JPH09210192A - Shift control device automatic transmission for vehicle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To carry out leaning according to the preference of a driver with a high accuracy so as to realize sable shift control by not adopting the output of a monitor and not updating a calculated leaning data, in the case where it is judged that a car runs on a traffic snarl road. SOLUTION: Leaning reflected the preference of a driver is carried out by a leaning type selecting means 3B, and an optimal shift map is selected according to a leaned result. The traveling condition of a vehicle and the will of the driver are judged by an optimal shifting position deciding means 1, and the means 1 is constitution as a fuzzy type optimal shifting position deciding means for correcting a target shifting position set by a purpose shifting position setting means 3. A leaning data obtained when car speed V not yet exceed prescribed speed is not employed, it is judged that traffic snarl traveling is completed and a traveling condition is moved into a normal traveling condition when car speed exceeds prescribed car speed, the leaning data at the time of just before shift-up is employed as an effective data. It is thus possible to stably hold the accuracy of shift control in which shifting preference of a driver is reflected.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shift control device for an automatic transmission for a vehicle, and more particularly to a shift control device for an automatic transmission in which shift control is performed based on learning.

[0002]

2. Description of the Related Art Conventionally, as a vehicle transmission, an automatic transmission in which a shift operation is automated has been frequently used. The mainstream of this automatic transmission has a torque converter as a joint instead of the clutch. Also, in the case of large vehicles such as buses and trucks, an actuator that automatically connects and disconnects the clutch is provided in a mechanical transmission similar to the manual transmission, in order to make it easier to operate the clutch. It is considered that the pedal is removed to form an automatic transmission.

Usually, in such a shift control device for an automatic transmission, a target shift speed is set according to the vehicle speed and the accelerator opening. However, it is difficult to sufficiently reflect the driver's intention and preference according to the vehicle speed and the accelerator opening degree in the target shift speed, and therefore, the shift according to the intentions of all the drivers is necessarily realized well. Not necessarily.

Therefore, in recent years, an automatic transmission has been developed which is mainly adapted to automatic gear shifting, but which is also capable of manual gear shifting according to the driver's preference. This allows the driver to manually shift gear according to his or her preference. Is also possible.
Further, there is a shift control device configured to learn the driver's preference based on the content of the shift using a neuronetwork or the like, and automatically shift according to the learned content.
No. 60, etc.

[0005]

By the way, in the shift control device as disclosed in the above publication, not only the shift based on the preference of the driver but also the shift depending on the road condition or the like is required. The forced shift information may also be used as data for learning. For example, when the vehicle is traveling at a low speed on a congested road, most gear shifting is performed according to the road conditions regardless of the driver's preference. However, all the gear shifting information at this time is used as learning data. It may be used.

As described above, if data that is not suitable for learning is mixed during learning, the accuracy of learning is reduced, and as a result, it is difficult to maintain good shift control according to the driver's preference. Become. In particular, when shifting up at low speeds such as when driving on congested roads, the driver's preferences differ subtly for each driver, and the learning effect is expected to be great. The shift control during traveling becomes unstable, which is not preferable.

The present invention has been made based on the above-mentioned circumstances, and an object of the present invention is to perform automatic learning for a driver according to a driver's preference with high accuracy and to realize stable gear shift control. It is to provide a shift control device for a transmission.

[0008]

In order to achieve the above object, the invention of claim 1 is an accelerator opening detecting means for detecting an accelerator opening of an engine, and an engine speed detecting for detecting an engine speed. Means, monitoring means for monitoring at least one of the accelerator opening degree and the engine speed at the time of switching the shift stage of the automatic transmission, and learning data calculation for calculating learning data based on the output from the monitoring means. Means, learning means for learning based on the learning data so as to switch gears reflecting the shift preference of the driver, and the automatic transmission based on the output from the learning means and the operating state of the vehicle. Shift control means for controlling the shift control of the vehicle, and traffic jam running judgment means for judging that the vehicle is running on a traffic jam road. If it is determined that the vehicle is running the congested road by road travel determining means, it is characterized in that not updated training data without employing the output from the monitoring means.

Therefore, at least one of the accelerator opening degree and the engine speed at the time of switching the shift stage of the automatic transmission is monitored and output, and the learning data is calculated based on this output. Then, learning is performed on the basis of the learning data so as to change the shift speed in which the driver's shift preference is reflected, and the shift control of the automatic transmission is properly performed. However, when the congested road traveling determination means determines that the vehicle is traveling on the congested road, the learning data calculation means does not adopt the output from the monitor means. Therefore, while the vehicle is traveling on a congested road, the learning data is not updated unnecessarily, and the accuracy of the shift control that reflects the driver's shift preference is preferably maintained.

According to the second aspect of the invention, the vehicle is provided with a manual mode which is connected to the engine and which switches the shift speed by a driver's operation and an automatic mode which automatically shifts the shift speed in accordance with the operating condition of the vehicle. In a shift control device for an automatic transmission for a vehicle, an accelerator opening detecting means for detecting an accelerator opening of an engine, an engine speed detecting means for detecting an engine speed, and the accelerator at a time point when a shift stage of the automatic transmission is switched. Monitoring means for monitoring at least one of the opening degree and the engine speed; learning data computing means for computing learning data based on the output from the monitoring means in at least one of the manual mode and the auto mode; Based on the data, shifting of gears that reflects the driver's shift preference in the auto mode And learning means for learning in order to practice,
Target shift speed setting means for setting a target shift speed based on the output from the learning means and the operating state of the vehicle, shift control means for performing shift control of the automatic transmission based on the target shift speed, and vehicle congestion Congested road traveling determination means for determining that the vehicle is traveling on a road, the learning data calculation means, when it is determined by the congested road traveling determination means that the vehicle is traveling on the congested road The learning data is not updated without using the output from the monitor means.

Therefore, in a shift control device for a vehicle automatic transmission having a manual mode and an automatic mode,
At least one of the accelerator opening degree and the engine speed at the time of switching the shift stage of the automatic transmission is monitored, and the learning data is calculated based on the monitor output in at least one of the manual mode and the automatic mode.
Then, based on this learning data, learning is performed in order to switch the shift speed in which the shift preference of the driver in the automatic mode is reflected, whereby the target shift speed is set and the shift control of the automatic transmission is performed. Good implementation. However, when the congested road traveling determination means determines that the vehicle is traveling on the congested road, the learning data calculation means does not adopt the output from the monitor means. Therefore, while the vehicle is traveling on a congested road, the learning data is not updated unnecessarily, and the accuracy of the shift control that reflects the driver's shift preference is preferably maintained.

Further, in the invention of claim 3, the learning data calculating means runs on the congested road when the congested road traveling judging means judges that the vehicle has left the traveling on the congested road. It is characterized in that the learning data is calculated based on the latest output monitored by the monitoring means during the operation. Therefore, after the vehicle has come out of the congested road, the monitor output at the time of the gear shift, which was last executed during the congested road, is effectively applied to the calculation of the learning data, and the learning continues to be performed satisfactorily.

Further, in the invention of claim 4, the vehicle speed detection means for detecting the vehicle speed is further provided, and the congested road traveling determination means is that the vehicle is traveling on the congested road when the vehicle speed is less than a predetermined vehicle speed. It is characterized by making a judgment. Therefore, the fact that the vehicle is traveling on the congested road is easily determined by whether or not the vehicle speed is lower than the predetermined vehicle speed.

Further, in the invention of claim 5, the learning data calculation means, when the vehicle speed is equal to or higher than the predetermined vehicle speed,
When the vehicle speed is less than the predetermined vehicle speed, the learning data is calculated based on the latest output monitored by the monitor means. Therefore, it is determined that the vehicle has exited from running on a congested road because the vehicle speed has exceeded the predetermined vehicle speed. The monitor output at that time is effectively applied to the calculation of the learning data, so that the learning is favorably continued.

Further, the invention of claim 6 is characterized in that the shift stage switching time is a shift up time. Therefore, at least one of the accelerator opening degree and the engine speed at the time of upshifting of the automatic transmission is monitored and output, and the learning data is calculated based on this output. Then, learning is carried out based on the learning data so as to carry out the shift-up in which the driver's shift preference is reflected, and the shift-up control of the automatic transmission is properly carried out. However, when the vehicle is traveling on a congested road, the learning data is not updated unnecessarily. Therefore, it is particularly desirable to reflect the driver's shift preference, so that the accuracy of shift-up control is preferable. Maintained at.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION A shift control device for an automatic transmission for a vehicle according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of the main part, and FIG. 2 is a schematic configuration diagram showing the overall configuration. First, the overall configuration of the shift control device of the present invention will be described with reference to FIG.

As shown in FIG. 2, the shift control device of the present invention is a diesel engine (hereinafter referred to as an engine).
This is a system for automatically changing the rotational driving force of 11 using a gear type transmission (hereinafter referred to as a speed change mechanism) 17 having a clutch 15. Here, the speed change mechanism 17 has seven forward speeds in addition to the reverse speed, and not only automatic speed change but also manual speed change is possible.

The engine 11 is provided with a fuel injection pump (hereinafter referred to as an injection pump) 21 having a pump input shaft 19 that rotates at a speed half that of the engine output shaft 13, and this injection pump 21. An electromagnetic actuator 25 is connected to the control rack 23. A rack position detection sensor 123 for detecting the position of the control rack 23 is also provided. Further, the pump input shaft 19 is provided with an engine rotation sensor (engine rotation speed detection means) 27 for detecting the rotation speed signal of the output shaft 13 of the engine 11 and detecting the engine rotation speed Ne.

An exhaust pipe 12a for guiding exhaust gas extends from the engine 11 through an exhaust manifold 12, and an exhaust brake device 119, which is one of engine auxiliary brakes, is interposed in the exhaust pipe 12a. .
The clutch 15 is provided with an air cylinder 33 as a clutch actuator. This clutch 15
Is a state in which the clutch plate 31 is pressed against the flywheel 29 by a well-known sandwiching means (not shown) to be in a connected state. That is, when the air cylinder 33 shifts from the non-actuated state to the actuated state, the holding means operates in the releasing direction, whereby the clutch 15 changes from the connected state to the disengaged state.

Further, the clutch 15 is provided with a clutch stroke sensor 35 for detecting disconnection and connection information based on the clutch stroke amount. A clutch touch sensor 37 may be used instead of the clutch stroke sensor 35. Transmission mechanism 1
7, an input shaft 39 has a clutch rotation sensor 41 for detecting the rotation speed of the input shaft 39, that is, the rotation speed NCL of the clutch 15.
Is attached.

By the way, an air passage 43 is connected to the air cylinder 33, and the air cylinder 33 is provided with a pair of air tanks 47 as an air source via a check valve 45.
It is connected to 49. In the middle of the air passage 43, solenoid valves X1 and X2 that are duty-controlled to supply operating air and function as opening / closing means, and solenoid valves Y1 and Y2 that are duty-controlled to open the air in the air cylinder 33 to the atmosphere.
Is provided, and a three-way solenoid valve W is provided upstream of the solenoid valves X1 and X2.

As shown, the solenoid valves X1,
X2 are connected in parallel with each other and are normally closed. The solenoid valves Y1 and Y2 are also connected in parallel with each other and are normally open. These solenoid valves X1 and X2 and solenoid valves Y1 and Y2 are used alternately. The solenoid valve W is controlled to connect the air tanks 47 and 49 to the air passage when the air cylinder 33 is turned on, and to open the air passage to the atmosphere when the air cylinder 33 is turned off.

Here, of the pair of air tanks 47, 49, the air tank 49 is an emergency tank, and if the main air tank 47 runs out of air for some reason, the solenoid valve 55 is opened and air is removed from the emergency air tank 49. Supply. For this reason, air sensors 57 and 59 are attached to the air tanks 47 and 49, which output an ON signal when the internal air pressure falls below a specified value.

The air tank 47 has an air passage 43.
An air passage that is different from the above, and is connected to an air passage that branches into two systems on the downstream side. The tip of the air passage is provided with a pair of air masters 10 in a braking system (air over hydraulic type) via a pair of solenoid valves MVQ111, 111 that operate according to a signal from a brake sensor 87, which will be described later.
It is connected to 9,109.

These air masters 109, 109 include
A strong brake pedal force sensor (BPS sensor) 106 is attached. The BPS sensor 106 is configured as a diaphragm-type opening / closing switch that is turned on when receiving information on a strong brake pedal force corresponding to an air pressure when a strong braking force exceeding a set value is required. The change lever 61 is a select lever serving as a gear shift operation means for the speed change mechanism 17, and can be moved in the select direction and a direction orthogonal to the select direction as shown in FIG. It is possible to move from the position moved in the direction to the shift direction parallel to the select direction.

The select pattern and shift pattern in each of these directions will be described. First, in the select direction, N (neutral) range and R (reverse) range are selected.
D corresponding to range and automatic shift mode (auto mode)
Each range of (drive) range is set. Further, in the shift direction, that is, in the M (manual) range corresponding to the manual shift mode (manual mode), the above D
A reference position set at a position where the change lever 61 is moved from the (drive) range in the direction orthogonal to the above, that is, a HOLD (hold) range (shown as HOLD in FIG. 3) is sandwiched between the UP (shift up) position and DOW
An I-type shift pattern including N (shift down) position is set.

In such a select pattern and shift pattern, the change lever 61 located in the N range, R range, and D range is held at that position even if the driver's hand is released after the select operation to that position. If the driver's hand is released after the operation when the M range is selected and the M range is selected and then the shift operation is performed to the UP position or the DOWN position, the change lever 61 automatically returns to the HOLD range. Held in that position. The detection of each range and position of the change lever 61 is performed by the shift stage selection switch 63, and this detection signal is supplied to the control unit 71 as shift control means. Then, the gear shift unit 65 is operated according to this detection signal, and the gears in the transmission mechanism 17 are switched according to the select range and the shift position.

The gear shift unit 65 includes a plurality of solenoid valves (only one of which is shown in FIG. 2) 73 operated by an operation signal from the control unit 71, a select fork and a shift fork in the speed change mechanism 17 (both shown in FIG. And a pair of power cylinders (not shown) for operating the power cylinder (not shown). This power cylinder operates when high-pressure operating air is supplied from the above-mentioned air tanks 47 and 49 via the solenoid valve 73. That is, the solenoid valve 7
Each of the power cylinders is operated by the operation signal provided to the gear 3, and the meshing state of the gears of the gear type speed change mechanism 17 is changed in the order of select and shift.

Further, the gear shift unit 65 is provided with a gear position switch 75 as a gear position sensor for detecting each shift stage, and a gear position signal from this gear position switch 75 is output to the control unit 71. Further, a vehicle speed sensor 79 for detecting a vehicle speed is attached to the output shaft 77 of the speed change mechanism 17, and an accelerator pedal 81 has an accelerator opening degree for detecting a depression amount (accelerator opening degree VA) as engine load information. A sensor (accelerator opening detection means) 85 is provided. The accelerator opening sensor 85 detects a resistance change according to the amount of depression of the accelerator pedal 81 as a voltage value (VA), and the A / D converter 83 converts the detected resistance value into a digital signal and outputs the digital signal. FIG. 4 shows a map showing the relationship between the accelerator opening and the voltage value (VA), and the accelerator opening VA is set based on this map.

A brake sensor 87, which outputs a high-level brake signal to the control unit 71 when the brake pedal 69 is depressed, is attached to the brake pedal 69. A starter 89 is mounted to start the engine 11 by meshing with each other at appropriate times. Starter 8
9 is provided with a starter relay 91, which is also connected to the control unit 71.

Reference numerals 120 and 121 in the figure respectively denote an exhaust brake device 119 which is an engine auxiliary brake.
And an exhaust brake on / off switch and an engine brake auxiliary device on / off switch for switching a compression release type engine auxiliary brake device (not shown) as an engine brake auxiliary device between a working standby state and a non-operating state. It is arranged in the vicinity.

Reference numeral 93 in FIG. 2 indicates an engine control unit provided separately from the control unit 71. This engine control unit 93
Is the injection pump 2 according to the accelerator opening information VA and the like from the control unit 71 based on the information from each sensor.
The electronic governor 25 in 1 is controlled. That is, when the electronic governor 25 receives a command signal from the engine control unit 93, the control rack 23 operates to increase / decrease the fuel.
The drive control of 1 is performed to control the increase / decrease in the rotation speed of the output shaft 13.

The control unit 71 comprises a microcomputer (CPU) 95, a memory 97 and an interface 99 as an input / output signal processing circuit. In the input port (input interface) 101 of the interface 99, the above-described gear position selection switch 63, brake sensor 87, accelerator sensor 85, engine rotation sensor 27, clutch rotation sensor 41, gear position switch 75, vehicle speed sensor 79, clutch. Stroke sensor 35, clutch touch sensor 37 (clutch 1
5 is used to output the connection / disconnection information in place of the clutch stroke 35), the air sensors 57 and 59, the BRS sensor 106 that outputs the strong brake pedal force information, the exhaust brake on / off switch 120, the engine brake auxiliary device on / off switch 121, and the rack. Position detection sensor 123
Alternatively, a slope start switch 103 described later is connected, and detection information is input to the control unit 71 from each of these sensors.

The slope start switch 103 is for operating a system (hereinafter referred to as AUS) for preventing backward movement when the vehicle starts uphill. This AUS is
Air master 1 for a plurality of wheel brakes 107, 107
09,109 is supplied with air by a pair of solenoid valves MVQ
This is a system for starting a vehicle while controlling it via 111, 111.

On the other hand, the output port (output interface) 113 is provided with the above-mentioned engine control unit 93, starter relay 91, solenoid valves X1, X2, Y.
1, Y2, W and solenoid valves 55, 73, 111 etc. are connected respectively. In the figure, reference numeral 115 indicates an air warning lamp which is turned on by receiving a detection signal from the air sensors 57 and 59 when the air pressure in the air tanks 47 and 49 has not reached a set value, and reference numeral 117 indicates wear of the clutch 15. A clutch warning lamp, which lights when receiving the detection signal when the amount exceeds a specified value, indicates a stop lamp switch which is turned on when the brake pedal 69 is depressed.

The memory 97 comprises a read-only ROM in which various flowcharts are written as programs and data and a writable RAM. In the ROM, in addition to the control program, the duty ratios of the solenoid valves X1, X2, Y1, Y2 corresponding to the accelerator opening information VA are stored in advance as a map.
Reads the appropriate value from this map as appropriate.

Then, the above-described gear stage selection switch 63
Outputs a select signal and a shift signal as a shift signal in response to the operation of the change lever 61. The ROM stores the shift speed positions corresponding to a pair of these signals in advance as a plurality of data maps. ing. Therefore, when the control unit 71 receives the select signal and the shift signal, the CPU 95 selects a predetermined map from a plurality of data maps, calculates an output signal from this map, and further outputs this output signal to the gear shift unit 65.
To each electromagnetic valve 73, and the gear is adjusted to the target shift speed corresponding to the shift signal. The gear position signal from the gear position switch 75 is output when the shift is completed, and it is determined whether all the gear position signals corresponding to the select signal and the shift signal have been output. That is, this gear position signal is mainly used for issuing a signal indicating whether or not the meshing is normal.

Further, the ROM also has a shift map for determining the optimum shift speed based on the values of the vehicle speed V, the accelerator opening VA and the engine speed Ne when the target shift speed in the D range exists. Multiple items are stored. The shift map is roughly divided into # 1, # 2, and # 3 as shown in Table 1 as a selection map according to the on / off states of the foot brake signal and the exhaust brake signal from the brake sensor 87 and the exhaust brake on / off switch 120. Of 3
The type is set. The three types are set in this manner in order to improve the driving feeling on a downhill. Normally, the selection map # 1 is applied when the braking is not performed.

[0039]

[Table 1]

In addition, since the driving torque required for buses and trucks usually changes according to the loading condition and the operating condition, the ROM indicates this loading condition (for example, vehicle load degree αVL, which will be described later). There are also stored a plurality of shift maps that can be switched automatically or based on the driver's intention according to the situation and the driver's preference described later. There are, for example, three types of shift maps corresponding to the loading condition and the operating condition, such as an economy mode for fuel efficiency, a normal mode, and a power mode for acceleration, which are respectively # 1, # 2, and # above. 3 is provided for each shift map. That is, the ROM stores a total of nine types of shift maps in terms of the relationship between the vehicle speed V and the accelerator opening degree VA, for example. FIGS. 5 to 13 show these nine types of shift maps, and FIGS.
Shift maps of # 1, # 2, and # 3 in the economy mode are shown, and FIGS. 8 to 10 show # 1 and # in the normal mode.
11 and 13 show shift maps of Nos. 2 and 3, respectively.
Shows each shift map of # 1, # 2, and # 3 in the power mode. That is, the CPU 95 appropriately selects a shift map according to the situation from these shift map groups. In these shift maps, shift-up shift characteristics are shown by solid lines, and shift-down shift characteristics are shown by broken lines.

By the way, the basic operation of the above-described automatic transmission is known, and a detailed description thereof will be omitted here.
This automatic transmission operates, for example, in the same manner as that disclosed in Japanese Utility Model Laid-Open No. 2-49663. Next, the control content of the control unit 71, which is an essential part of the present invention, will be described.

Referring to FIG. 1, for the control unit 71 of this automatic transmission, a suitable shift map is selected from the above shift map group, vehicle speed information V from the vehicle speed sensor 79 and an accelerator from the accelerator opening sensor 89 are selected. Target shift speed setting means 3 for setting a target shift speed from the selected map based on the opening degree information VA, and optimum shift speed determining means 1 capable of correcting the target shift speed by reflecting the driver's intention. ing.

Further, the target shift speed setting means 3 has a storage means 3A for storing a shift map and a learning formula selection means 3B.
Are provided, and the storage means 3A stores the five shift maps described later in addition to the nine shift map groups shown in FIGS.
-4 downshifts and 4-3 downshifts Downshift restriction map (see FIG. 24) for restricting each selection map #
It is stored for each mode of # 1, # 2, and # 3. Therefore,
The target shift speed setting means 3 is configured as a map type storage means.

Further, the learning-type selecting means 3B of the target shift speed setting means 3 determines the driver based on the vehicle speed information V, the engine speed information Ne, the detection signal from the shift speed selecting switch 63, the accelerator opening information VA, and the like. It has a function of performing learning that reflects driving preference and selecting an optimum shift map from among the above modes in accordance with the learning result.
The driver's driving preference refers to, for example, a driving method such as whether he / she wants to shift up early with an emphasis on fuel consumption, or whether he / she wants to drive with emphasis on acceleration. Therefore, the learning formula selecting means 3B learns the driving operation of the driver using the driving operation information by the driver as input information, and selects the optimum one from the plurality of shift maps in the storage means 3A according to the driver's preference and personality. The shift map of the mode (economy mode with emphasis on fuel consumption, normal mode, or power mode with emphasis on acceleration) will be selected.

More specifically, the learning formula selection means 3B is shown in FIG.
21 to 21. It is configured by including a plurality of neural networks as shown in FIG. 21, and when various driving operation information is input to these neural networks, arithmetic processing including learning is performed in the neural network, and as a result, the driver is driven. The optimum output signal (judgment output) according to the user's preference is output. Then, the shift map of the mode is selected according to the output signal from each neural network, and the target shift speed corresponding to the vehicle speed V and the accelerator opening degree VA is set based on this shift map.

Further, the optimum shift stage determining means 1 determines the running state of the vehicle and the driver's intention using the fuzzy theory, and can correct the target shift stage set by the target shift stage setting means 3 by the fuzzy formula optimum. It is configured as a gear stage determining means. In this fuzzy type optimum shift speed determination means 1, the running state of the vehicle is estimated based on the fuzzy rule set in advance by experiments or the like using the vehicle information having the vehicle load information as a parameter, thereby inferring the driver's intention. However, in the vehicle load information, the difference between the acceleration α0 when the vehicle accelerates on a straight flat road in an empty state and the actual acceleration α when the vehicle actually accelerates (= vehicle load degree αVL)
Is used as a parameter.

Therefore, as shown in FIG. 1, the control unit 71 is provided with a vehicle load degree calculating means 2 for calculating the vehicle load degree αVL. The vehicle load degree calculating means 2 comprises an engine torque calculating means 4, a driving force calculating means 5, an air resistance calculating means 6, a straight flat road empty vehicle equivalent acceleration calculating means 7, and a subtracting means 8. Of these, the engine torque calculation means 4 calculates the engine torque Te from the position information SRC of the control rack 23 and the engine speed information Ne supplied from the rack position detection sensor 123, and is a target shift speed setting means. Similar to No. 3, it is configured as a map type storage means.

As described above, the engine torque Te is obtained from the position information SRC of the control rack 23 and the engine speed information Ne on the basis of a preset map (not shown). Since it varies depending on the use of the compression opening type engine auxiliary brake, in a situation where the exhaust brake or the compression opening type engine auxiliary brake is used, a map (not shown) set separately according to this situation is used.

Further, the driving force calculation means 5 has the engine torque information Te obtained by the engine torque calculation means 4.
The driving force F of the vehicle is calculated based on The driving force F is calculated, for example, by the following formula. F = (Te · it · if · η) / R where it is the gear ratio of the gear, if is the final reduction gear ratio (differential gear ratio), η is the power transmission efficiency, and R is the tire radius.

The air resistance calculating means 6 calculates the air resistance Rl as the running resistance of the vehicle from the actual vehicle speed information V, which is calculated by the following equation. Rl = λ · A · V ² where λ is the air resistance coefficient, A is the front projected area of the vehicle, and V is
Is the actual vehicle speed. Next, the straight flat road empty vehicle equivalent acceleration calculation means 7 (hereinafter, referred to as acceleration calculation means) 7 will be described. The acceleration calculation means 7 includes the driving force information F calculated by the driving force calculation means 5 and the air resistance. From the air resistance coefficient information Rl calculated by the calculation means 6, the acceleration α0 when the vehicle accelerates on a straight flat road in an empty state, that is, the straight flat road empty vehicle equivalent acceleration is calculated. This straight flat road empty vehicle equivalent acceleration α0 is calculated by the following equation using the driving force F of the vehicle.

Α0 = g {F- (μW0 + Rl)} / (W0 + Wr) where g is gravitational acceleration, μ is a road surface friction coefficient, W0 is an empty vehicle weight, and Wr is a rotating portion weight. Then, the subtracting means 8 calculates the vehicle load degree information αVL by the following equation based on the straight flat road empty vehicle equivalent acceleration information α0 calculated by the acceleration calculating means 7 and the actual acceleration information α from the clutch rotation sensor 41. The vehicle load degree information αVL corresponds to the weight of the vehicle and the gradient resistance of the vehicle.

ΑVL = α0−α That is, it can be said that the vehicle load is heavy when αVL> 0, and the vehicle load is light when αVL <0. Then, from the magnitude of the value of αVL, it is determined how heavy (or light) the vehicle load is. FIG. 14 shows an example in which the vehicle load degree αVL is calculated by simulation on a flat road. Here, the horizontal axis is the total vehicle weight gvw (gloss vehicle weight), and the vertical axis is the vehicle load αVL.

When the vehicle load degree .alpha.VL is calculated by the vehicle load degree calculating means 2 in this way, the fuzzy optimum gear shift stage determining means 1 adds the accelerator opening information VA and accelerator to the vehicle load degree .alpha.VL. The target shift speed setting means 3 is loaded with various information such as the opening change information ΔVA, the vehicle speed information V, the engine speed Ne, the brake information and the current shift speed information.
Correction is performed for the target shift speed set in. As the brake information, in addition to the foot brake operation information input from the brake sensor 87, operation information such as an exhaust brake and a compression opening type engine auxiliary brake is also input. The information is also taken into account for correction.

More specifically, the optimum shift stage determining means 1 quantifies each information by the fuzzy inference method using the min-max combined center of gravity method, and these quantified values are preset by a predetermined fuzzy logic. The target gear is corrected in light of the rules. Since the content of this fuzzy inference is not directly related to the present invention, its detailed description is omitted here.

As described above, in the shift control device for the automatic transmission for a vehicle according to the embodiment of the present invention, the target shift speed according to the driver's preference set by using the neural network is finally fuzzy. The optimum shift speed is corrected by the control, and the optimum shift speed is determined by the control unit 71 in FIG.
It is determined according to the flowchart of the main routine as shown in FIG. Hereinafter, the procedure for determining the automatic shift, that is, the optimum shift stage in the D range will be described in detail with reference to FIG.

First, in step S10 of FIG.
As described above, the engine torque Te is calculated by the engine torque calculating means 4. Then, step S12
Then, the driving force F of the vehicle is calculated by the driving force calculation means 5,
In the next step S14, the air resistance calculating means 6 calculates the air resistance Rl of the vehicle. In step S16, the linear flat road empty vehicle equivalent acceleration α0 is calculated by the linear flat road empty vehicle equivalent acceleration calculating means 7 from the driving force information F calculated in step S12 and the air resistance information Rl calculated in step S14.

Then, in step S18, step S
The load degree information αVL of the vehicle is calculated based on the straight flat road empty vehicle equivalent acceleration α0 calculated in 16 and the actual acceleration information α from the clutch rotation sensor 41. Next step S19
Then, the learning formula selection means 3B of the target shift speed setting means 3 performs the learning control of the shift according to the driver's preference and selects the optimum shift map from the storage means 3A. Hereinafter, the learning control of the shift according to the driver's preference will be described in detail with reference to the flowchart of FIG. Note that here, since the effect of learning is great, 2-3 upshift and 3-
The four-upshift is targeted for learning, and further, when the vehicle is traveling on a flat road where the driver's preference is easily dispersed.

First, in step S50 of FIG. 16, data necessary for learning the shift according to the driver's preference is calculated (learning data calculating means). The following are examples of learning data. (1) 2-3 average accelerator opening AVA (23) (2) 2-3 standard deviation SVA (23) (3) 3-4 average accelerator opening AVA (34) (4) 3-4 standard deviation SVA ( 34) (5) 2-3 engine rotation average ANe (23) (6) 3-4 engine rotation average ANe (34) (7) Hold range usage frequency FH during deceleration (8) M range usage frequency FMD during deceleration Referring to FIG. 17, there is shown a flowchart showing a routine for calculating the accelerator opening average AVA, the standard deviation SVA, and the engine rotation average ANe, and the AVA (23), AVA (listed in (1) to (6) above. 34), SVA (23),
SVA (34), ANe (23) and ANe (34) are calculated based on the flowchart of FIG. Hereinafter, based on FIG. 17, AVA (23), AVA (34), SVA (23),
The calculation procedure of SVA (34), ANe (23) and ANe (34) will be described.

First, in step S70 in FIG. 17, the vehicle speed V is equal to or higher than a predetermined vehicle speed V2 (for example, 20 km / h) (V ≧ V).
2) is determined (congested road running determination means). This means that the vehicle is traveling on a congested road at a low vehicle speed less than the predetermined vehicle speed V2. Here, a threshold value (for example, 2
0 km / h) is the lowest vehicle speed at which the vehicle can normally be regarded as not traveling on a congested road. The judgment result is true (Yes)
If the vehicle speed V is equal to or higher than the predetermined vehicle speed V2, it can be determined that the vehicle is not traveling on a congested road. In this case, the process proceeds to step S71.

In step S71, the vehicle speed V is the predetermined vehicle speed V.
It is determined whether or not it is 2 or more immediately after. If the determination result is false, the process proceeds to step S78, and AVA (23), A
VA (34), SVA (23), SVA (34), ANe (23), ANe
Calculate (34). AVA (23), AVA (34), SVA (2
The calculation procedure of 3), SVA (34), ANe (23), ANe (34) is as follows and will be described in order. 1) 2-3 accelerator opening average AVA (23): 2-3 accelerator opening average AVA (23) is monitored by the control unit 71 (monitoring means) from the second gear to the third gear in the D range. It is the average value of the accelerator opening degree VA at the start of the shift (at the time of switching the shift stage), and the average of the past 10 times on a flat road is obtained from the following equation so that sufficient accuracy can be obtained.

AVA (23) = (VA (23) 0 + VA (23) 1 + ·
.. + VA (23) 9) / 10 2) 2-3 standard deviation SVA (23): 2-3 standard deviation SVA
(23) is the standard deviation (3σ) of the accelerator opening VA when shifting from the 2nd gear to the 3rd gear in the D range. Again, the standard deviation for the past 10 times on a flat road is calculated from the following formula. .

SVA (23) = 3 · ((1/10) · Σ (VA
(23) i-AVA (23)) ² ) ^1/2 , i = 0 to 9 However, here, in order to facilitate the calculation processing by the CPU 95, the actual value is ( SVA
The square of (23), that is, the following equation is used for convenience.

(SVA (23)) ² = 9 · (1/10) · Σ (V
A (23) i-AVA (23)) ² , i = 0-9 3) 3-4 Average accelerator opening AVA (34): 3-4 Average accelerator opening AVA (34) is also the control unit 7
It is the average value of the accelerator opening degree VA at the start of shifting from the 3rd speed to the 4th speed in the D range monitored by 1, and the average of the past 10 times on a flat road is calculated from the following equation.

AVA (34) = (VA (34) 0 + VA (34) 1 + ·
.. + VA (34) 9) / 10 4) 3-4 standard deviation SVA (34): 3-4 standard deviation SVA
(34) is the standard deviation (3σ) of the accelerator opening VA at the time of shifting from the 3rd speed to the 4th speed in the D range. Again, the standard deviation for the past 10 times on a flat road is calculated from the following formula. .

SVA (34) = 3 · ((1/10) · Σ (VA
(34) i-AVA (34)) ² ) ^1/2 , i = 0 to 9 However, here, as in the case of the above SVA (23), the actual value is (SVA (23) instead of (SVA (23)). The square of (34), that is, the following equation is used for convenience. (SVA (34)) ² = 9 · (1/10) · Σ (VA (34) i-A
VA (34)) ² , i = 0 to 95) 2-3 Engine rotation average ANe (23): 2-3 engine rotation average ANe (23) is still control unit 7
It is the average value of the engine speed at the time of shifting from the 2nd speed to the 3rd speed by the manual operation in the M range monitored by 1, and the average of the past 5 times on a flat road is obtained from the following formula. Specifically, the engine speed Ne just before the shift starts
Calculated from the average value of.

ANe (23) = (Ne (23) 0 + Ne (23) 1 + ...
・ + Ne (23) 4) / 5 6) 3-4 engine rotation average ANe (34): 3-4 engine rotation average ANe (34) is still control unit 7
It is the average value of the engine speed at the time of shifting from the 3rd speed to the 4th speed by the manual operation in the M range monitored by 1. Also, the average of the past 5 times on the flat road is obtained from the following formula. Similarly to the above, it is obtained from the average value of the engine speed Ne immediately before the shift is started.

ANe (34) = (Ne (34) 0 + Ne (34) 1 + ...
-+ Ne (34) 4) / 5 The determination result of step S71 is true and the vehicle speed V is the predetermined vehicle speed V.
If it is immediately after 2 or more, next step S7
Proceed to 2. In step S72, it is determined whether or not the upshift in the D range has been performed at the stage where the determination result in step S70 is true and the vehicle speed V is less than the predetermined vehicle speed V2 (V <V2) before becoming equal to or higher than the predetermined vehicle speed V2. Determine. If the determination result is true, the process proceeds to step S73.

In step S73, the vehicle speed V is the predetermined vehicle speed V.
The accelerator opening VA at the latest shift up in the D range, which was last executed in the stage of less than 2, is adopted as effective learning data. Then, the process proceeds to step S78. On the other hand, if the determination result of step S72 is false (No), and the upshift in the D range has not been performed at a stage below the predetermined vehicle speed V2, the process proceeds to step S74.

In step S74, the vehicle speed V is the predetermined vehicle speed V.
At a stage where the vehicle speed is less than the predetermined vehicle speed V2 before being equal to or more than 2, it is determined whether or not the manual shift up in the M range is performed. If the determination result is true, then step S
Proceed to 76. In step S76, the vehicle speed V is the predetermined vehicle speed V.
The engine speed Ne at the latest manual shift up in the M range, which was last executed in a stage of less than 2, is adopted as effective learning data. And step S7
Proceed to 8.

If the determination result of step S74 is false, that is, if the vehicle speed V is less than the predetermined vehicle speed V2, the upshift is not executed in the D range (step S7).
2) Further, in the case where the manual shift up in the M range is not performed (step S74), for example, after the vehicle starts, the vehicle speed V suddenly reaches the predetermined vehicle speed V2 or more with the second speed being the second speed. In such a case, the routine is ended without doing anything. That is, in this case, the calculation of the learning data as performed in step S78 is not performed.

By the way, the determination result in step S70 is false, that is, the vehicle speed V is still the predetermined vehicle speed V2 (for example,
Even if it has not reached 20 km / h), the routine is ended without doing anything. That is, in a situation where the vehicle speed V has not yet reached the predetermined vehicle speed V2 and the vehicle can be regarded as traveling on a congested road, the driver's favorite learning is not performed. Normally, when a vehicle is driving on a congested road, upshifting and downshifting are often carried out according to the traffic situation rather than the driver's preference and intention, and thus the driver's preference is good. If it is not reflected in, the adoption of learning data is canceled.

In Table 2, for example, the relationship between the time change of the vehicle speed V in 2-3 upshifting in the D range and the adoption or non-adoption of the learning data (here, the accelerator opening VA is shown as an example) is simply shown. There is. As shown in the table, when the vehicle speed V has not yet reached the predetermined vehicle speed V2 (for example, 20 km / h), the learning data is not adopted and the vehicle speed V is the predetermined vehicle speed V.
When it becomes 2 (for example, 20km / h) or more,
It is determined that the traffic congestion has ended and the vehicle has shifted to normal driving, and the learning data immediately before the upshift is adopted as valid data.

As described above, the learning data at the time of upshifting immediately before the vehicle speed V becomes equal to or higher than the predetermined vehicle speed V2 is adopted as valid data. That is, at this time, the gear shift from the second gear to the third gear This is because the driver's preference for shifting during acceleration appears in the accelerator opening VA at that time. As for the M range, as in the case of the D range, when the vehicle speed V becomes equal to or higher than the predetermined vehicle speed V2, the learning data at the immediately preceding shift up is adopted as effective data.
This is because, at this time, the driver's preference for shifting at the time of acceleration has already appeared in the engine speed Ne at the time of shifting from the second speed to the third speed.

[0074]

[Table 2]

That is, in this learning, when the vehicle is traveling on a traffic jam road in which the vehicle speed V does not reach the predetermined vehicle speed V2 (for example, 20 km / h), upshifting or downshifting is actually performed. If so, these shifts are treated as if they were not performed and the learning data is not updated. In addition, (7) HOLD
Range use frequency FH and (8) M range use frequency FMD
Is calculated based on the flowchart of the FH, FMD calculation routine shown in FIG. Hereinafter, the calculation procedure of the HOLD range usage frequency FH and the M range usage frequency FMD will be described with reference to FIG.

The HOLD range use frequency FH and the M range use frequency FMD are defined as the average values of the individual HOLD range use frequency FHi and the M range use frequency FMDi. Therefore, first, in step S80 of FIG.
The OLD range use frequency FHi and the M range use frequency FMDi are calculated. The HOLD range use frequency FHi is determined by the change lever 61 from the D range, the UP position, or the DOWN position in response to the target gear shift information in the normal D range shift down mode when the vehicle is in the predetermined deceleration operation. It is calculated based on the frequency of the number of times the gear is switched to the HOLD range and the gear is held at the current gear.

Specifically, the frequency of use of the HOLD range FHi
Means that when the vehicle speed V is equal to or higher than a predetermined vehicle speed V1 (for example, 30 km / h), the accelerator pedal 81 is released and the accelerator opening V
The vehicle speed V is gradually decreased after A is set to 0%, and the engine speed Ne is set to the predetermined engine speed Ne1 and is performed during a predetermined operating state until the clutch 15 is automatically disengaged. The total number of downshifts NSi for the fifth gear or lower in the normal downshift mode of the D range, NSi, and H
It is calculated from the following formula based on the number NHi of canceling the downshift using the OLD range. The normal downshift mode in the D range refers to a mode that uses a map other than the above-mentioned downshift restriction map (hereinafter, referred to as a downshift enable mode).

FHi = (number of cancellations NHi) / (total number of downshifts NSi in 5th gear and lower), i = 0 to 4 where the predetermined engine speed Ne1 is equal to the brake pedal 6
When 9 is operated and the foot brake is used, the speed is, for example, 600 rpm (Ne1 = 600 rpm), and when the foot brake is not used, the value is smaller than 600 rpm, for example, 450 rpm (Ne1 = 450 rpm). . At this time, the vehicle speed V corresponding to Ne1 = 600 rpm is
For each gear, for example, the speed is 20 km / h for the fifth speed, 14 km / h for the fourth speed, and 8.5 km / h for the third speed.

On the other hand, the M range use frequency FMDi during downshifting is that the accelerator pedal 81 is released and the accelerator opening degree VA is 0% when the vehicle speed V is equal to or higher than a predetermined vehicle speed V1 (for example, 30 km / h). The vehicle speed V gradually decreases from
The total number of downshifts NDi of the fifth speed or lower actually performed during a predetermined operating state until the engine speed Ne reaches the predetermined engine speed Ne1 and the clutch 15 is automatically disengaged (the above total shift). The number of downshifts NSi is different from the number of downshifts NSi, including manual operation) and the number of times NMDi of downshifting by manual operation using the M range are calculated from the following equation. Here, the calculation of the M range use frequency FMDi is performed by calculating the total number of downshifts and the number of NDi NDi.
Is set on the basis of the detection signal from the shift stage selection switch 63, the output frequency of the shift state switching signal is obtained.

FMDi = NMDi / NDi, i = 0 to 4 In the next step S82, it is determined whether or not the downshift mode is a mode using the downshift restriction map (hereinafter referred to as downshift non-operating mode). . If the determination result is true and the downshift mode is the downshift mode, the process proceeds to step S84.

In step S84, the above step S80 is performed.
The value 1 is set to the HOLD range use frequency FHi calculated in step irrespective of the above calculation result. Details will be described later,
In the case where the downshift mode is the downshift mode, the 5-4 downshift of the fifth speed or lower, 4-
Since the 3 downshift is canceled in the region where the vehicle speed is low and the accelerator opening is small (see FIG. 24), the usage frequency FHi becomes extremely close to the value 1 in light of the above formula. Therefore, in this step S84, the value 1 is set again for the HOLD range usage frequency FHi to clarify that the HOLD range usage frequency is high. As a result, the learning by the learning formula selection means 3B, that is, the learning preferred by the driver is made more suitable and highly accurate.

On the other hand, the determination result of step S82 is false,
When the downshift mode is not the downshift mode but the normal downshift enabled mode, the process proceeds to step S86. In step S86, it is determined whether or not the downshift of the fifth speed or lower is manually performed in the M down range in the shift down possible mode. If the determination result is true and it is determined that the downshift is performed by the manual operation, the process proceeds to step S88, and this time the HOLD range use frequency FHi.
Is set to 0. In other words, if the downshift of the 5th gear or lower in the M range is manually performed even once, the driver is using the HOLD range to cancel the downshift in the D range. Is unlikely to occur, and in this case, it can be determined that the manual operation is actively performed. Therefore, here, the value 0 is set to the HOLD range use frequency FHi and it is not used as an input parameter, and it is made clear that the driver intends to perform a manual operation.

By the way, the determination result of step S86 is false, it is determined in step S82 that the downshift mode is the downshift enabled mode, and it is determined that the downshift by the manual operation of the fifth speed or lower is not performed. In this case, the value of the HOLD range use frequency FHi calculated in step S80 is applied as it is, and the process proceeds to step S90.

In step S90, the determination result in step S82 is true, and the execution of step S84 results in HO.
Whether the LD range use frequency FHi is 1 or whether the determination result in step S82 is false and the determination result in step S86 is also false, and there was no downshift by the manual operation of the fifth speed or lower. To determine. If the determination result is true, that is, if it is determined in step S88 that the use frequency FHi is not set to the value 0, the process proceeds to step S92.

In step S92, the value 0 is set to the M range use frequency FMDi this time. That is, step S90
If the result of the determination is true and the frequency of use FHi is not 0, the driver wants to cancel the downshift using the HOLD range, and it can be determined that the driver is reluctant to manually shift. Therefore, for one of the M-range use frequencies FMDi, the value 0 is set so that the learning result by the learning formula selection means 3B is clearer and more accurate, and is not used as an input parameter.

On the other hand, if the determination result in step S90 is false, the value of the use frequency FMDi calculated in step S80 is applied as it is, and the process proceeds to step S94. In step S94, based on the usage frequency FHi and usage frequency FMDi set as described above, finally HO
Obtain the LD range usage frequency FH and the M range usage frequency FMD. Specifically, the frequency of use FH and the frequency of use FMD are the past 5 of the frequency of use FHi and the frequency of use FMDi from the following equations, respectively.
Calculate by taking the average of the times.

FH = (FH0 + FH1 + ... + FH4) / 5 FMD = (FMD0 + FMD1 + ... + FMD4) / 5 In this way, the average values of the individual HOLD range usage frequency FHi and M range usage frequency FMDi for the past five times are calculated. By finally setting the holding range use frequency FH and the M range use frequency FMD, learning is stabilized.

Returning to FIG. 16, in the next step S52,
For the reasons described above, it is determined whether or not the vehicle is traveling on a flat road. Here, it is determined whether or not the vehicle load degree αVL is within a predetermined range (near the zero value) set for each shift speed. By thus determining whether or not the vehicle is traveling on a flat road, it is possible to limit the learning to only the flat road on which the driver's preference is significantly exhibited.

If the determination result of step S52 is false (No),
That is, when it is determined that the vehicle load degree αVL is outside the predetermined range and the vehicle is not traveling on a flat road, the process proceeds to step S64 without performing the upshift learning. On the other hand, the determination result of step S52 is true (Yes
s), that is, when it is determined that the vehicle load degree αVL is within the predetermined range and the vehicle is traveling on a flat road, the process proceeds to step S54.

In step S54, it is determined whether or not 2-3 upshifts or 3-4 upshifts have been continuously performed 10 times in the D range. If the determination result is true, it can be determined that almost only the D range is used during upshifting and the M range is used very rarely. In this case, the process proceeds to step S56. In step S56, the upshift preference is learned based on the upshift by the D range operation, and the output necessary for selecting the shift map, that is, the determination output is set. That is,
Here, since the frequency of use of the M range is low, learning based on the automatic shift is not performed, but learning based on the use of the M range is not performed. More specifically, the first neural network shown in FIG. 19 performs learning based on the back propagation learning method and calculation of the judgment output. The processing in the first neural network will be described below.

In the input layer of the first neural network of FIG. 19, the previously selected shift map information, the above-mentioned 2
-3 Accelerator opening average AVA (23), 2-3 standard deviation SV
A (23), 3-4 accelerator opening average AVA (34) and 3-4
Input values I1, I2, I3, I4, I of standard deviation SVA (34)
5 is entered. Here, the shift map information selected last time is the last judgment output, and has the character as the reference value for the current learning. Then, these input values of I1 to I5 are normalized by the following equation to obtain Ii '
And input to the middle layer consisting of three pieces.

Ii '= (Ii-Iimin) / (Iimax-Iimi
n), i = 1 to 5 where Iimin is the minimum value of Ii, that is, the minimum value of each input value of I1 to I5, and Iimax is the maximum value of Ii, that is, I1.
It is the maximum value of each input value of I5. Then, when inputting to the three intermediate layers, different coupling coefficients Wij (i = 1 to 5, j = 1 to 3) are added to the respective intermediate layer input values Ii '. That is, 1 is input to each of the input values Ii ′ input from the 5 input layers to the 3 intermediate layers in advance by experiments or the like.
Five coupling coefficients Wij (W11, W12, W13, W2)
1, W22 ... W53) corresponding coupling coefficient Wij is integrated.

Output values O1, O2, O3 calculated by the sigmoid function of the following equation are output from the three intermediate layers. Oj = 1 / (1 + exp (-Xj)), j = 1 to 3, where Xj = W1j.I1 '+ W2j.I2' + ... + W5j.
I5 '-[theta] j, j = 1 to 3 Here, [theta] j (j = 1 to 3) is a predetermined threshold value set in advance by experiments or the like.

For these output values O1, O2 and O3, the preset coupling coefficient Wj4 (j = 1 to 3) is integrated and input to the output layer. From the output layer, the judgment calculated by the following equation Output O4 is output. O4 = (1 / (1 + exp (-Y4))). (Omax-Om
in) + Omin However, Y4 = W14 * O1 + W24 * O2 + W34 * O3- [theta] 4 where Omax is the maximum value of the judgment output O4 and Omin is the minimum value of the judgment output O4. Further, θ4 is a predetermined threshold value set in advance by experiments or the like.

The above-mentioned Iimin, Iimax, Omin, Om
Detailed descriptions of ax, Wij, Wj4, θj, and θ4 are omitted here. However, since the input value I1 and the judgment output value O4 corresponding to the previously selected shift map information among the input values Ii take values between 1 and 3 as described later, the minimum of the input value I1 is obtained. The value I1min and the minimum value Omin of the output value O4 are the value 1, while the maximum value I1max of the input value I1 and the maximum value Omax of the output value O4 are value 3.
Becomes

As described above, when the frequency of use of the M range at the time of upshifting by the driver is small, the learning is carried out based on the shiftup by the automatic shift in the D range, not by the manual upshift in the M range. Therefore, the output for selecting the shift map, that is, the determination output O4 is set satisfactorily and stably. On the other hand, if the result of the determination in step S54 is false, the frequency of use of the M range is relatively high during upshifting, and the driver is not satisfied with only the D range and frequently uses manual shifting in the M range. In this case, the process proceeds to step S58.

In step S58, it is determined whether or not 2-3 upshift or 3-4 upshift by manual operation in the M range has been continuously performed five times. If the determination result is true, the process proceeds to step S60. In step S60, the preference for shift up is learned based on the M range operation, that is, the shift up by the manual operation, and thereby the output for selecting the shift map, that is, the determination output is set. In detail, learning and calculation of the judgment output are performed by a second neural network as shown in FIG. 20, which is provided separately from the first neural network. The processing in the second neural network will be described below.

In the input layer of the second neural network shown in FIG. 20, the previously selected shift map information, the above-mentioned 2
Input values I'1, I'2, I'3 of the -3 engine rotation average ANe (23) and 3-4 engine rotation average ANe (34) are input. Here, similarly to the above, the shift map information selected last time is the determination output of the previous time, and has the character as the reference value of the current learning. Then, these input values of I1 to I5 are normalized by the following equation to be I'i ', and are input to the intermediate layer composed of three pieces.

Then, the output value O'1, through the intermediate layer,
O'2 and O'3 are obtained, and the judgment output O'4 is finally derived. The arithmetic processing in this second neural network is the same as that in the first neural network. The detailed description is omitted here. In this way, when the M range is frequently used during shift up, learning is performed based on the manual shift up by the M range operation, and it is determined whether to select the shift map according to the driver's preference. The output O'4 will be set appropriately.
In other words, depending on the operating condition of the M range by the driver,
The output indicating which of the economy mode, standard mode, and power mode shift maps is selected is set.

When the determination outputs O4 and O'4 regarding the upshift are set by the processing of the first neural network or the second neural network as described above, the process proceeds to step S62. Note that step S5
If the determination result of 8 is false, the process proceeds to step S62 without passing through the first and second neural networks.

In step S62, the judgment output O4,
A shift map is selected according to O'4. That is, here, the shift map of any one of the economy mode, the standard mode, and the power mode is selected. In particular,
The shift map is selected based on the graph showing the relationship between the judgment outputs O4 and O'4 and the selection map as shown in FIG. For example, when the judgment output O4, O'4 value from the first neural network or the second neural network is 1 or more and less than 1.5 (1≤O4, O'4 <1.5), the economy mode shift map is Selected, the judgment output O4, O'4 value is 1.5 or more and less than 2.5 (1.5≤O
When 4, O'4 <2.5), the standard mode shift map is selected and the judgment output O4, O'4 value is 2.5 or more 3
If less than (2.5 ≦ O4, O′4 <3), the power mode shift map is selected.

Here, the economy mode selection map is
As described above, in the shift map that emphasizes fuel economy, the shift timing at each shift stage is set to the low vehicle speed side with respect to the standard mode having the shift timing similar to the normal D range. Further, the power mode selection map is a shift map that emphasizes acceleration, and the shift-up timing of each shift stage is set to the high vehicle speed side with respect to the standard mode. That is, when the judgment outputs O4 and O'4 change due to the learning, the shift pattern in the D range, which is the automatic shift range, can be suitably switched to the shift pattern according to the driver's preference.

If the determination result of step S58 is false, the determination outputs O4 and O'4 are not newly set, and the previous value is applied as it is. That is,
In this case, the shift map currently adopted is continuously used as it is. After the next step S64,
This is a step of learning the driver's preference during downshifting.

In step S64, the vehicle speed V is the predetermined vehicle speed V.
Accelerator pedal 8 when the speed is 1 (eg, 30 km / h) or more
1 is released (OFF) and the accelerator opening VA is 0.
After that, the vehicle speed V gradually decreases, and the engine speed Ne becomes the predetermined engine speed Ne1.
It is determined whether or not 5 is the automatic clutch disengagement. In addition,
The details of the value of the predetermined engine speed Ne1 are as described above, and the description thereof is omitted here.

When it is determined that the determination result of step S64 is true and the clutch 15 is automatically disengaged after the accelerator opening VA is set to 0%, that is, when the shift is determined to be the downshift mode. Next in step S6
Proceed to 6. In step S66, the downshift preference is determined. That is, the shift-down preference learning is performed according to the driver's use of the M range, and the judgment output necessary for selecting the shift map is set. Specifically, learning and calculation of the judgment output are performed by the third neural network shown in FIG. The processing in the third neural network will be described below.

In the input layer of the third neural network of FIG. 21, the previously selected shift map information, the above-mentioned H
The input values I "1, I" 2, I "3 of the OLD range use frequency FH and the M range use frequency FMD are input. Here, similarly to the above, the previously selected shift map information is the reference value for this learning. And these I "1,
The input values of I "2 and I" 3 are normalized by the following equation and I "i '
And input to the middle layer consisting of three pieces.

Then, output values O "1, O" 2,
O "3 is obtained and the judgment output O" 4 is finally derived. The arithmetic processing in this third neural network is similar to the arithmetic processing in the first and second neural networks. The description is omitted here. However, here, the input value I "1 and the judgment output value O" 4 corresponding to the previously selected shift map information among the input values I "i are values between 0 and 1 as will be described later. Since it is taken, the minimum value I "1 of the input value I" 1 and the output value O "
The minimum value O "min of 4 is the value 0, while the maximum value I" 1max of the input value I "1 and the maximum value O" max of the output value O "4 are the value 1.

In this way, the learning regarding the downshift is carried out based on the use state of the M range during the downshift, and the judgment output O "4 for selecting the shift map according to the driver's preference is set. In other words, here, 5-4 downshift and 4-3
The output as to whether or not to select the shift-down restriction map corresponding to the down-shift restriction mode for restricting the down-shift is set. Here, the shift-down restriction map is provided for each of the economy mode, standard mode, and power mode selection maps # 1, # 2, and # 3, as described above.

FIG. 24 exemplifies a shift-down regulation map corresponding to the selection map # 1 in the economy mode as a representative, but as shown in the figure, the accelerator opening VA is small in the shift-down regulation map. In the case of No. 5, the shift timing of the 5-4 downshift and the 4-3 downshift of the fifth speed or lower shifts to the low vehicle speed side.
That is, the timing of the 5-4 downshift and the timing of the 4-3 downshift are the same as the timing of the 3-2 downshift. Therefore, in this shift-down regulation map, unless the vehicle speed V becomes extremely low, the fifth speed or the fourth speed is set.
The speed is maintained and the 5-4 downshift and the 4-3 downshift are not executed.

Then, in step S68, it is determined whether or not the shift down is possible in accordance with the determination output O "4. In the down shift enable mode, the normal economy mode, the standard mode selected in step S62,
The power mode selection maps # 1, # 2, and # 3 are applied as they are. On the other hand, in the case of the shift-down no mode,
Any one of the above shift down regulation maps provided for each mode is selected and applied. Specifically, the shift down permission / prohibition according to the determination output O "4 is set on the basis of a graph showing the relationship between the determination output O" 4 and the shift down permission / prohibition as shown in FIG. The map is selected appropriately. In other words, if the judgment output O "4 value is 0 or more and less than 0.5 (0≤O" 4 <0.5), the shift down restriction map is selected because the shift down is prohibited, while the judgment output O " If the four values are 0.5 or more and 1 or less (0.5 ≦ O ″ 4 ≦ 1), downshifting is possible, and the normal economy mode, standard mode, and power mode shift maps are selected. Become.

For example, when the driver does not use the exhaust brake and frequently performs a manual downshift in the M mode during deceleration, that is, when the M range use frequency FMD is large, The normal economy mode, the standard mode, and the power mode shift map selected by the shift-up learning, that is, the selection map # 1 as the shift-down possible mode are applied as they are. At this time, in the D range, 5-4 and 4-3 downshifts are sequentially executed as the vehicle decelerates, whereby the engine brake works properly.

On the other hand, at the time of deceleration, the driver dislikes downshifting and often uses the HOLD range.
When the HOLD range use frequency FH is high, that is, when the driver does not like the shock of downshifting or the like, the downshift restriction map as the downshift mode is selected. At this time, in the D range, the fifth speed or the fourth speed is maintained irrespective of the decrease in the vehicle speed V, so that discomfort such as shift shock is appropriately prevented.

By the way, when the determination result in step S64 is false and the vehicle speed V is equal to or higher than the predetermined vehicle speed V1 (for example, 30 km / h), the accelerator pedal 81 is released (OFF), and thereafter the engine speed Ne is increased. Under a situation where the engine speed Ne1 is reached and the clutch 15 does not become the automatic clutch disengagement, it can be determined that it is not necessary to carry out the desired learning of the downshift, and in this case, the routine is exited without doing anything. ..

As described above, the selection of the shift map is completed based on the learning according to the driver's preference. Next, returning to FIG. 15 and proceeding to step S22. Step S22 and subsequent steps show a routine for setting the optimum shift speed based on the operating state of the vehicle. In step S22, a target suitable for the traveling state is selected from the shift map selected as described above by the target shift stage setting means 3 on the basis of the vehicle speed V and the accelerator opening degree VA, which are factors that determine the operating state of the vehicle. Find the gear.

Then, in step S24, step S
10 Based on the vehicle load degree αVL obtained in step S18, the information from each sensor, and the target shift speed, the running state of the vehicle is estimated according to the above-mentioned fuzzy rule, and the driver's driving intention is inferred accordingly. I do. In the next step S30, the shift command supplied to the target shift speed is corrected based on the driver's intention estimated in step S24. In addition to the correction based on the fuzzy rule, this correction includes, for example, correction contents that restrict continuous execution of the shift immediately after the end of the shift or immediately after turning on and off the exhaust brake and the auxiliary brake. Description is omitted.

As described in detail above, in the shift control device for an automatic transmission for a vehicle according to the present invention, learning is performed based on the back propagation learning method using a neural network, so that each driver is subtly delicate. The optimum shift map according to the driver's preference is selected from a plurality of shift maps in consideration of different driver's tastes, and the target shift speed obtained by this shift map is further used as a desired fuzzy value with the vehicle load αVL as a parameter. The correction can be made according to a rule or the like, and the optimum shift speed can be appropriately determined by reflecting the driver's intention as well as the driver's preference. Therefore, in the shift control device for an automatic transmission for a vehicle, it is possible to perform shift control extremely close to shift-up and shift-down of a manual transmission mainly for manual operation, even though it is an automatic transmission,
The drivability can be greatly improved regardless of the driver.

In addition, among the learning data for performing the driver's favorite learning control, 2-3 accelerator opening average AVA
(23), 2-3 standard deviation SVA (23), 3-4 accelerator opening average AVA (34), 3-4 standard deviation SVA (34), 2-3 engine rotation average ANe (23), 3-4 Engine rotation average A
Regarding Ne (34), in calculating each data, when the vehicle speed V is less than a predetermined vehicle speed V2 (for example, 20 km / h), each learning value is not calculated until the vehicle speed V exceeds the predetermined vehicle speed V2. Therefore, the vehicle speed V is the predetermined vehicle speed V2
It is possible to prevent the data of upshifts, which are carried out regardless of the driver's preference, from being taken in as learning values when driving on congested roads that do not reach the limit, preventing unnecessary shift map switching when driving on congested roads. , The driver's favorite learning can be better implemented.

Of the learning data, regarding the HOLD range usage frequency FH during deceleration and the M range usage frequency FMD during deceleration, the driver performs a manual downshift even once when calculating these learning data. If the M-range usage frequency FMDi is not 0 due to the above fact, the driver judges that the HOLD range is not being used only to cancel the shift in the D range, and the HOLD range is used. The frequency FHi is set to a value of 0. On the other hand, if there is no manual downshift, the driver holds the HOL for the purpose of canceling the shift in the D range.
Since it is determined that the D range is being used and the M range usage frequency FMDi is set to a value of 0, the HOLD range usage frequencies FHi and M are always used when learning the downshift preference. The magnitude relationship of the range usage frequency FMDi is clarified. Therefore, the average value of each of them, H
The OLD range usage frequency FH and the M range usage frequency FMD are also set to clear values that reflect the driver's preference, which improves the learning accuracy and properly and reliably separates the shift down possible mode and the shift down non-mode. To be done.

In learning the above-mentioned downshift, when the vehicle speed V is equal to or higher than a predetermined vehicle speed V1 (for example, 30 km / h), the driver releases the accelerator pedal 81, the vehicle speed V is reduced, and the clutch 15 is automatically operated. When the clutch is disengaged, it is determined that the vehicle is in the deceleration state and learning regarding the downshift is performed. Therefore, it is accurately detected that the vehicle is in the deceleration state, and the desired learning during the downshift is in progress. It will be implemented reliably without interruption. In this case, the vehicle speed V is a predetermined vehicle speed V1 (for example,
The condition is 30 km / h) or more, but this is because if the vehicle speed V is a predetermined vehicle speed V1 (for example, 30 km / h) or more, downshifting is expected to be performed at least once.

Further, in the above-mentioned shift-down learning, the shift to be learned is the 5-4 downshift of the fifth speed or lower,
4-3 downshifts, 3-2 downshifts, etc., which are often made by other factors such as traffic conditions than the driver's preference, that is, larger than the 5th gear 6
Since learning is not performed for the -5 downshift and the 7-6 downshift, it is possible to prevent the CPU 95 from performing unnecessary calculation and avoid the shortage of the capacity of the memory 97, thereby reducing the load on the control unit 71. Therefore, it is possible to continuously carry out good learning.

In the above embodiment, the gear type transmission 17 capable of performing not only the automatic gear shift but also the manual gear shift is used to learn the driver's preference, but the present invention is not limited to this. Even if the learning is applied to a feasible automatic transmission, a sufficient effect can be obtained. Further, in the above embodiment, as described with reference to FIG. 17, the vehicle speed V indicates that the vehicle is traveling on a congested road.
(For example, it is determined whether or not the speed is less than 20 km / h), but the present invention is not limited to this. The vehicle may be provided with means for recognizing and grasping the road condition so that it can be determined that the vehicle is traveling on a congested road.

[0122]

As described above, according to the shift control device for a vehicle automatic transmission of the first aspect, the accelerator opening detecting means for detecting the accelerator opening of the engine and the engine for detecting the engine speed. Rotation speed detection means, monitor means for monitoring at least one of the accelerator opening degree and engine speed at the time of switching the shift stage of the automatic transmission, and learning data calculation for calculating learning data based on the output from the monitoring means. Means, learning means for performing learning to carry out a shift speed change that reflects the driver's shift preference based on the learning data, and shift control of the automatic transmission based on the output from the learning means and the operating state of the vehicle. And a congested road traveling determination means for determining that the vehicle is traveling on a congested road, and the learning data calculation means uses the congested road traveling determination means. If it is determined that both are traveling on a congested road, the output from the monitor means is not adopted and the learning data is not updated.Therefore, the accelerator opening at the time of switching the shift stage of the automatic transmission is performed. At least one of the engine speed and engine speed is monitored and output. Based on this output, learning data is calculated and learning is performed automatically so that the gear shift stage that reflects the driver's shift preference is implemented. While the vehicle is traveling on a congested road where the driver's shift preference is not likely to appear while the transmission gear shift control is being performed well, the learning data calculation means does not adopt the output from the monitor means, The learning data in consideration of the driver's shift preference can be prevented from being updated unnecessarily, and therefore, the accuracy of the shift control in which the driver's shift preference is reflected can be preferably maintained.

According to the shift control device for an automatic transmission for a vehicle of the second aspect of the invention, the shift mode is automatically changed according to the operating mode of the vehicle and the manual mode in which the shift stage is switched by the operation of the driver. In a shift control device for an automatic transmission for a vehicle having an automatic mode for switching gears, an accelerator opening detection means for detecting an accelerator opening of an engine, an engine rotation speed detection means for detecting an engine speed, and an automatic transmission. Learning data calculation for calculating learning data based on the output from the monitor means for monitoring at least one of the accelerator opening degree and the engine speed at the time of switching the gear shift stage of the machine and the monitor means in at least one of the manual mode and the automatic mode Based on the learning method and learning data, the gear shift is reflected in the auto mode in which the driver's shift preference is reflected. Learning means for carrying out learning, target shift speed setting means for setting a target shift speed based on the output from the learning means and the operating state of the vehicle, and shift control of the automatic transmission based on the target shift speed. Shift control means,
The vehicle is provided with a traffic jam running judgment means for judging that the vehicle is traveling on a traffic jam road, and the learning data calculation means, when the traffic jam road judging means judges that the vehicle is running on a traffic jam road. Since the output from the monitor means is not adopted and the learning data is not updated, in the shift control device of the automatic transmission for a vehicle having the manual mode and the automatic mode, the shift stage of the automatic transmission is switched. At least one of the accelerator opening degree and the engine speed at the time point is monitored, the learning data is calculated based on the monitor output in at least one of the manual mode and the automatic mode, and the driver's operation in the automatic mode is performed based on the learning data. Learning can be carried out so that the gear shift that reflects the gearshift preference is carried out. The shift control of the transmission can be satisfactorily performed. However, when the vehicle is traveling on a congested road in which the driver's shift preference does not easily appear, the learning data calculation unit does not adopt the output from the monitor unit, and the learning data considering the driver's shift preference is acquired. It can be prevented from being updated unnecessarily, and therefore, the accuracy of the shift control in which the driver's shift preference is reflected can be preferably maintained.

Further, according to the shift control device for an automatic transmission for a vehicle of claim 3, the learning data calculating means determines that the vehicle has departed from traveling on the congested road when the congested road traveling judging means judges that the vehicle has left the traveling on the congested road. Calculates the learning data based on the latest output monitored by the monitoring means while traveling on a congested road. The gearshift preference can be effectively reflected in the calculation of the learning data, and therefore, the gearshift control in which the driver's gearshift preference is reflected can be favorably executed.

According to another aspect of the present invention, there is further provided a vehicle speed detecting means for detecting a vehicle speed, and the congested road running determining means is configured so that the vehicle is congested when the vehicle speed is less than a predetermined vehicle speed. Since it is determined that the vehicle is traveling on a road, it can be easily determined that the vehicle is traveling on a congested road by determining whether the vehicle speed is lower than a predetermined vehicle speed. In addition, claim 5
According to the shift control device for an automatic transmission for a vehicle of above, the learning data calculation means is based on the latest output monitored by the monitoring means when the vehicle speed is equal to or higher than the predetermined vehicle speed and the vehicle speed is lower than the predetermined vehicle speed. Since the learning data is calculated,
It is possible to determine that the vehicle has exited from driving on a congested road because the vehicle speed has exceeded the prescribed speed, and the driver's shift preference when accelerating from a low vehicle speed such as a congested road to normal driving is effective for calculating learning data. Therefore, the shift control in which the driver's shift preference is reflected can be favorably performed.

Further, according to the shift control device for an automatic transmission for a vehicle of claim 6, since the shift stage is the shift up stage, it is particularly desirable to reflect the driver's shift preference. The accuracy of the up control can be preferably maintained.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main configuration of a shift control device for an automatic transmission for a vehicle according to an embodiment of the present invention.

FIG. 2 is a configuration diagram showing an overall configuration of a shift control device for an automatic transmission for a vehicle as an embodiment of the present invention.

FIG. 3 is a diagram showing a select pattern of a change lever.

FIG. 4 is a graph showing the relationship between accelerator opening and VA.

FIG. 5 is a graph showing shift map # 1 in economy mode.

FIG. 6 is a graph showing shift map # 2 in economy mode.

FIG. 7 is a graph showing shift map # 3 in economy mode.

FIG. 8 is a graph showing shift map # 1 in a normal mode.

FIG. 9 is a graph showing shift map # 2 in the normal mode.

FIG. 10 is a graph showing shift map # 3 in the normal mode.

FIG. 11 is a graph showing shift map # 1 in power mode.

FIG. 12 is a graph showing shift map # 2 in power mode.

FIG. 13 is a graph showing shift map # 3 in power mode.

FIG. 14 is a diagram showing a characteristic of a vehicle load degree as a shift control parameter on a flat road.

FIG. 15 is a flowchart showing a main routine of shift control.

16 is a flowchart showing a learning control routine in FIG.

FIG. 17: AVA, SVA, ANe of learning data
It is a flow chart which shows a calculation routine.

FIG. 18 is a flowchart showing an FH / FMD calculation routine of the learning data.

FIG. 19 is a diagram showing a first neural network.

FIG. 20 is a diagram showing a second neural network.

FIG. 21 is a diagram showing a third neural network.

FIG. 22 is a graph showing the relationship between judgment outputs O4 and O′4 and a selection map.

FIG. 23 is a graph showing the relationship between the judgment output O ″ 4 and whether downshifting is possible.

FIG. 24 is a graph showing a shift-down regulation map corresponding to shift map # 1 in economy mode.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Optimum shift stage determination means 2 Vehicle load degree calculation means 3 Target shift stage setting means 3A Storage means 3B Learning type selection means (learning means) 4 Engine torque calculation means 5 Driving force calculation means 6 Air resistance calculation means 7 Straight flat empty vehicle Equivalent acceleration calculating means 8 Subtracting means 11 Diesel engine (engine) 17 Gear transmission (transmission mechanism) 27 Engine rotation sensor (engine speed detection means) 61 Change lever 63 Gear shift stage selection switch 65 Gear shift unit 71 Control unit (shift control means) ) 79 vehicle speed sensor 81 accelerator pedal 85 accelerator opening sensor (accelerator opening detecting means) 87 brake sensor

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁶ Identification code Agency reference number FI Technical display F16H 59:66

Claims (6)

[Claims] 1. An accelerator opening detecting means for detecting an accelerator opening of an engine, an engine speed detecting means for detecting an engine speed, the accelerator opening and the engine at a time point of shifting a shift stage of an automatic transmission. Monitor means for monitoring at least one of the number of revolutions, learning data calculation means for calculating learning data based on the output from the monitor means, and a gear shift stage in which the driver's shift preference is reflected based on the learning data. Learning means for learning to perform switching, gear shift control means for performing gear shift control of the automatic transmission based on the output from the learning means and the operating state of the vehicle, and the vehicle traveling on a congested road. A traffic congestion road judging means for judging that the vehicle is traveling on the traffic jam road by the traffic jam road judging means. Been the case, the shift control apparatus for a vehicular automatic transmission, characterized in that does not update the learning data without employing the output from the monitoring means. 2. A shift control of an automatic transmission for a vehicle, which is connected to an engine and has a manual mode in which a shift stage is switched by a driver's operation and an automatic mode in which a shift stage is automatically switched according to a driving state of a vehicle. In the apparatus, an accelerator opening detection means for detecting an accelerator opening of the engine, an engine rotation speed detection means for detecting an engine rotation speed, the accelerator opening and the engine rotation speed at the time of shifting the shift stage of the automatic transmission. A monitoring means for monitoring at least one of the following: a learning data calculating means for calculating learning data based on an output from the monitoring means in at least one of the manual mode and the auto mode; and the auto mode based on the learning data. Sometimes learns to change gears that reflect the driver's shift preferences. Learning means, target shift speed setting means for setting a target shift speed based on the output from the learning means and the operating state of the vehicle, and shift control means for performing shift control of the automatic transmission based on the target shift speed. A traffic-congested road determination unit that determines that the vehicle is traveling on a congested road, and the learning data calculation unit determines that the vehicle is traveling on the congested road by the traffic-congested road determination unit. In this case, the output of the monitor means is not adopted and the learning data is not updated. 3. The monitor when the vehicle is traveling on the congested road when the learning data calculation means determines that the vehicle has exited from traveling on the congested road by the congested road traveling determination means. 3. The shift control device for an automatic transmission for a vehicle according to claim 1, wherein the learning data is calculated based on the latest output monitored by the means. 4. A vehicle speed detecting means for detecting a vehicle speed is further provided, and the congested road traveling judging means judges that the vehicle is traveling on the congested road when the vehicle speed is less than a predetermined vehicle speed. A shift control device for an automatic transmission for a vehicle according to claim 1 or 2. 5. The learning data calculation means calculates the learning data based on the latest output monitored by the monitor means when the vehicle speed is equal to or higher than the predetermined vehicle speed and the vehicle speed is lower than the predetermined vehicle speed. Characterized by doing,
A shift control device for an automatic transmission for a vehicle according to claim 4. 6. The shift control device for an automatic transmission for a vehicle according to claim 1, wherein the shift stage switching time is a shift up time.