelerated and the signal period is shortened in the combustion stroke, the prediction by the period prediction unit is prohibited, Can be treated.

  Here, a predetermined angle section in an angular section in which the engine speed is accelerated in the combustion stroke and a predetermined angle section in an angle section in which the engine speed is decelerated in the compression stroke of the same cylinder are acceleration and deceleration. Although the change characteristic of the engine rotation speed is different, it can be considered to be the opposite change characteristic.

  Therefore, according to the invention described in claim 30, the period predicting means converts the deceleration characteristic of the engine rotational speed in a predetermined angular section among the angular sections in which the engine rotational speed decelerates in the compression stroke of at least one cylinder of the engine. Then, an adjustment amount with respect to the reference signal period is calculated in a predetermined angle section among the angle sections in which the engine speed is accelerated in the combustion stroke of the corresponding cylinder.

  As a result, the engine rotation speed in the compression stroke in an angle section in which the signal period cannot be properly predicted with the same correction method as the previous angle section, such as a predetermined angle section among the angle sections in which the engine rotation speed is accelerated in the combustion stroke. The reference signal period can be appropriately adjusted by converting the deceleration characteristics of the above to acceleration characteristics in the combustion stroke.

  In addition, in a predetermined angle section among the angle sections in which the engine speed decreases in the compression stroke, the characteristics of the amount of change in the engine speed and the previous angle section are different. It may be difficult to predict the signal period.

  Here, a predetermined angle section among the angular sections in which the engine rotational speed is decelerated in the compression stroke and a predetermined angular section among the angular sections in which the engine rotational speed is accelerated in the combustion stroke of other cylinders are deceleration and acceleration. However, it can be considered that the change characteristic of the engine speed is different, but the opposite change characteristic.

  Therefore, according to the invention described in claim 31, the cycle predicting means converts the acceleration characteristics of the engine in a predetermined angle section among the angle sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine. Then, an adjustment amount with respect to the reference signal period is calculated in a predetermined angle section among the angle sections in which the engine speed decreases in the compression stroke of the other cylinders.

  As a result, the engine rotation speed in the combustion stroke in an angle section in which the signal period cannot be appropriately predicted by the same correction method as the previous angle section, such as a predetermined angle section among the angle sections in which the engine speed decreases in the compression stroke. The reference signal period can be appropriately adjusted by converting the acceleration characteristics of the above to the deceleration characteristics in the compression stroke.

  According to the thirty-second aspect of the present invention, when the period predicting unit corrects the reference signal period by converting one of the deceleration characteristic or the acceleration characteristic, the absolute value of the adjustment amount of the reference signal period is set as the deceleration characteristic or The absolute value of the amount of change in one of the acceleration characteristics is set larger.

  As a result, the change characteristic of the engine rotation speed can be changed in the same cylinder, such as when the compression stroke is decelerated to the combustion stroke acceleration interval, or when different cylinders are shifted from the combustion stroke acceleration interval to the compression stroke deceleration interval. The reference signal cycle can be adjusted with high accuracy in the angle interval in which the absolute value of the change amount increases.

  According to a thirty-third aspect of the present invention, the period predicting means is configured such that the reference signal period is larger in an angular section where the absolute value of the rotational acceleration of the engine is equal to or smaller than a predetermined value than in an angular section where the absolute value of the rotational acceleration is larger than the predetermined value. Decrease the adjustment amount for.

  In this way, the amount of change in the engine rotational speed is small, that is, in an angle section where the rotational acceleration of the engine is less than or equal to a predetermined value, that is, the absolute value of the rotational acceleration of the engine is smaller than the predetermined value, and the signal period varies due to noise. By making the absolute value of the adjustment amount small in an easy angle section, it is possible to minimize misadjustment due to noise when adjusting the reference signal period.

  According to the thirty-fourth aspect of the present invention, when the engine rotational speed is equal to or higher than the first speed, the cycle predicting means makes the adjustment amount smaller than when the engine rotational speed is less than the first speed, and the engine rotational speed is the first. When the speed is less than the second speed, which is slower than the speed, the adjustment amount is made larger than when the engine speed is equal to or higher than the second speed.

  According to this configuration, when the engine rotation speed becomes equal to or higher than the first speed and the adjustment amount increases, the adjustment amount does not change until the engine rotation speed becomes less than the second speed that is slower than the first speed. Further, when the engine rotation speed becomes less than the second speed and the adjustment amount becomes small, the adjustment amount does not change until the engine rotation speed becomes equal to or higher than the first speed higher than the second speed.

  As a result, even if the engine rotation speed fluctuates near the first speed or the second speed that is slower than the first speed, it is possible to prevent the adjustment amount from increasing or decreasing frequently. Abnormal operation can be prevented and reliability is improved.

  According to a thirty-fifth aspect of the present invention, the period predicting means obtains a predicted signal period by adding the reference signal period and the adjustment amount. According to this configuration, the predicted signal period can be obtained by a simple calculation by addition.

  According to a thirty-sixth aspect of the present invention, the period predicting means is determined by an adjustment coefficient set according to at least the crank angle of the crank angle and the engine speed and a time change of the signal period measured by the measuring means. The adjustment amount is obtained by multiplying the change amount.

According to this configuration, the adjustment amount can be easily calculated by multiplying the adjustment coefficient by the change amount of the signal period.
According to the thirty-seventh aspect of the present invention, the adjustment coefficient is a value determined by addition / subtraction of a power of two. According to this configuration, the adjustment amount can be easily calculated by multiplying the adjustment coefficient and the change amount by shift and addition / subtraction. Further, the adjustment amount can be calculated with a simple circuit using a shifter and an adder / subtracter.

  According to a thirty-eighth aspect of the invention, the period predicting means obtains the change amount of the signal period based on the difference between the two measured signal periods. According to this configuration, the change amount can be obtained by a simple calculation of subtracting the signal period.

According to a thirty-ninth aspect of the present invention, the period predicting means uses the difference between two consecutive signal periods as the amount of change.
According to this configuration, since the amount of change in the signal period is not averaged compared to the case where the amount of change is calculated from the difference between two signal periods with one abnormal signal period in between, the signal period greatly changes. In this case, the amount of change can be obtained with high accuracy according to the change in the signal period.

  According to the invention of claim 40, the period predicting means sets an adjustment amount for adjusting the reference signal period to 0 in a predetermined angle section. According to this configuration, it is possible to easily predict the signal period in an angle section where the signal period hardly changes.

According to the invention of claim 41, the period predicting means adjusts the reference signal period among the signal periods measured by the measuring means so far, and predicts a signal period after this time.
According to this configuration, it is possible to adjust the reference signal period among the signal periods measured up to this time and predict a signal period after this time with high accuracy.

  According to the invention of claim 41, for example, by storing a signal cycle as close as possible to the current time or the current time before this time, and adjusting the reference signal cycle based on the signal cycle, Later signal periods can be predicted. As a result, in order to predict the signal period after this time for each crank angle, it is not necessary to store all the corresponding signal periods before this time, so for predicting the signal period after this time The storage capacity can be made as small as possible.

  According to the invention of claim 42, the period predicting means sets the signal period measured this time by the measuring means as the reference signal period. According to this configuration, the signal cycle after this time is predicted by adjusting the latest signal cycle measured this time, so that the prediction accuracy is improved.

  According to the invention of claim 43, the period predicting means predicts the signal period immediately after this time. According to this configuration, for example, the signal period immediately after this time can be predicted with high accuracy according to the change characteristic of the signal period measured up to this time.

  According to the invention of claim 45, the cycle calculating means determines the adjustment amount by an adjustment coefficient set in accordance with at least the crank angle of the crank angle and the engine speed, and stores a coefficient storage unit for storing the adjustment coefficient. Have.

According to this configuration, since the crank angle corresponds to the adjustment coefficient stored in the coefficient storage unit, it is not necessary to calculate the adjustment coefficient for each crank angle.
According to the invention of claim 46, the period predicting means has a selector for selecting either the reference signal period or the signal period calculated by adjusting the reference signal period by the period calculating means, Selector output is switched by angle or software command.

  According to this configuration, for example, an adjustment amount calculated when an abnormal adjustment amount different from the predicted adjustment amount is calculated, the adjustment amount calculated is invalidated, and an unadjusted reference signal cycle is selected and a signal after this time is selected. It can be a period.

The block diagram which shows the engine control apparatus by this embodiment. The block diagram which shows a period estimation part and a period learning part. The time chart which shows the relationship between a crank angle and a signal period. The flowchart which shows a period prediction process. The time chart which shows the period prediction process according to a crank angle. (A) is the time chart which shows the change of the signal period at the time of low rotation, (B) at the time of medium rotation, and (C) at the time of high rotation. The flowchart which shows another period prediction process. The flowchart which shows change amount calculation processing. The flowchart which shows other variation | change_quantity calculation processing. The flowchart which shows other variation | change_quantity calculation processing. The flowchart which shows other variation | change_quantity calculation processing. The block diagram which shows a period adjustment part. The block diagram which shows a log | history memory | storage part. The block diagram which shows a period calculation part. (A) is a block diagram showing an adjustment amount calculation unit, (B) is a diagram showing the relationship between the function of the shifters 1 and 2 and the calculated angle characteristic coefficient. The block diagram which shows another period adjustment part. The time chart which shows a learning correction process. The other time chart which shows a learning correction process. The flowchart which shows a learning correction process. The flowchart which shows a correction | amendment determination process. The flowchart which shows a correction process. (A) is a correspondence diagram showing the correspondence between the angle characteristic coefficient and the adjustment amount, (B) is a characteristic diagram showing the relationship between the periodic error and the coefficient judgment threshold, (C) is an explanatory diagram showing the change in the learning value. The other flowchart which shows a learning correction process. The other block diagram which shows a period estimation part and a period learning part.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows an engine control apparatus according to this embodiment.
(Engine control device 10)
The engine control device 10 is an electronic control unit (ECU) that executes engine control such as fuel injection control in a four-cylinder diesel engine, for example. Hereinafter, the engine control device 10 is also simply referred to as “ECU 10”.

The ECU 10 includes a microcomputer 20, an EEPROM 50, an input circuit 60, an output circuit 62, a power supply circuit 64, and the like.
The ECU 10 inputs various sensor input signals and switch input signals from the input circuit 60, and outputs an output signal to other ECUs and a control signal to the injector 90 from the output circuit 62.

  The microcomputer 20 includes a CPU 22, a ROM 24, a RAM 26, an A / D 28, an I / O 30, a timer unit 40, and the like. The CPU 22 executes a control program stored in the ROM 24, and outputs various sensor detection signals, detection signals of the crank angle sensor 80 and the cam angle sensor 82, and input signals of switches and the like from the input circuit 60. The signal is input via O30, and an output signal to another ECU and a control signal to the injector 90 are output from the output circuit 62 via the I / O 30.

  The CPU 22 uses the RAM 26 as a storage area for various control work data, and data that needs to be stored even if the power supplied from the battery 70 is shut off by the power supply circuit 64 in the EEPROM 50 is stopped in the EEPROM 50. Remember.

  The timer unit 40 measures the signal period of the crank angle signal, which is a signal sequence input from the crank angle sensor 80, predicts the subsequent signal period, and corrects the predicted signal period predicted based on the period error. The prediction unit 100 is included. Details of the angle prediction unit 100 will be described later.

(Crank angle)
The crank angle sensor 80 is an electromagnetic pickup type sensor, and is installed facing the outer periphery of the crank rotor 4 fixed to the crankshaft of the engine. Teeth are formed on the outer periphery of the crank rotor 4 at predetermined angular intervals, for example, at 10 ° intervals. The crank angle sensor 80 outputs a pulsed crank angle signal at a position facing the teeth of the crank rotor 4.

The ECU 10 calculates the rotation angle per unit time as the engine rotation speed based on the number of pulses of the crank angle signal per unit time.
The cam angle sensor 82 is an electromagnetic pickup sensor that is the same as the crank angle sensor 80, and is installed to face the outer periphery of the cam rotor 6 fixed to the cam shaft that rotates once while the crank shaft rotates twice.

  On the outer periphery of the cam rotor 6, teeth indicating a specific cylinder position among the four cylinders are formed. As for the cam angle signal output from the cam angle sensor 82, one cycle is 720 ° CA in which the crankshaft rotates twice.

  The ECU 10 detects, for example, the top dead center (TDC) of the # 1 cylinder as a reference angular position based on the crank angle signal and the cam angle signal. Then, the number of pulses, which is the number of crank angle signal pulses from the reference angle position, is counted at a 720 ° CA period, which is an angular period of one combustion cycle including an intake stroke, a compression stroke, a combustion stroke, and an exhaust stroke. Based on this pulse count, the ECU 10 detects a crank angle that indicates which angular position in which stroke each cylinder is in one combustion cycle.

(Angle prediction unit 100)
Next, the angle prediction unit 100 of the timer unit 40 will be described. As shown in FIG. 2, the angle prediction unit 100 includes a cycle prediction unit 110 and a cycle learning unit 300. First, the period prediction unit 110 will be described. The period learning unit 300 will be described after the description of the period prediction unit 110 based on FIGS.

(Period prediction unit 110)
The cycle prediction unit 110 includes a cycle measurement unit 120, a cycle adjustment unit 130, a learning control unit 132, a multiplication angle clock generation unit 140, and the like.

(Periodic measurement unit 120)
The period measuring unit 120 includes a counter 122 and a measuring unit 124. The counter 122 measures the number of pulses of the crank angle signal from the reference angle position of the crank angle signal every 720 ° CA. The crank angle is detected based on the pulse count number of the crank angle signal measured by the period measuring unit 120.

  The measuring unit 124 measures the time interval between the pulses of the crank angle signal generated every 10 ° according to the pulse count number of the crank angle signal with a timer, thereby determining the signal period of the crank angle signal (the signal of the crank angle signal). The period is simply referred to as “signal period”). The timer actually measures the period from the clock immediately after the crank angle signal is generated to the clock immediately after the next crank angle signal is generated.

(Cycle adjustment unit 130)
The cycle adjustment unit 130 detects the current crank angle of each cylinder from the pulse count number of the current crank angle signal measured by the cycle measurement unit 120, and calculates an adjustment amount calculation method for adjusting the signal cycle measured this time. It is determined according to the crank angle. Then, the cycle adjustment unit 130 calculates an adjustment amount based on the signal cycle of the crank angle signal measured so far by the current calculation method, adjusts the current signal cycle, and predicts the next signal cycle. In addition, not only the signal period measured this time but the reference signal period to be adjusted when the signal period is predicted is also referred to as a reference signal period.

  When the correction amount for the adjustment amount is input from the correction determination unit 320 of the period learning unit 300 described later, the cycle adjustment unit 130 corrects the adjustment amount for the reference signal cycle when predicting the signal cycle at the corresponding clan angle. Predict the signal period.

  The learning control unit 132 outputs a learning permission signal that instructs whether or not to allow the learning process in the periodic learning unit 300. The learning permission signal is 0 when an abnormal prediction signal period is predicted by the period adjustment unit 130, for example, when the prediction signal period is larger than the reference signal period in an angle interval where the prediction signal period should be smaller than the reference signal period. The learning process in the period learning unit 300 is prohibited.

(Multiplication angle clock generator 140)
The multiplication angle clock generation unit 140 generates a multiplication angle clock having a period obtained by dividing the prediction signal period predicted by the period adjustment unit 130 by the multiplication number. The ECU 10 executes engine control such as fuel injection control in synchronization with a crank angle that is more detailed than the angular interval of teeth formed on the crank rotor 4 based on the multiplication angle clock output from the multiplication angle clock generator 140. . Thereby, fuel consumption and drivability can be improved.

  In addition, by executing engine control based on the generated multiplication angle clock, it is possible to reduce the process of calculating the execution timing of the engine control from the predicted next signal cycle, and highly accurate engine control can be performed with a small processing load. Is possible.

  1 and 2 described above, the ECU 10 corresponds to the engine control device of the present invention, the cycle measuring unit 120 corresponds to the measuring unit of the present invention, and the cycle adjusting unit 130 corresponds to the cycle predicting unit of the present invention. The multiplication angle clock generator 140 corresponds to the angular position predicting means of the present invention.

(Change in signal cycle)
Next, changes in the signal period of the crank angle signal will be described. The rotational speed of the engine varies greatly under the influence of the in-cylinder pressure in the compression stroke and the combustion stroke of each cylinder. Therefore, the engine speed, that is, the change in the signal period, has a strong relationship with the crank angle as shown in FIG. FIG. 3 shows the relationship between the signal period and the crank angle of each cylinder in a 4-cylinder engine. Ignition is in the order of # 1 cylinder → # 3 cylinder → # 4 cylinder → # 2 cylinder.

  The signal period varies greatly depending on the load torque received by the engine and the in-cylinder pressure of each cylinder, particularly the in-cylinder pressure in the compression stroke and the combustion stroke. For example, when the # 1 cylinder is in the compression stroke, the in-cylinder pressure of the # 1 cylinder increases as the piston approaches # 1 TDC, and the resultant force of the load torque and the in-cylinder pressure prevents the piston from rising. Slow down. As a result, the signal period of the crank signal output at a predetermined angular interval becomes longer.

  When the crank angle exceeds # 1 TDC of the # 1 cylinder, the # 1 cylinder shifts to the combustion stroke. Then, the in-cylinder pressure that has acted in the direction of decelerating the rotation of the crankshaft before TDC acts in the direction of accelerating the rotation after TDC. At this time, the in-cylinder pressure and the load torque act in opposite directions. Since the effective work of the engine becomes the largest when the in-cylinder pressure becomes maximum after about 10 ° CA from TDC, the in-cylinder pressure of the injector 90 is maximized after about 10 ° CA from TDC in consideration of the combustion delay. The injection timing is controlled. Therefore, in the example of FIG. 3, at the time when the crank angle signal immediately after TDC before about 10 ° CA has elapsed from TDC is output, the pressure increase due to combustion is small and the rotational change is small.

  Thereafter, when the in-cylinder pressure of the # 1 cylinder explosively increases due to combustion, the engine speed is rapidly accelerated and the signal period is shortened. As described above, the engine rotation speed and the signal period largely change in the acceleration direction or the deceleration direction before and after the angle section near the TDC due to the influence of the in-cylinder pressure in the compression stroke and the combustion stroke of each cylinder.

  The in-cylinder pressure of the # 1 cylinder that increased explosively due to combustion gradually decreases as the piston descends. On the other hand, in the # 3 cylinder, the compression stroke proceeds, and the in-cylinder pressure of the # 3 cylinder gradually increases. In the vicinity of the intermediate angle between # 1 TDC and # 3 TDC, the in-cylinder pressure of the # 1 cylinder and the in-cylinder pressure of the # 3 cylinder are balanced with the load torque, and the change in the engine speed becomes small. In this angular section, irregular noise components such as crankshaft torsion and mechanical vibration appear.

  When the compression stroke of # 3 cylinder further proceeds, the resultant force of the in-cylinder pressure of # 3 cylinder and the load torque exceeds the in-cylinder pressure of # 1 cylinder, and the engine speed starts to decelerate again. Thus, there is a strong relationship between the engine speed and signal cycle variation characteristics and the crank angle.

(Cycle prediction process 1)
FIG. 4 shows a flowchart for executing the signal cycle prediction process according to the present embodiment. In FIG. 4 and the flowcharts of each figure described below, “S” represents a step. The flowchart of FIG. 4 is executed at the timing when the pulse of the crank angle signal is detected.

  In FIG. 4, the ECU 10 first reads the current crank angle θ (S400). As described above, the crank angle θ is detected based on the pulse count number of the crank angle signal.

  Next, as the signal period measured up to this time, for example, an amount of change α between the signal period of the crank angle signal of this time and the previous time and the signal period of the crank angle signal of the previous time and the previous time is calculated (S402). The angle characteristic coefficient K (θ) set according to the crank angle is read (S404). Each signal period measured so far to correct the signal period is stored in a register or the like, and the angle characteristic coefficient is stored in a rewritable nonvolatile EEPROM or the like. The angular characteristic coefficient K (θ) may be read before the change amount α is calculated.

  The angle characteristic coefficient is set in advance according to the crank angle based on the relationship between the signal cycle change characteristic and the crank angle described above. The angle characteristic coefficient is a reference signal among the signal periods measured by the period measuring unit 120 when a signal period later than this time is predicted based on the signal period of the crank angle signal measured by the period measuring unit 120 until this time. It is a coefficient that determines the adjustment amount for the period.

Next, the ECU 10 multiplies the change amount α and the angle characteristic coefficient K (θ) to calculate the adjustment amount H from the following equation (1) (S406).
Adjustment amount H = change amount α × angle characteristic coefficient K (θ) (1)
The ECU 10 reads the reference signal period T (i) from a register or the like (S408), adds the adjustment amount H to the reference signal period T (i), and a prediction signal that is the signal period of the current and next crank angle signals. The period T ′ (i + 1) is calculated (S410).

The multiplication angle clock generation unit 140 generates a multiplication angle clock having a period obtained by dividing the prediction signal period calculated in the period prediction process 1 of FIG. 4 by the multiplication number.
Thus, by determining the adjustment amount H of the reference signal cycle based on the current change amount α and the angle characteristic coefficient K (θ) set in accordance with the crank angle, the change in the signal cycle in accordance with the crank angle is determined. Based on the characteristics, the next signal cycle after this time can be predicted with high accuracy.

  For example, in a predetermined angle section among the angle sections in which the engine rotation speed is reduced and the signal period becomes longer in the compression stroke of FIG. 3, the deceleration amount of the engine rotation speed is larger than the preceding and subsequent angle sections. When the next signal period is predicted in the angle section, the adjustment amount for adjusting the reference signal period is set larger than before and after the predetermined angle section.

  Specifically, in FIG. 5, when predicting the predicted signal periods T ′ (3) and T ′ (4) in the angle interval in which the signal periods T (2) and T (3) are measured, the signal period T (3 ), T (1) and T (4) before and after the angle section for measuring T (4), the angle characteristic coefficient is made larger than the prediction of T ′ (2) and T ′ (5), and the adjustment amount is increased. It is getting bigger. In FIG. 5, the prediction formula for predicting T ′ (2) at the timing of measuring T (1) is not described, but it is assumed to be smaller than 1.25.

  Further, in the predetermined angular section among the angular sections in which the engine rotational speed is accelerated and the signal period is shortened in the combustion stroke of FIG. 3, the acceleration amount of the engine rotational speed is smaller than the previous time. When the signal cycle is predicted, the absolute value of the adjustment amount for correcting the reference signal cycle is made smaller than the previous time.

  Specifically, in FIG. 5, when predicting the predicted signal periods T ′ (8) and T ′ (9) in the angle interval for measuring the signal periods T (7) and T (8), respectively, When T (6) and T (7) are measured, the angle characteristic coefficient is made smaller than when T ′ (7) and T ′ (8) are predicted, and the absolute value of the adjustment amount is made smaller.

  Further, in the predetermined angular section before and after each TDC in FIG. 3, the absolute value of the change amount of the engine rotational speed, that is, the absolute value of the engine rotational acceleration is equal to or smaller than the predetermined value. Therefore, when the signal period of the predetermined angle section is predicted, the absolute value of the adjustment amount is made smaller than immediately before and immediately after the predetermined angle section.

  Specifically, in FIG. 5, prediction signal periods T ′ (5) and T ′ (6), which are prediction periods of signal periods T (5) and T (6) in a predetermined angular section before and after TDC, are represented. When obtaining the signal periods T (4) and T (5) at the timing of measurement, the predicted signal period T ′ (4) of the signal period T (4) immediately before the signal period T (5) and the signal period T (6 The angle characteristic coefficient is made smaller and the absolute value of the adjustment amount is made smaller than when the prediction signal period T ′ (7) of the signal period T (7) immediately after) is predicted.

(Relationship between engine speed and signal cycle)
FIG. 6 shows the change characteristics of the signal period when the engine speed is low (A), medium (B), and high (C).

  In FIG. 6, Tmax, Tave, and Tmin respectively indicate the maximum value, average value, and minimum value of the signal cycle when the engine rotation state is constant and the engine is in steady operation. The signal cycle is affected by the in-cylinder pressure in the compression stroke and the combustion stroke of each cylinder, and changes in a cycle of 720 / k [° CA] (k is the number of cylinders).

  When the engine is running at a low speed, the in-cylinder pressure acts for a long time, so the change in the engine speed is large. As the engine speed increases to medium and high, the change in the engine speed becomes small.

  The dominant factors that change the signal period during low and medium rotations are the in-cylinder pressure and load torque during the compression and combustion strokes, and are characterized by regular, relatively smooth and large changes depending on the crank angle. On the other hand, the periodic rotation change is small at high rotations, and irregular noise components are dominant.

  This irregular noise component is caused by crankshaft torsion or mechanical vibration. When the signal cycle changes smoothly at low rotation, the next signal cycle can be predicted to some extent from the signal cycle up to this time. Components may be emphasized and errors may increase.

  Therefore, in this embodiment, by setting the angle characteristic coefficient of the signal period in accordance with the engine rotation speed in addition to the crank angle, for example, the adjustment amount of the signal period can be reduced in a high rotation region that exceeds a predetermined rotation speed. Deterioration of signal cycle prediction accuracy due to noise components is reduced. In addition to the noise reduction processing using the angle characteristic coefficient, high frequency components may be removed by filtering processing such as averaging.

  According to the cycle prediction process 1 in FIG. 4 described above, the change amount of the signal cycle is calculated from the signal cycle measured so far, and the calculated change amount is multiplied by the angle characteristic coefficient set according to the crank angle. Thus, the adjustment amount is calculated. Then, the adjustment amount is added to the reference signal cycle to be adjusted among the signal cycles measured so far, and the reference signal cycle is adjusted according to the crank angle. Thereby, based on the signal period measured until this time, the signal period after this time can be predicted with high accuracy according to the crank angle.

  In addition, since the signal period after this time can be predicted based on the three signal periods of the current time, the previous time and the previous time, which are measured up to this time, the storage capacity for storing the signal period measured up to this time can be increased. Can be reduced.

  In the cycle prediction process 1 of FIG. 4 described above, the processing of S400 to S410 corresponds to the function executed by the cycle prediction means of the present invention. In the description of FIG. 4, the angle characteristic coefficient corresponds to the adjustment coefficient of the present invention.

(Cycle prediction process 2)
FIG. 7 shows a flowchart of the cycle prediction process 2 for setting the angle characteristic coefficient of the signal cycle according to the engine rotation speed in addition to the crank angle described above.

7, S420, S424, S430, and S432 correspond to S400, S402, S408, and S410 of FIG. 4 and are substantially the same processing.
In the cycle prediction process 2 of FIG. 7, in addition to reading the crank angle θ in S420, the engine speed NE is read in S422. Then, an angle characteristic coefficient K (θ, NE) set in advance according to the crank angle θ and the engine speed NE is read (S426), and the change amount α is multiplied by the angle characteristic coefficient K (θ, NE). The adjustment amount H is calculated (S428).

  In this way, by calculating the adjustment amount H using the angle characteristic coefficient K (θ, NE) set in advance according to the crank angle θ and the engine rotational speed NE, for example, at the time of a high speed of a predetermined speed or higher. The value of the angle characteristic coefficient is made smaller than that at the time of low rotation and middle rotation to reduce the absolute value of the adjustment amount. Thereby, at the time of high rotation, the deterioration of the prediction accuracy of the signal period due to noise components can be reduced.

  When the engine rotational speed is equal to or higher than the first speed, the angle characteristic coefficient is made smaller than when the engine rotational speed is less than the first speed to reduce the absolute value of the adjustment amount of the signal period. When the speed is less than the second speed, which is lower than the first speed, it is desirable to increase the angle characteristic coefficient and increase the absolute value of the adjustment amount as compared with the case where the engine speed is equal to or higher than the second speed.

  Thereby, when the engine rotation speed becomes equal to or higher than the first speed and the adjustment amount increases, the adjustment amount does not change until the engine rotation speed becomes less than the second speed that is slower than the first speed. Further, when the engine rotation speed becomes less than the second speed and the adjustment amount becomes small, the adjustment amount does not change until the engine rotation speed becomes equal to or higher than the first speed higher than the second speed.

  As a result, even if the engine speed fluctuates near the first speed or the second speed, it is possible to prevent the adjustment amount from frequently increasing or decreasing, thus preventing the occurrence of abnormal operation caused by frequent increase or decrease in the adjustment amount. And reliability is improved.

In the period prediction process 2 of FIG. 7 described above, the processes of S420 to S432 correspond to the function executed by the period prediction means of the present invention.
(Change amount calculation process 1)
FIG. 8 shows a flowchart of the change amount α calculation process executed in S402 of FIG. 4 and S424 of FIG.

  The ECU 10 reads the two signal periods T (im) and T (in) of the crank angle signal measured up to this time (S440, S442), and the change amount α of the signal period from the following equation (2). Is calculated (S444). In Expression (2), the difference α is obtained by dividing the difference between the two signal periods T (im) and T (in) by the absolute value of the difference in the order in which the two signal periods are measured. Is calculated.

α = (T (im) −T (in)) / | n−m | (2)
The signal period T (i) represents the time interval between the previous and current crank angle signal pulses. The change amount α is obtained, for example, by calculating the difference between the latest two consecutive signal periods. In this case, m = 0, n = 1, and | n−m | = 1.

  The denominator of equation (2) is one interval to average the difference between two signal periods separated by a plurality of intervals and to correspond to the next signal period after this time when | n−m | ≧ 2. This is a term for converting to a difference in signal period.

  Irregular noise components can be reduced by dividing the difference between the signal periods with one or more signal periods between them in the order in which the signal periods are measured. For example, when the change amount α is obtained from the latest two consecutive signal periods, that is, when m = 0, n = 1, and | n−m | = 1, for example, the noise component N is present in the signal period T (i). If it is included, the change amount α is calculated by the following equation (3).

α = T (i) −T (i−1) + N (3)
In Expression (3), the noise component N is added to the change amount α as it is. On the other hand, when the change amount α is calculated from two signal periods separated from each other by one interval, that is, when m = 0, n = 2, and | n−m | = 2, the change amount α is expressed by the following equation (4). Calculated.

α = {(T (i) −T (i−2)) / 2} + N / 2 (4)
In the equation (4), the noise component N can be halved. Therefore, if the interval between the two signal periods is further separated, the influence of the noise component can be further reduced.

  However, if the interval between the two signal periods is too far apart, the change characteristics of the signal period are averaged and cannot be appropriately adapted to the change characteristics of the signal period measured so far. Therefore, it is necessary to appropriately set the number of intervals between the signal periods based on the generation amount of the noise component and the change characteristics of the signal period.

  In the equation (2), instead of dividing the difference between signal periods with one or more signal periods between them by | n−m |, the angle characteristic coefficient K is larger than when dividing by | n−m |. The noise component may be reduced by setting to a small value. In this case, the amount of change α is calculated by the following equation (5) even if one or more signal periods are interposed therebetween.

α = T (im) −T (in) (5)
In the change amount calculation process 1 of FIG. 8 described above, the processes of S440 to S444 correspond to the function executed by the period predicting means of the present invention.

(Change amount calculation process 2)
FIG. 9 shows a flowchart of another calculation process of the change amount α. In the change amount calculation process 2 in FIG. 9, the next prediction is performed in the angle interval in which the compression stroke ends in FIG. 3 and the engine rotation speed rapidly increases from the TDC to the combustion stroke, specifically in the angle interval A in FIG. 5. The change amount calculation process for predicting the signal period is added to the change amount calculation process 1 in FIG.

In S450 of FIG. 9, the ECU 10 determines whether or not the crank angle θ detected this time is included in the angle section A of FIG.
When the crank angle θ is not included in the angle section A (S450: No), the ECU 10 executes the same processing as S440 to S444 in FIG. 8 in S452 to S456.

  When the crank angle θ is included in the angle section A (S450: Yes), the ECU 10 reads the two signal periods T (ip) and T (iq) in the deceleration section of the compression stroke (S458, S460). ), The change amount α is obtained from the following equation (6) (S462).

α = (T (ip) −T (iq)) / | q−p | (6)
In the angle section A of FIG. 5, when p = 2 and q = 1 are set and the current signal period T (6) is measured, the amount of change is calculated from the following equation (7).

α = (T (4) −T (5)) / | 1 | (7)
That is, in the angle interval A, instead of reading the signal period measured this time and the previous time in the normal angle interval and calculating the change amount from the difference, the signal period measured the previous time and the previous time is read, and the signal period When calculating the difference, the order is reversed. Accordingly, the sign of the change amount α is “negative” in the equation (7).

Thus, when the change amount is calculated in order to predict the next signal cycle in the angle section A, the change amount α is calculated by converting the deceleration characteristic in the compression stroke into the acceleration characteristic of the combustion stroke.
In the angular interval A of the combustion stroke in FIG. 5, the next prediction signal cycle T ′ (7) is calculated from the following equation (8) from the variation calculated in the variation calculation processing 2 in FIG.

T ′ (7) = T (6) + {(T (4) −T (5)) / | 1 |} × 1.5 (8)
Thus, in the angular interval A of the combustion stroke, the angle characteristic coefficient is set to 1.5, so that the amount of change in the signal period (T (5) −T (4)) in the deceleration interval of the compression stroke is absolute. The absolute value | (T (4) −T (5)) / 1 | × 1.5 of the adjustment amount is made larger than the value.

In the variation calculation process 2 of FIG. 9 described above, the processes of S450 to S462 correspond to the function executed by the period predicting means of the present invention.
(Change amount calculation process 3)
FIG. 10 shows a flowchart of another calculation process of the change amount α. In the change amount calculation process 3 in FIG. 10, in FIG. 3, the acceleration interval ends in the combustion stroke of one cylinder, the deceleration interval of the compression stroke of the other cylinder starts, and the change in engine rotation speed is greater than the angle interval before and after this. In the small angle section B (see also the angle section B in FIG. 5), the change amount calculation process for predicting the next prediction signal cycle is added to the change amount calculation process 2 in FIG.

  In S470 of FIG. 10, the ECU 10 determines whether or not the crank angle θ detected this time is included in the angle section A of FIG. When the crank angle θ is included in the angle section A (S470: Yes), the ECU 10 executes the same processing as S458 to S462 of FIG. 9 in S472 to S476.

  When the crank angle θ is not included in the angle section A (S470: No) and is not included in the angle section B (S478: No), the ECU 10 executes the same processing as S452 to S456 in FIG. 9 in S480 to S484. .

  When the crank angle θ is not included in the angle section A (S470: No) and is included in the angle section B (S478: Yes), the ECU 10 has two signal periods T (i -R) and T (is) are read (S486, S488), and the change amount α is obtained from the following equation (9) (S490).

α = (T (ir) −T (is)) / | r−s | (9)
In the angle section B shown in FIG. 5, r = 0 and s = 2 are set, and the amount of change is calculated from the following equation (10).

α = (T (i) −T (i−2)) / | 2 | (10)
As described above, in the angle section B, the difference between two signal periods with one or more signal periods sandwiched between them is divided by the difference in the order in which the signal periods are measured, and is described by Expression (4). As described above, the influence of noise components can be reduced.

In the variation calculation process 3 of FIG. 10 described above, the processes of S470 to S490 correspond to the function executed by the period predicting means of the present invention.
(Change amount calculation process 4)
FIG. 11 shows a flowchart of another calculation process of the change amount α.

  S500 to S504 in FIG. 11 are the same processes as S440 to S444 in FIG. In S506, the ECU 10 determines whether or not the value of the change amount α calculated in S504 is negative.

  If the value of the change amount α is positive (S506: No), the signal period is predicted in the deceleration zone, and the predicted signal period is considered to be larger than the reference signal period. At this time, if the detected crank angle θ indicates an acceleration section (S508: Yes), it is determined that the amount of change calculated in S504 is abnormal due to the occurrence of noise or sensor abnormality, and the amount of change α is set to “0”. (S510), and prediction is prohibited.

  If the value of the change amount α is negative (S506: Yes), the signal period is predicted in the acceleration section, and the predicted signal period is considered to be smaller than the reference signal period. At this time, if the detected crank angle θ indicates a deceleration zone (S512: Yes), the change amount calculated in S504 is determined to be abnormal due to the occurrence of noise or sensor abnormality, and the change amount α is set to “0”. (S510), and prediction is prohibited.

  In the change amount calculation process 4 in FIG. 11, the change amount α calculated in the deceleration interval of the compression stroke and the acceleration interval in the combustion stroke in FIG. 3 satisfies the change amount considered to be calculated in the corresponding angle interval. If not, it is determined that the amount of change caused by noise or sensor abnormality is abnormal, and prediction is prohibited. Therefore, the next predicted signal cycle has the same value as the signal cycle (reference signal cycle) measured this time. As a result, erroneous adjustment of the signal period can be prevented.

  In the change amount calculation process 4 of FIG. 11 described above, the processes of S500 to S504 correspond to the functions executed by the cycle predicting means of the present invention, and the processes of S506 to S510 are functions performed by the predictive prohibiting means of the present invention. Equivalent to.

(Circuit configuration of the cycle adjustment unit)
FIG. 12 shows a cycle adjustment unit 150 which is an example of a circuit configuration of the cycle adjustment unit 130 shown in FIG. The cycle adjustment unit 150 includes an angle determination unit 152, a history storage unit 154, a change amount calculation unit 156, and a cycle calculation unit 158.

  The angle determination unit 152 inputs the pulse count number of the crank angle signal measured by the period measurement unit 120, and determines which angle position of which stroke of each cylinder the crank angle detected this time is.

  Then, the angle determination unit 152 outputs a history output control signal to the history storage unit 154 in accordance with the crank angle detected this time. The history output control signal designates the reference signal cycle output from the history storage unit 154, the history signal cycles 1 and 2, and changes the signal cycle stored in the history storage unit 154 by the change amount calculation unit 156. Specifies the preprocessing to be performed on the signal period to output as history signal periods 1 and 2 used to calculate.

  Further, the angle determination unit 152 outputs an angle characteristic coefficient selection signal for selecting an angle characteristic coefficient to be used when the period calculation unit 158 calculates the adjustment amount according to the crank angle detected this time.

  The history storage unit 154 stores a plurality of signal cycles measured by the cycle measuring unit 120 so far in a register or the like. Based on the history output control signal output from the angle determination unit 152, the reference signal cycle to be output and the history signal cycles 1 and 2 are selected, and the preprocessing is performed on the signal cycle output as the history signal cycles 1 and 2. To do.

The change amount calculation unit 156 calculates the change amount α described so far from the signal cycle up to this time stored in the history storage unit 154.
The period calculation unit 158 selects an angle characteristic coefficient corresponding to the crank angle based on the angle characteristic coefficient selection signal output from the angle determination unit 152, and selects the change amount α calculated by the change amount calculation unit 156 and the selected angle. Multiply by the characteristic coefficient and add to the reference signal period measured this time to calculate the predicted signal period.

  In FIG. 12 described above, the cycle adjustment unit 150 corresponds to the cycle prediction unit of the present invention, the angle determination unit 152 corresponds to the angle detection unit and control unit of the present invention, and the history storage unit 154 stores the history storage of the present invention. The change amount calculation unit 156 corresponds to a change amount calculation unit of the present invention, and the cycle calculation unit 158 corresponds to a cycle calculation unit of the present invention.

(Circuit configuration of history storage unit)
FIG. 13 shows a history storage unit 160 which is an example of the circuit configuration of the history storage unit 154 shown in FIG. The history storage unit 160 stores the signal period measured by the period measurement unit 120 in the history registers 162a, 162b, and 162c. In the present embodiment, three history registers are provided. The number of history registers may be four or more depending on the number of signal periods necessary for correcting the reference signal period.

  The latest signal cycle measured this time is stored in the history register 162a, and the signal cycles previously stored in the history registers 162a and 162b are sequentially transferred from the history register 162b to the history register 162c and from the history register 162a to the history register 162b. Moved.

  The MUX (Multiplexer) 166 selects a history register that stores a signal period necessary for calculating the change amount from the history registers 162a, 162b, and 162c based on the history output control signal output from the angle determination unit 152. And a reference signal cycle is output. The reference signal cycle may be a signal cycle measured this time among the history registers 162a, 162b, and 162c, or may be a signal cycle other than that.

  The shifters 164a and 164b are based on the history output control signal output from the angle determination unit 152, and the signal period indicates the difference between the signal periods with one or more signal periods between the signal periods output from the MUX 166. When dividing in the measured order, a shift process corresponding to the division is executed. Therefore, it is desirable that the difference between the signal periods with one or more signal periods in between is a power of 2.

  When the angle characteristic coefficient is reduced instead of reducing the amount of change, that is, the amount of adjustment by dividing the difference between the signal periods with one or more signal periods in between in the order in which the signal periods are measured, The shifters 164a and 164b in FIG. 13 can be omitted.

In FIG. 13 described above, the history storage unit 160 corresponds to the history storage unit of the present invention, and the history registers 162a to 162c correspond to the periodic storage unit of the present invention.
(Circuit configuration of period calculation unit)
FIG. 14 shows a cycle calculation unit 170 which is an example of a circuit configuration of the cycle calculation unit 158 shown in FIG. The counter 172 of the period calculation unit 170 counts the angle characteristic coefficient selection signal output from the angle determination unit 152, actually the number of pulses of the pulse signal. Then, the angle characteristic coefficient storage unit 174 sequentially outputs the angle characteristic coefficients stored in the order of use according to the crank angle, using the count number counted by the counter 172 as a pointer.

  The angle characteristic coefficient storage unit 174 includes a rewritable storage unit such as a register. Therefore, the angle characteristic coefficient stored in the angle characteristic coefficient storage unit 174 can be adjusted by software without adjusting the angle characteristic coefficient storage unit 174 or software.

  The fixed coefficient storage unit 176 stores a fixed angle characteristic coefficient (hereinafter also referred to as “fixed coefficient”). When a fixed angle characteristic coefficient is used for a plurality of crank angles, the fixed coefficient is stored in the fixed coefficient storage unit 176 as a fixed coefficient, whereby the storage capacity of the angle characteristic coefficient storage unit 174 is stored. Can be reduced.

  For example, 1.0 is normally stored as an angle characteristic coefficient in the fixed coefficient storage unit 176 as an angle characteristic coefficient, and the angle corresponding to the crank angle stored in the angle characteristic coefficient storage unit 174 at other crank angles. Use characteristic factors.

  The selector 178 selects the fixed coefficient output from the fixed coefficient storage unit 176 when the angle characteristic coefficient selection signal pulse is “0”, and the angle characteristic coefficient selection signal pulse is generated. In the case of “1”, the angle characteristic coefficient output from the angle characteristic coefficient storage unit 174 is selected.

  The adjustment amount calculation unit 180 calculates the adjustment amount by multiplying the change amount α output from the change amount calculation unit 156 of FIG. 12 by the angle characteristic coefficient output from the selector 178 by a multiplier or the like. This adjustment amount is, for example, a positive value in the deceleration zone of the compression stroke and a negative value in the acceleration zone of the combustion stroke.

  The adder 182 adds the reference signal cycle output from the history storage unit 154 in FIG. 12 and the adjustment amount output from the adjustment amount calculation unit 180, and calculates a predicted signal cycle predicted as the next signal cycle. To do.

In FIG. 14 described above, the period calculation unit 170 corresponds to the period calculation means of the present invention.
(Circuit configuration of adjustment amount calculation unit)
FIG. 15A shows an adjustment amount calculation unit 190 which is an example of a circuit configuration of the adjustment amount calculation unit 180 shown in FIG. The adjustment amount calculation unit 190 realizes the process of calculating the adjustment amount H by multiplying the change amount α and the angle characteristic coefficient by two shifters 194a and 194b and one adder / subtractor 196. In this case, the angle characteristic coefficient is set to a value calculated by a power of 2 and its addition / subtraction.

The adjustment control unit 192 switches the shift amount of the shifters 194a and 194b according to the value of the angle characteristic coefficient, and switches the addition or subtraction of the adder / subtractor 196.
FIG. 15B shows an example of the relationship between the angle characteristic coefficient, the multiplication value multiplied by the change amount by the shifters 194a and 194b, and the addition or subtraction of the adder / subtractor 196.

  The multiplication value of the shifters 194a and 194b is a power of 2 that can be shifted. For example, when the angle characteristic coefficient is 1.75, the shifter 194a performs a left 1-bit shift operation, the shifter 194b performs a 2-bit shift operation, and the adder / subtractor 196 subtracts the output of the shifter 194b from the output of the shifter 194a.

  Thus, the angle characteristic coefficient can be set in the range of 0 to 2.0 by switching the shift amount of the shifters 194a and 194b and the addition or subtraction of the adder / subtractor 196 according to the angle characteristic coefficient. In addition, by further increasing the shift amount of the shifters 194a and 194b, the setting range of the angle characteristic coefficient can be increased and the multiplication value can be reduced.

  In FIG. 15A described above, the shifters 194a and 194b and the adder / subtractor 196 change the angle characteristic into the amount of change when the angle characteristic coefficient is set to a value calculated by a power of 2 and its addition / subtraction. A circuit for multiplying coefficients.

(Circuit configuration of the cycle adjustment unit)
FIG. 16 shows a cycle adjustment unit 200 which is another example of the circuit configuration of the cycle adjustment unit. The period adjustment unit 200 is configured to select the output of the period calculation unit 158 or the reference signal period by the selector 202 according to a software command, using the latest signal period as the reference signal period.

  According to this configuration, when an abnormality occurs in the cycle adjustment unit 200 other than the selector 202, the signal cycle prediction process by the cycle adjustment unit 200 can be prohibited, and the reference signal cycle before correction can be output as the predicted signal cycle. As a result, it is possible to prevent an abnormal predicted signal period from being set as the next signal period when the period adjusting unit 200 is abnormal, so that reliability is improved.

  Further, for example, when the engine rotation speed is high, the reference signal cycle may be selected by the selector 202 according to a software command, and the prediction process for the reference signal cycle may be prohibited. Thereby, it is possible to prevent deterioration in the accuracy of prediction of the signal period due to the prediction process at the time of high rotation in which an unnecessary noise component is emphasized by the prediction process and an error increases.

  Further, in the vicinity of the TDC and in an angle section where the change in the signal period is small, such as an intermediate portion between the TDC and the TDC, a control signal is output from the angle determination unit 152 to the selector 202 according to the crank angle, and the reference signal The period may be selected and the prediction process for the reference signal period may be prohibited. As a result, it is possible to prevent deterioration of the prediction accuracy of the signal period due to the prediction process in an angle section in which the change of the signal period is small and is susceptible to noise by the prediction process.

  In FIG. 16 described above, the period adjusting unit 200 corresponds to the period predicting unit of the present invention, and the selector 202 selects either the reference signal period or the predicted signal period predicted by adjusting the reference signal period. This corresponds to the selector of the present invention.

(Periodic learning section)
Next, the period learning unit 300 shown in FIG. 2 will be described. When the period predicting unit 110 predicts the signal period and the period learning unit 300 needs to correct the adjustment amount with respect to the reference signal period, the period learning unit 300 determines the correction amount and sends the correction determination unit 320 to the period adjusting unit 130. Output. The period learning unit 130 is configured by an integrated circuit.

  FIG. 17 shows a time chart of the learning correction process executed by the periodic learning unit 300 in FIG. In FIG. 17, the crank angle 1 (CA1) corresponding to the crank angle signal 1 indicates the learning target angle in FIG. 2, and the crank angle 2 (CA2) corresponding to the crank angle signal 2 is (learning target angle in FIG. + 10 ° CA).

In the time chart of FIG. 17, every time a crank angle signal is generated, the cycle measuring unit 120 measures the signal cycle of the previous and current crank angle signals.
Then, when the signal period is measured, as described above, the period adjustment processing is executed in the period adjustment unit 130 based on the signal period measured so far including the signal period measured this time. The signal period of the crank angle signal is predicted. The cycle prediction process is executed in a period indicated by hatching in FIG. 17 in synchronization with the rise of the clock φ immediately after the crank angle signal is generated. The prediction signal period predicted by the period adjustment unit 130 is stored in the prediction signal period storage unit 142 of the multiplication angle clock generation unit 140.

  The period learning unit 300 inputs the prediction signal period output from the multiplication angle clock generation unit 140 to the selector 302 and stores the prediction signal period in the register 304 as the previous prediction signal period.

  The selector 302 is a multiplication angle clock generation unit when the pulse count number of the crank angle signal output from the period measurement unit 120 is the crank angle 1 (learning target angle) and the selection signal output from the comparator 310 is 1. The prediction signal cycle output from 140 is selected, and when the count number of the crank angle signal is crank angle 2 (learning target angle + 10 ° CA) and the selection signal output from the comparator 310 is 0, it is stored in the register 304. Select the previous predicted signal period.

  The subtractor 306 is an actual signal measured by the period measurement unit 120, either the prediction signal period output from the multiplication angle clock generation unit 140 selected by the selector 302 or the prediction signal period stored in the register 304. Subtract from the period.

  The subtractor 306 has an abnormal signal from the outside of 0, an output signal of the learning control unit 132 of the cycle prediction unit 110 is 1, and the count number of the crank angle signal is the learning target angle or (learning target angle + 10 When all the subtraction conditions (° CA) are satisfied and 1 is output from the logic gate 314, the subtraction result is output. When any of the subtraction conditions is not satisfied, the subtraction result is not output.

  The subtraction condition in which the pulse count number of the crank angle signal is the learning target angle or (learning target angle + 10 ° CA) is when either of the outputs of the comparators 310 and 312 is 1 and the output of the logic gate 316 is 1. To establish.

  In accordance with the above conditions, when the crank angle signal generated this time is the crank angle signal 1 (CA signal 1) which is the learning target angle, the selector 302 outputs the current prediction signal cycle, and the subtractor 306 outputs the current prediction signal. The actual signal period measured this time is subtracted from the signal period.

  Here, T (i) is an actual signal cycle measured when the crank angle signal 1 is generated, and prediction between the crank angle signal 1 and the crank angle signal 2 predicted when the crank angle signal 1 is generated. Assuming that the signal period is T ′ (i + 1), the output of the subtractor 306 when the crank angle signal 1 is generated is T ′ (i + 1) −T (i). This represents an adjustment amount from the reference signal cycle T (i) when predicting the predicted signal cycle T ′ (i + 1) from the actual signal cycle T (i) which is the reference signal cycle.

  Next, when the crank angle signal generated this time is the crank angle signal 2 (CA2) which is (learning target angle + 10 ° CA), the selector 302 outputs the previous predicted signal cycle stored in the register 304 and subtracts it. The unit 306 subtracts the previous predicted signal period from the actual signal period measured this time.

  The output of the subtracter 306 at this time is T (i + 1) −T ′ (i + 1), where T (i + 1) is the actual signal period measured when the crank angle signal 2 is generated. This represents a cycle error obtained by subtracting the predicted signal cycle T ′ (i + 1) predicted when the crank angle signal 1 is generated from the actual signal cycle T (i + 1) measured when the crank angle signal 2 is generated. ing.

If the period error output from the subtracter 306 is within a predetermined error range, the correction determination unit 320 determines that the predicted signal period predicted by the period prediction unit 110 is appropriate.
On the other hand, when the period error output from the subtractor 306 is out of a predetermined error range, the correction determination unit 320 determines that the adjustment amount for the reference signal period determined by the period adjustment unit 130 should be corrected. The correction amount is determined and output to the period adjustment unit 130. In this embodiment, the adjustment amount is corrected by correcting the angle characteristic coefficient.

In the time chart of FIG. 17, the adjustment amount is obtained at the timing when the crank angle signal 1 is generated, and the cycle error is obtained at the timing when the crank angle signal 2 is generated.
On the other hand, as shown in FIG. 18, the adjustment amount and the cyclic error may be obtained at the same time when the crank angle signal 2 is generated.

The periodic learning unit 300 in FIG. 2 described above functions as learning means of the present invention, and the logic gate 314 corresponds to learning prohibiting means.
Next, flowcharts of the signal period learning correction processing 1 and the correction determination processing and correction processing executed in the learning correction processing 1 according to the present embodiment are shown in FIGS. The flowcharts shown in FIGS. 19 to 21 are executed by the period learning unit 300.

  In S520 of FIG. 19, the ECU 10 executes a correction determination process. In the correction determination process in S520, at least a majority of preset maximum learning times is executed. For example, if the maximum learning count is set to 10, the correction determination process in S520 is executed at least 6 times.

(Correction judgment processing)
Hereinafter, details of the correction determination process executed in S520 will be described based on the flowchart of FIG.

  In S540 of FIG. 20, the ECU 10 obtains a coefficient determination threshold for determining whether or not to correct the angle characteristic coefficient from the following equation (11). The period learning unit 300 acquires the adjustment amount of Expression (11) and the current angle characteristic coefficient from the period prediction unit 110.

Coefficient determination threshold = | Adjustment amount / Current angle characteristic coefficient | × Coefficient resolution (11)
In the equation (11), the adjustment amount is positive in the angle interval where the engine rotation speed decreases, becomes negative in the angle interval where the engine rotation speed increases, and the value of the angle characteristic coefficient is positive regardless of the angle interval. . The term | adjustment amount / current angle characteristic coefficient | represents the absolute value of the change before multiplication by the angle characteristic coefficient. As shown in FIG. 22A, when the angle characteristic coefficient is increased or decreased every 0.25, the coefficient resolution is 0.25.

  Therefore, the coefficient determination threshold represents the absolute value of the increase / decrease amount when the angle characteristic coefficient is increased / decreased in units of resolution. For example, in FIG. 22, when the adjustment amount is 10 or −10, the coefficient determination threshold value is 2.5.

  In S542, the ECU 10 determines whether or not the absolute value of the cyclic error is larger than the coefficient determination threshold obtained by the equation (11). This determination is based on the actual signal period actually measured and the predicted signal period obtained when the adjustment amount is 10 or -10 when the current angular characteristic coefficient is 1.0, as shown in FIG. It is determined whether or not the absolute value of the cyclic error that is the difference between the two is greater than 2.5.

  If the absolute value of the cyclic error is larger than the coefficient determination threshold (S542: Yes), if the adjustment amount is positive and the cyclic error is also positive, the predicted signal period is too small than the actual signal period, so the angle characteristic coefficient is increased. If the adjustment amount is positive and the period error is negative, the predicted signal period is too large compared to the actual signal period, so the angle characteristic coefficient is reduced and added to the reference signal period. It is necessary to reduce the adjustment amount.

  If the adjustment amount is negative and the cycle error is negative, the predicted signal cycle is too large compared to the actual signal cycle, so the angle characteristic coefficient is increased and the absolute value of the adjustment amount subtracted from the reference signal cycle is increased. When the adjustment amount is negative and the cycle error is positive, the predicted signal cycle is too small compared to the actual signal cycle, so it is necessary to reduce the angle characteristic coefficient and reduce the absolute value of the adjustment amount to be subtracted from the reference signal cycle. .

  Therefore, when the absolute value of the cyclic error is larger than the coefficient determination threshold (S542: Yes) and the sign of the cyclic error and the adjustment amount matches (S544: Yes), the ECU 10 adds 1 to the learning value ( In S546), if the sign of the cyclic error and the adjustment amount does not match (S544: No), the ECU 10 subtracts 1 from the learning value (S548).

If the learning value is positive, it becomes a threshold value for determining whether to increase the angle characteristic coefficient, and if it is negative, it becomes a threshold value for determining whether to decrease the angle characteristic coefficient.
When the absolute value of the cycle error is equal to or smaller than the coefficient determination threshold (S542: No), it can be determined that the predicted signal cycle is appropriately set. In this case, in the present embodiment, the learning value set in S546 and S548 is set to be close to 0, and the angle characteristic coefficient is set to be less likely to increase or decrease.

  Therefore, when the absolute value of the cyclic error is equal to or smaller than the coefficient determination threshold value (S542: No), if the learning value is negative (S550: Yes), the ECU 10 adds 1 to the learning value and brings the learning value close to 0 ( If the learning value is positive (S552), the ECU 10 subtracts 1 from the learning value to bring the learning value close to 0 (S548).

When the learning value is 0 (S550: No, S552: No), the ECU 10 ends this process.
The correction determination process of FIG. 20 described above corresponds to the function executed by the learning means of the present invention. The learning value corresponds to the correction determination value of the present invention. The coefficient determination threshold corresponds to a predetermined error range of the present invention.

  In the correction determination process of FIG. 20, not only the learning value is increased or decreased when the absolute value of the periodic error is larger than the coefficient determination threshold value, but the learning value is brought close to 0 when the absolute value of the cyclic error is equal to or smaller than the coefficient determination threshold value. Was executed. On the other hand, when the absolute value of the cyclic error is larger than the coefficient determination threshold, only the process of increasing or decreasing the learning value may be executed.

(Learning correction process 1)
When the correction determination process of FIG. 20 is executed in S520 of FIG. 19, in S522 of FIG. 19, the ECU 10 determines whether or not the absolute value of the learning value having a positive or negative value has reached a majority of the maximum number of learning times. judge. The maximum learning number is set to an arbitrary number by the software.

  When the absolute value of the learning value reaches a majority of the maximum number of learning times (S522: Yes), the ECU 10 does not return to the process of S520, but proceeds to S528 to execute the correction process. This is because, in the present embodiment, when the absolute value of the learning value reaches a majority of the maximum number of learning times, it is determined whether or not to increase or decrease the angle characteristic coefficient. Thereby, the frequency | count of performing a correction | amendment determination process can be reduced, and processing time can be reduced.

  If the absolute value of the learning value does not reach the majority of the maximum number of learning times (S522: No), the ECU 10 adds 1 to the learning number (S524), and determines whether the learning number is equal to or greater than the maximum learning number. (S526). The initial value of the number of learning is 0. When the number of learning is less than the maximum number of learning (S526: No), the ECU 10 returns the process to S520.

When the number of learning is equal to or greater than the maximum number of learning (S526: Yes), the ECU 10 shifts the process to S528 and executes the correction process.
In S528, the ECU 10 executes correction for increasing or decreasing the angle characteristic coefficient by the correction process shown in FIG. The correction process in FIG. 21 will be described later. When S528 is executed, the ECU 10 updates the current learning target angle to set the next learning target angle (S530), sets the number of learnings to 0, and ends this process (S532). As described above, when the learning correction processing is completed at one learning target angle, the learning target angle is sequentially updated, so that the cycle error can be reduced in the entire signal cycle change characteristic.

In S530, the learning target angle may be sequentially set by increasing the crank angle every 10 ° CA, or only a predetermined angle may be selected and set.
The learning correction process 1 of FIG. 19 described above corresponds to the function executed by the learning means of the present invention.

  In the learning correction process 1 of FIG. 19, when a crank angle signal corresponding to one crank angle that is a learning target angle is generated, a correction determination process is executed, and whether the absolute value of the learning value reaches a majority of the maximum number of learning times, When the number of learning times reached the maximum number of learning times, the correction processing of S528 was executed to update the learning target angle.

  On the other hand, the learning correction process including the correction determination process and the determination process may be executed in parallel every time the corresponding crank angle signal is generated at all the crank angles that are the learning target angles.

(Correction process)
Details of the correction processing executed in S528 of FIG. 19 will be described based on the flowchart of FIG.

  In S560 of FIG. 21, the ECU 10 determines whether or not the learning value set by the correction determination process of FIG. 20 is larger than the increase determination threshold when increasing the angle characteristic coefficient. If the maximum number of learning set in the learning correction process 1 of FIG. 19 is, for example, 10 times, the increase determination threshold is executed when the absolute value of the learning value reaches a majority of the maximum number of learning times. For example, it is set to 5 times.

  When the learning value is larger than the increase determination threshold value (S560: Yes), the ECU 10 determines whether or not the current angle characteristic coefficient is the maximum value in the correction range (S562). When the angle characteristic coefficient is not the maximum value of the correction range (S562: No), the ECU 10 increases the angle characteristic coefficient in units of resolution (S564). In the example of FIG. 22, the angle characteristic coefficient is increased by 0.25. After execution of S564, the ECU 10 shifts the process to S576.

  When the angle characteristic coefficient is already the maximum value of the correction range (S562: Yes), the ECU 10 cannot increase the angle characteristic coefficient any more, and from the prediction signal cycle range that can be predicted by correcting the angle characteristic coefficient. In the correction range, it is determined that the period error between the actual signal period and the predicted signal period is not within the predetermined error range, and an abnormality signal is output to notify the abnormality (S566). Possible causes of the abnormality include erroneous measurement of the signal period by the period measurement unit 120. By notifying the abnormality, appropriate measures can be taken.

  When the learning value is equal to or smaller than the increase determination threshold value (S560: No), the ECU 10 determines whether the learning value is smaller than the decrease determination threshold value (S568). If the maximum learning frequency set in the learning correction processing 1 in FIG. 19 is 10 times, the decrease determination threshold value is executed when the absolute value of the learning value reaches a majority of the maximum learning frequency. For example, it is set to -5 times.

  When the learning value is smaller than the decrease determination threshold value (S568: Yes), the ECU 10 determines whether or not the current angle characteristic coefficient is the minimum value of the correction range (S570). When the angle characteristic coefficient is not the minimum value of the correction range (S570: No), the ECU 10 decreases the angle characteristic coefficient in units of resolution (S572). In the example of FIG. 22, the angle characteristic coefficient is decreased by 0.25. After execution of S572, the ECU 10 shifts the process to S576.

  When the angle characteristic coefficient is already the minimum value of the correction range (S570: Yes), the ECU 10 cannot further decrease the angle characteristic coefficient, and the actual signal period and the prediction signal are not corrected in the correction range for correcting the angle characteristic coefficient. It is determined that there is an abnormality in which the period error with the period cannot be within a predetermined error range, and an abnormality signal is output to notify the abnormality (S574).

    When the learning value is equal to or less than the increase determination threshold value (S560: No) and is equal to or greater than the decrease determination threshold value (S568: No), the ECU 10 determines that the prediction signal cycle is appropriately set, and proceeds to S576.

In S576, the ECU 10 sets the learning value to 0 and ends this process.
In the correction processing of FIG. 21, the increase determination threshold value and the decrease determination threshold value are arbitrarily set according to, for example, the model of the engine. As a result, the adjustment amount can be corrected by appropriately correcting the angle characteristic coefficient for each engine model.

  In the correction process of FIG. 21, one increase determination threshold and one decrease determination threshold are set. In contrast, at least one of the increase determination threshold value and the decrease determination threshold value may be set to a plurality of different values. An increase / decrease amount of the angle characteristic coefficient is set for each of a plurality of values set for at least one of the increase determination threshold and the decrease determination threshold. For example, in the example of FIG. 22, the angle characteristic coefficient may be increased or decreased by ± 0.25 units and ± 0.5 units according to a plurality of set values. As a result, the angle characteristic coefficient can be corrected with high accuracy according to the learning value, and the adjustment amount can be corrected. As a result, the signal period can be predicted with high accuracy.

  In the correction processing of FIG. 21 described above, the processing of S560 to S564 and S568 to S572 corresponds to the function executed by the learning means of the present invention, and the processing of S566 and S574 corresponds to the function of the notification means of the present invention. To do.

The increase determination threshold corresponds to the predetermined upper limit of the present invention, and the decrease determination threshold corresponds to the predetermined lower limit of the present invention.
Further, by executing the learning correction process, the correction determination process, and the correction process of FIGS. 19 to 21, even if the change characteristic of the signal period changes due to engine difference or aging, the actual signal period and the predicted signal period The angle characteristic coefficient is corrected with high accuracy by learning the cyclic error, and the adjustment amount with respect to the reference signal period can be corrected. As a result, it is possible to predict the signal period after the reference signal period with high accuracy by adjusting the reference signal period.

(Learning correction process 2)
FIG. 23 shows a flowchart of another learning correction process 2. In the learning correction process 1 of FIG. 19, the learning correction process for the angle characteristic coefficient is performed for each crank angle. On the other hand, in the learning correction process 2 in FIG. 23, the learning correction process is summarized for a plurality of crank angles in the same angle section in an angle section in which the adjustment amount is set using the same angle characteristic coefficient in a plurality of crank angles. And correct the common angle characteristic coefficient.

  First, in S580, the ECU 10 executes a correction determination process. This correction determination process is substantially the same as the correction determination process executed in FIG. Next, the ECU 10 adds 1 to the number of learning (S582), and if the number of learning is less than the maximum number of learning (S584: No), the process returns to S580.

  When the number of learning is equal to or greater than the maximum number of learning (S584: Yes), the ECU 10 adds the learning value set in the correction determination process to the learning value total (S586). The initial value of the learning value sum is zero. Next, the ECU 10 updates the current learning target angle to set the next learning target angle (S588), and sets the learning value and the number of learnings to 0 (S590, S592). In the learning correction process 2 of FIG. 23, the learning target angle is sequentially set by increasing the crank angle every 10 ° CA.

  In S594, the ECU 10 adds 1 to the learning angle number, and when the learning angle number is less than the maximum learning angle number (S596: No), the process returns to S580. The initial value of the learning angle number is zero. The maximum learning angle number is the number of crank angles as the learning target angle using a common angle characteristic coefficient in one angle section.

  When the number of learning angles becomes equal to or greater than the maximum learning angle (S596: Yes), the ECU 10 sets the total learning value as the learning value (S598), and executes correction processing (S600). Since this correction process is substantially the same as the correction process executed in FIG. However, since the learning value learned for one crank angle is the learning value learned for a plurality of crank angles, the correction processing in FIG. 21 executed in S600 performs S528 in FIG. The increase determination threshold is set to be large and the decrease determination threshold is set to be smaller than the correction processing of FIG.

When the correction process is executed in S600, the ECU 10 sets the total learning value and the learning angle number to 0 (S602, S604), and ends this process.
In the learning correction process of FIG. 23, when the learning correction process is completed in one angle section, the ECU 10 sequentially executes the learning correction process in the next angle section.

The learning correction process 2 in FIG. 23 described above corresponds to the function executed by the learning means of the present invention.
In the learning correction process 2 of FIG. 23, when a crank angle signal corresponding to one crank angle that is a learning target angle is generated in the same angle section, a correction determination process is executed, and the absolute value of the learning value is the maximum number of learning times. When the majority is reached or the learning count reaches the maximum learning count, the learning target angle is updated within the angle interval.

  On the other hand, correction determination processing may be executed in parallel every time a corresponding crank angle signal is generated at all crank angles that are learning target angles in the same angle section. In this case, when the correction determination process is completed for all the crank angles in the angle section, the correction process is executed.

(Other periodic learning section)
In the periodic learning unit 300 illustrated in FIG. 2, the learning determination process is executed by hardware configured with an integrated circuit. On the other hand, in the period learning unit 330 of FIG. 24, calculation of the period error and the adjustment amount is executed by software in the calculation unit 340, and calculation of the angle characteristic coefficient correction amount and correction abnormality determination are performed by the correction determination unit 350 in software. Run.

The period learning unit 330 in FIG. 24 functions as learning means of the present invention.
[Other embodiments]
In the above embodiment, the amount of change, which is the difference between the signal cycles measured up to this time, is calculated, and the amount of adjustment is adjusted to the currently measured signal cycle by adjusting the amount of change with the angle characteristic coefficient corresponding to the crank angle. Is added to predict the next signal period.

  On the other hand, the signal period measured this time may be directly corrected according to the crank angle. For example, the signal characteristic measured this time may be directly multiplied by an angle characteristic coefficient. Moreover, you may estimate the signal period after several times instead of the next signal period.

Further, the difference between the signal periods is not limited to one set of differences based on the two signal periods measured up to this time, and a plurality of sets of differences may be weighted according to the crank angle.
Further, in the above embodiment, based on the signal period measured up to this time, the reference signal period to be adjusted is adjusted according to the crank angle among the signal periods measured up to this time, and the signal period after this time is adjusted. Predicted.

  On the other hand, based on the signal period measured before this time, the reference signal period to be adjusted among the signal periods measured before this time is adjusted according to the crank angle. You may predict the past signal period later and before this time.

  In the above embodiment, the signal period is predicted by determining the adjustment amount for adjusting the reference signal period based on the crank angle and the angle characteristic coefficient set according to at least the crank angle of the engine speed.

  On the other hand, regardless of the crank angle, the signal period may be predicted by determining the adjustment amount with respect to the reference signal period based on the ratio before and after the measured signal period or the ratio of the change amount of the signal period. In this case, it is conceivable that the adjustment amount is corrected by adjusting the ratio based on the period error between the actual signal period and the predicted signal period.

  In contrast to the above-described embodiment, the learning correction processing is executed only in a predetermined angle section among the angle sections in which the engine rotation speed is reduced in the compression stroke of at least one cylinder of the engine, or in the compression stroke of at least one cylinder of the engine You may increase the learning frequency of a predetermined angle area among the angle areas where an engine speed decelerates from another angle area.

  Further, the learning correction processing is executed only in a predetermined angle section among the angular sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine, or the engine speed is increased in the combustion stroke of at least one cylinder of the engine. You may increase the learning frequency of a predetermined angle area among the angle areas to accelerate from another angle area.

  In these angle sections, the amount of change in the engine rotation speed is large, so the amount of change in the signal period is also large. Therefore, the error of the prediction signal period can be reduced by executing learning correction processing in these angle sections or increasing the learning frequency as compared with other angle sections.

The crank angle for executing the learning correction process may be set for each cylinder of the engine. Thereby, an important crank angle to be learned can be set for each cylinder.
Further, the crank angle for executing the learning correction process may be set at the same position between the cylinders of the engine. Thereby, in each cylinder, the period error between the actual signal period and the predicted signal period can be reduced as much as possible at the same crank angle. As a result, variations in engine control between cylinders can be reduced.

  In addition, the electric motor also has a change characteristic according to the rotation angle in the rotation speed at which the rotor rotates. Therefore, the signal cycle of the rotation angle signal, which is a signal sequence generated at a predetermined rotation angle with the rotation of the rotation shaft of the electric motor, also has a change characteristic according to the rotation angle.

  Therefore, also in the electric motor, based on the signal period measured up to this time, the reference signal period to be corrected among the signal periods measured up to this time according to the rotation angle of the rotating shaft represented by the rotation angle signal. May be corrected to correct the signal period after this time.

As a result, a rotational angle position that is more detailed than the angular interval of the rotational angle signal can be obtained within the predicted signal cycle range.
In the said embodiment, although applied to the engine control apparatus of the diesel engine for vehicles, application of this invention is not limited to this, You may apply to the engine control apparatus of a gasoline engine.

  As described above, the present invention is not limited to the above-described embodiment, and can be applied to various embodiments without departing from the gist thereof.

10: ECU (engine control device, measurement unit, cycle prediction unit, learning unit, angular position prediction unit, prediction prohibition unit, learning prohibition unit, notification unit), 120: cycle measurement unit (measurement unit), 130, 150, 200 : Period adjustment unit (cycle prediction unit), 140: multiplication angle clock generation unit (angle position prediction unit), 152: angle determination unit (angle detection unit, control unit), 154, 160: history storage unit (history storage unit) 156: change amount calculation unit (change amount calculation unit), 158, 170: cycle calculation unit (cycle calculation unit), 162a, 162b, 162c: history register (cycle storage unit), 202: selector, 300, 330: cycle Learning unit (learning means), 314: logic gate (learning prohibition means)

Claims (46)

Measuring means for measuring a signal period of a crank angle signal, which is a signal train generated at a predetermined angular interval with rotation of the crankshaft of the engine;
A period predicting means for adjusting the reference signal period to be adjusted among the signal periods measured by the measuring means and predicting the signal period after the reference signal period;
Learning a comparison result between an actual actual signal cycle measured by the measuring unit and a predicted signal cycle which is the signal cycle predicted by the cycle predicting unit, and the cycle predicting unit predicts the predicted signal cycle Learning means for correcting the adjustment amount with respect to the reference signal period based on the learning result;
Angular position prediction means for obtaining a crank angle position more detailed than the predetermined angular interval in the range of the predicted signal period after correction by the learning means;
An engine control device comprising: The period predicting means determines the adjustment amount by an adjustment coefficient,
The learning means corrects the adjustment amount by correcting the adjustment coefficient;
The engine control apparatus according to claim 1.   The learning means learns a period error that is a difference between the actual signal period and the predicted signal period as the comparison result, and corrects the adjustment coefficient so that the period error is within a predetermined error range. The engine control device according to claim 2, wherein   The engine control apparatus according to claim 3, wherein the learning unit learns the period error of the prediction signal period to be learned a plurality of times and corrects the adjustment coefficient.   The learning means learns the periodic error a plurality of times, and at least one of the number of times the periodic error is outside the predetermined error range and the number of times the periodic error is within the predetermined error range. The engine control apparatus according to claim 3 or 4, wherein a correction determination value for determining whether or not to correct the adjustment coefficient based on the number of times is set.   The learning means includes at least the number of times that the cyclic error is out of the predetermined error range among the number of times that the cyclic error is out of the predetermined error range and the number of times that the cyclic error is in the predetermined error range. When the correction determination value reaches a majority of a plurality of learning times set in advance to learn the periodic error during learning of the periodic error, the correction determination value reaches the majority of the periodic error. The engine control apparatus according to claim 5, wherein learning is stopped and the process proceeds to a correction process for the adjustment coefficient.   The learning means includes at least the number of times that the cyclic error is out of the predetermined error range among the number of times that the cyclic error is out of the predetermined error range and the number of times that the cyclic error is in the predetermined error range. The correction determination value is set based on the value, the adjustment coefficient is increased when the correction determination value is larger than a predetermined upper limit value, and the adjustment coefficient is decreased when the correction determination value is smaller than a predetermined lower limit value. The engine control device according to claim 5 or 6, wherein   The engine control apparatus according to claim 7, wherein the predetermined upper limit value and the predetermined lower limit value are arbitrarily set in advance.   A plurality of at least one of the predetermined upper limit value and the predetermined lower limit value are set for the correction determination value, and a correction amount for correcting the adjustment coefficient is set for each set value. The engine control device according to claim 7 or 8.   The notification means for notifying an abnormality when the period error is outside the predetermined error range within a range in which the learning means can correct the adjustment amount and predict the prediction signal period. The engine control device according to any one of 3 to 9. The common adjustment coefficient is set in an angle section composed of a plurality of the crank angles,
The learning means learns the comparison result at each crank angle in the angle interval, and corrects the adjustment coefficient common to the angle interval based on the comparison result.
The engine control device according to any one of claims 2 to 10, wherein   The engine control device according to any one of claims 2 to 11, wherein the adjustment coefficient is set according to at least the crank angle of the crank angle and the engine speed.   The engine control apparatus according to any one of claims 1 to 12, further comprising a learning prohibiting unit that prohibits processing by the learning unit when a prohibition signal is input from the outside.   The engine control device according to any one of claims 1 to 13, wherein the learning unit performs the comparison result learning process and the adjustment amount correction process at a plurality of the crank angles.   The engine control device according to any one of claims 1 to 14, wherein the learning unit sequentially executes the learning process and the correction process for each crank angle separated by the predetermined angle interval.   The engine control apparatus according to any one of claims 1 to 15, wherein the crank angle at which the learning unit learns the comparison result is set for each cylinder of the engine.   The engine control apparatus according to any one of claims 1 to 15, wherein the crank angle at which the learning unit learns the comparison result is set to the same position between cylinders of the engine.   18. The learning device according to claim 1, wherein the learning unit learns the comparison result only in a predetermined angle section among the angle sections in which the engine rotation speed is reduced in the compression stroke of at least one cylinder of the engine. The engine control device according to claim 1.   2. The learning device according to claim 1, wherein a learning frequency of a predetermined angle section among the angle sections in which the engine rotation speed is reduced in a compression stroke of at least one cylinder of the engine is increased as compared with other angle sections. The engine control device according to any one of 1 to 18.   20. The learning device according to claim 1, wherein the learning unit learns the comparison result only in a predetermined angle section among the angle sections in which the engine rotation speed is accelerated in a combustion stroke of at least one cylinder of the engine. The engine control device according to claim 1.   2. The learning device according to claim 1, wherein a learning frequency of a predetermined angle section among the angle sections in which the engine speed is accelerated in a combustion stroke of at least one cylinder of the engine is increased as compared with other angle sections. 21. The engine control device according to any one of 1 to 20.   The engine control apparatus according to any one of claims 1 to 21, wherein the processing by the learning unit is executed by a circuit.   The engine control apparatus according to claim 22, wherein the circuit is an integrated circuit.   The engine control apparatus according to any one of claims 1 to 21, wherein the process by the learning unit is executed by software.   The cycle predicting means, when predicting the signal cycle in a predetermined angle section among the angular sections in which the engine speed is reduced in the compression stroke of at least one cylinder of the engine, is more preferable than before and after the predetermined angle section. The engine control device according to any one of claims 1 to 24, wherein an adjustment amount is increased.   A prediction prohibiting unit that prohibits the prediction by the cycle prediction unit when the prediction signal cycle is smaller than the reference signal cycle in an angular interval in which the engine speed is reduced in the compression stroke of at least one cylinder of the engine; The engine control device according to any one of claims 1 to 25, wherein:   The cycle prediction means, when predicting the signal cycle of a predetermined angular section before and after the top dead center of at least one cylinder of the engine, makes the absolute value of the adjustment amount more immediately before and after the predetermined angular section. The engine control device according to any one of claims 1 to 26, wherein the value is reduced.   The cycle predicting means, when predicting the signal cycle in a predetermined angle section among the angle sections in which the engine rotation speed is accelerated in the combustion stroke of at least one cylinder of the engine, sets the absolute value of the adjustment amount as compared with the previous time. The engine control device according to any one of claims 1 to 27, wherein the engine control device is made smaller.   When the prediction signal period is larger than the reference signal period in a predetermined angle section among the angular sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine, the prediction by the period prediction unit is prohibited. The engine control apparatus according to any one of claims 1 to 28, further comprising a prediction prohibiting unit.   The cycle prediction means converts a deceleration characteristic of the engine rotational speed in a predetermined angle section among the angular sections in which the engine rotational speed decelerates in the compression stroke of at least one cylinder of the engine, and the engine in the combustion stroke of the corresponding cylinder 30. The engine control device according to any one of claims 1 to 29, wherein the adjustment amount is calculated in a predetermined angle section among the angle sections in which the rotation speed is accelerated.   The cycle prediction means converts the acceleration characteristics of the engine in a predetermined angle section among the angular sections in which the engine speed is accelerated in the combustion stroke of at least one cylinder of the engine, and the engine in the compression stroke of other cylinders The engine control device according to any one of claims 1 to 30, wherein the adjustment amount is calculated in a predetermined angle section among the angle sections in which the rotation speed is decelerated.   The cycle predicting means converts one characteristic of the deceleration characteristic or the acceleration characteristic and predicts the signal period in the other characteristic, and calculates the absolute value of the adjustment amount as one of the deceleration characteristic or the acceleration characteristic. 32. The engine control device according to claim 30 or 31, wherein the engine control device is set to be larger than an absolute value of a change amount of the signal period in the characteristic.   The period predicting means reduces the adjustment amount in an angular section where the absolute value of the rotational acceleration of the engine is equal to or less than a predetermined value than in an angular section where the absolute value of the rotational acceleration is larger than the predetermined value. The engine control device according to any one of claims 1 to 32.   When the engine speed becomes equal to or higher than the first speed, the cycle predicting means makes the adjustment amount smaller than when the engine speed is less than the first speed, and the engine speed is lower than the first speed. The engine control device according to any one of claims 1 to 33, wherein when the speed is less than the speed, the adjustment amount is made larger than when the engine rotation speed is equal to or higher than the second speed.   The engine control device according to any one of claims 1 to 34, wherein the cycle prediction unit obtains the prediction signal cycle by adding the reference signal cycle and the adjustment amount.   The period predicting unit multiplies an adjustment coefficient set according to at least the crank angle out of the crank angle and the engine speed by a change amount determined by a time change of the signal period measured by the measuring unit. 36. The engine control device according to claim 1, wherein the adjustment amount is obtained.   The engine control apparatus according to claim 36, wherein the adjustment coefficient is a value determined by addition / subtraction of a power of 2.   38. The engine control device according to claim 36 or 37, wherein the period predicting unit obtains the amount of change based on a difference between two signal periods measured by the measuring unit.   39. The engine control apparatus according to claim 38, wherein the period predicting means uses the difference between two consecutive signal periods as the amount of change.   The engine control device according to any one of claims 1 to 39, wherein the period predicting unit sets the adjustment amount to 0 in a predetermined angle section.   41. The cycle prediction unit adjusts the reference signal cycle among the signal cycles measured by the measurement unit up to this time, and predicts the signal cycle after the current time. The engine control device according to any one of the above.   The engine control apparatus according to any one of claims 1 to 41, wherein the cycle predicting unit sets the signal cycle measured this time by the measuring unit as the reference signal cycle.   The engine control device according to any one of claims 1 to 42, wherein the cycle predicting unit predicts the signal cycle immediately after the current time. The period predicting unit calculates a change amount of the signal period from a history storage unit that stores the history of the signal period measured by the measuring unit in a period storage unit, and the signal period stored in the period storage unit. Change amount calculating means for adjusting the reference signal period based on the change amount calculated by the change amount calculating means and calculating the signal period after the reference signal period. ,
Angle detection means for detecting the crank angle from the number of signals of the crank angle signal;
Based on the crank angle detected by the angle detection means, selection of the signal period output by the period storage unit, calculation processing in which the change amount calculation means calculates the change amount, and the cycle calculation means in the Control means for controlling adjustment processing for adjusting the reference signal period;
44. The engine control device according to any one of claims 1 to 43, comprising:   The cycle calculating means includes a coefficient storage unit that determines the adjustment amount based on an adjustment coefficient set according to at least the crank angle of the crank angle and the engine speed, and stores the adjustment coefficient. The engine control device according to claim 44.   The period predicting unit includes a selector that selects one of the reference signal period and the signal period calculated by adjusting the reference signal period by the period calculating unit, and the crank angle or the soft command 46. The engine control device according to claim 44, wherein the output of the selector is switched by the control.

Images