JP3214697B2 - Gear ratio control device for automatic transmission for vehicle - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a transmission ratio control device for an automatic transmission for a vehicle provided with a neural network.

2. Description of the Related Art In a vehicle equipped with an automatic transmission such as an automatic transmission that automatically switches to a gear ratio corresponding to a plurality of gear stages and a continuously variable transmission that automatically changes the gear ratio in a stepless manner, Generally, a gear position or speed ratio of an automatic transmission is determined based on a throttle valve opening and a vehicle speed corresponding to an actual required output of an engine from a predetermined shift diagram, and the gear position or speed ratio is obtained. The automatic transmission is controlled as described above. However, in such a control device, since the shift characteristics are determined uniformly, the driving tendency of each driver cannot be reflected, and it cannot be said that an optimum speed ratio is obtained.

On the other hand, as described in JP-A-2-21058, for the purpose of improving the switching accuracy of the shift map, the running state of the vehicle is determined using an interconnected neural network, and There has been proposed a transmission ratio control device for an automatic transmission of a type in which one shift diagram is selected from a plurality of prepared shift diagrams, and a shift command is issued using the selected shift diagram.

Problems to be Solved by the Invention Incidentally, according to the above-described conventional gear ratio control device, switching is performed according to the determination by the neural network in addition to the shift diagram for determining the shift based on parameters such as the throttle opening and the vehicle speed. Since it is necessary to prepare another shift diagram, there is a disadvantage that the storage capacity increases by a power. Further, since a plurality of types of shift diagrams are switched, shift commands are essentially stepwise, and fine shift control cannot be performed. In addition, in the case of the interconnected neural network, when the same number of neuron elements are compared, the number of connections between them is generally increased, so that the calculation time tends to be longer.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a gear ratio control device for an automatic transmission for a vehicle that can perform fine gear ratio control without requiring a large storage capacity. Is to do.

Means for Solving the Problems The gist of the present invention for achieving the above object is to provide an automatic transmission for a vehicle that controls a speed ratio of an automatic transmission based on a plurality of types of traveling parameters representing a traveling state of the vehicle. (A) storage means for preliminarily storing a shift diagram for determining a gear ratio of the automatic transmission based on an engine required output and a vehicle speed among the travel parameters. A) a neural network that determines the gear ratio of the automatic transmission based on the traveling parameters and a parameter that represents a gear ratio determined from the gear shift diagram; and (c) the automatic network based on the determination of the neural network. Command value calculating means for outputting a speed ratio command value of the transmission.

In this manner, in the neural network, the gear ratio is determined based on the traveling parameters and the parameter representing the gear ratio determined from the gear shift diagram, and the command value calculating means determines the speed ratio in the neural network. By outputting the gear ratio command value of the automatic transmission based on the determination of the above, the gear ratio of the automatic transmission is controlled. Therefore, a fine shift command can be issued between the shift positions determined by the shift diagram without newly adding a shift diagram, so that a high-accuracy speed ratio control can be obtained without requiring a large storage capacity. is there.

Embodiment Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 2 is a block diagram showing a schematic configuration of the speed ratio control device of the present embodiment. In the figure, an automatic transmission 10 of a vehicle includes a plurality of planetary gear sets, a plurality of brakes and clutches, and switches to a plurality of gears by a combination of engagement states of the brakes and clutches. It is supposed to be. Further, the automatic transmission 10 is provided with a torque converter 14 having a lock-up clutch 12, and the power of the engine 16 is input via the torque converter 14. The lock-up clutch 12 and brakes and clutches for switching gears are controlled by a gear switching electromagnetic valve device 20 and a lock-up clutch switching electromagnetic valve device 22 included in the hydraulic control circuit 18. . Control 24 consists of CPU 24a, ROM 24b, RA
It is constituted by a so-called microcomputer including M24c and the like, and input signals supplied from the sensor group 26, that is, the throttle valve opening θa, the vehicle speed V, the road surface inclination angle θr
d, steering angle θs, vehicle acceleration G, current shift operation position S
, And based on them, the throttle valve opening degrees θa (0), θa
(1), θa (2)... Θa (Na) and vehicle speed V
(0), V (1), V (2)... V (Nv) are generated, the throttle valve opening θas at the time of the current shift operation position, and at the time of the current shift operation position. The vehicle speed Vs, the elapsed time ts after the shift operation position is reached, the current lock-up clutch engagement state L, the previous lock-up clutch engagement state L ⁻¹ , the current lock-up state The throttle valve opening θaL at the time, the vehicle speed VL at the time of the current lock-up state, and the elapsed time tL since the current lock-up state are determined. Then, the controller 24 determines a shift command value and a shift command value obtained based on the actual vehicle speed and the throttle opening from the shift diagram and the engagement diagram previously stored in the ROM 24b functioning as storage means. An algorithm constituted by a neural network is executed based on the combined command value, and a final shift command value and a final engagement command value output from the neural network are determined. And the lock-up clutch switching electromagnetic valve device 22 is controlled. Thereby, the optimum gear position (speed ratio) of the automatic transmission 10 is selected in relation to the running state of the vehicle, and the engagement state of the lock-up clutch 12 is switched. Note that the neural network may be configured on a dedicated semiconductor chip,
It may be configured as software.

FIG. 1 is a functional block diagram illustrating functions of the speed ratio control device of the present embodiment. In the drawing, the time-series parameter generating means 30 outputs time-series data relating to the throttle valve opening θa and the vehicle speed V among the parameters representing the running state of the vehicle, that is, the throttle valve opening θa (0),
θa (1), θa (2)... θa (Na) and vehicle speed V
(0), V (1), V (2)... V (Nv) are generated and supplied to the hierarchical neural network 32.
The vehicle running state parameter generation means 34 outputs a parameter indicating the running state of the vehicle, for example, the road surface inclination angle θ.
rd, steering angle θs, vehicle acceleration G, current shift operation position S, previous shift operation position S ⁻¹ , throttle valve opening θas at the time of the current shift operation position, and current shift operation position Vehicle speed Vs, elapsed time ts after the current shift operation position, current lock state L, current lock state L
L ⁻¹ , the throttle valve opening θaL at the time of the current lockup state, the vehicle speed VL at the time of the current lockup state, and the elapsed time tL since the current lockup state are generated. , Hierarchical neural network
Supplied to 32.

Further, the shift determining means 36 selects any predetermined gear position or gear ratio based on the actual throttle valve opening θa and the vehicle speed V from a shift diagram previously stored in the ROM 24b shown in FIG. 3, for example. Shift command value sf
One of t1, sft2, sft3, and sft4 is generated and supplied to the hierarchical neural network 32.
The lock-up clutch 12 is determined based on the actual throttle valve opening θa and the vehicle speed V based on the engagement diagram stored in the ROM 24b.
Release command value l _off for commanding engagement command value l _on or releasing commanding engagement is caused to occur, and is supplied to the hierarchical neural network 32.

As shown in FIG. 5, for example, the hierarchical neural network 32 includes an input layer, an intermediate layer, and an output layer composed of neuron elements corresponding to neurons, and a connection corresponding to a synapse between these layers. And a transmission element (corresponding to a nerve fiber) that transmits the state of the nerve cell element from the input layer to the output layer by being connected via the body. That is, the input layer includes a parameter representing the vehicle running state, that is, θa (0),
θa (1), θa (2)... θa (Na), V (0),
V (1), V (2)... V (Nv), θrd, θs, G,
S, S ⁻¹ , θas, Vs, ts, L, L ⁻¹ , θaL, VL, tL,
Shift command value sft1, sft2, sft3, sft4 and engagement command value
n neuron elements X _i (X _{1 to} X _n ) which are assigned corresponding to l _on and release command value l _off , respectively, and whose state is changed according to the contents of the parameters or command values.
It is composed of The middle layer is m neuron elements Y
_j (Y _{1 to} Y _m ). The output layer is composed of one neuron element Z _k (Z _{1 to} Z _L ). In the present embodiment, the number of the neuron elements _Zk is equal to the total number of the four types of shift commands for the gear position of the automatic transmission 10 and the two types of engagement commands and release commands for the lock-up clutch 12. Therefore, l = 6. Then, above the n-number of between neuronal elements X _i and the m neurons element Y _j, the coupling coefficient (weight) via a conjugate with a WX _ij those neurons element X _i and neurons Connect the element Y _j to the neuron element X _i
A transmission element DX _ij for transmitting the state of の to the neuron element Y _j is provided. Further, the m between the number of neurons element Y _j and l-number of neurons element Z _k, the coupling coefficient (weight) thereof neuronal elements through a coupling body having a WY _jk Y _j and nerve cells Connect the element Z _k and the state of the neuron element Y _j to the neuron element
A transmission element DY _jk for transmitting to Z _k is provided.

Returning to FIG. 1, in the final shift command value calculating means 40,
Among the neuronal elements Z _k in the output layer of the neural network 32, the final shift command signal is outputted to the gear switching solenoid valve device 20 with is calculated based on the state of the neuron elements Z ₁ to Z ₄ . Also, the final lock-up command value calculating unit 42, among the neuronal elements Z _k in the output layer of the neural network 32, the final lock-up command signal is calculated based on the state of the neuron elements Z ₅ and Z ₆ Is output to the lock-up clutch switching electromagnetic valve device 22.

Hereinafter, the operation of the shift control device for the automatic transmission according to the present embodiment shown in FIG. 2 will be described with reference to the flowchart in FIG.

In step S1, signals detected and output by the sensor group 26, that is, signals representing the throttle valve opening θa, the vehicle speed V, the road surface inclination angle θrd, the steering angle θs, the vehicle acceleration G, and the current shift operation position S are The time series data, that is, the throttle valve opening degrees θa (0), θa (1), θa (2) ·
..Θa (Na) and vehicle speeds V (0), V (1), V
(2)... V (Nv), throttle valve opening θas at the current shift operation position, vehicle speed Vs at the current shift operation position, after the current shift operation position Elapsed time ts, current lock-up clutch engagement state L, previous lock-up clutch engagement state L ⁻¹ ,
Throttle valve opening θaL at the time of the current lockup state, vehicle speed V at the time of the current lockup state
L, elapsed time tL since the current lockup state
Are determined respectively. In the present embodiment, step S1 corresponds to the time-series parameter generating means 30 and the vehicle running state parameter generating means 34.

In step S2, a shift command value sft1 for commanding the first speed gear based on the actual throttle valve opening θa and the vehicle speed V of the vehicle running state parameters is obtained from the previously stored shift diagram shown in FIG. One of the shift command value sft2 for commanding the second speed gear, the shift command value sft3 for commanding the third speed gear, and the shift command value sft4 for commanding the fourth speed gear is determined. For example, the shift point vehicle speed corresponding to the actual throttle valve opening θa is determined from the shift diagram, and when the actual vehicle speed V passes through the shift point vehicle speed, the command value is switched. In the present embodiment, this step S2 corresponds to the shift determining means 36.

In step S3, a command for engaging the lock-up clutch 12 based on the actual throttle valve opening θa and the vehicle speed V of the vehicle running state parameters is obtained from the previously stored engagement diagram shown in FIG. A combined command value l _on or a release command value l _off for instructing release is generated. For example,
The engagement point vehicle speed or the disengagement point vehicle speed corresponding to the actual throttle valve opening θa is determined from the engagement diagram, and when the actual vehicle speed V passes the engagement point vehicle speed or the disengagement point vehicle speed, the command value is switched. is there. In the present embodiment, this step S3 corresponds to the lock-up determining means 38.

In the following step S4, each state of the neuron element _Yj belonging to the intermediate layer of the neural network 32 is determined based on the equation (1) indicating the transmitted state and the transmitted state based on a predetermined function. Each is calculated according to the equation (2) that determines the state in the intermediate layer. In step S5, the states of neurons elements Z _k belonging to the output layer of the neural network 32, on the basis of a transmitted state indicating the equation (3), with the delivered state from a predetermined function , Respectively, according to the equation (4) that determines the state in the output layer. The functions f (x _j ) and f (y _k ) in the equations (2) and (4)
Are the four functions f ₁ , f ₂ , f ₃ ,
those that have been set in advance from among f _4.

Here, the functions f ₁ , f ₂ , f ₃ and f _{4 in} FIG. 7 are represented by the following equations.

As described above, the state of each neuron element Z _k belonging to the output layer of the neural network 32 is calculated, in step S6, the neuron elements Z ₁ to Z ₄ of each neuron element Z _k states , The final shift command value is determined. For example, the function f ₃ or by f ₄ is used state of neurons elements Z _k assumed to be represented by two values of -1 or +1, the state of the neuron elements Z ₁ is -1, nerve cells state elements Z ₂ -1, the state of the neuron elements Z ₃ +
1. If the state of the neuron element Z ₄ is −1, the third
A final shift command signal for commanding a high gear is output.
Further, when the by the function f ₁ or f ₂ is used neuronal elements Z _k state is assumed to be expressed by a continuous value between -1 and +1, the state of the neuron elements Z ₁ to Z ₄ Is output as a final shift command signal for commanding the gear corresponding to the gear having the largest value.

Any subsequent step S7, in the same manner as described above, the final lock-up command signal is outputted based on the state of the neuron elements Z ₅ and Z _6. As is apparent from this, step S7 corresponds to the final lock-up command value calculating means 42, and step S6 corresponds to the final shift command value calculating means 40.

As described above, in the gear ratio control device of the present embodiment, in the neural network 32, parameters representing the vehicle running state, that is, θa (0), θa (1), θa
(2)... Θa (Na), V (0), V (1), V
(2)... V (Nv), θrd, θs, G, S, S ⁻¹ , θa
Based on s, Vs, ts, L, L ⁻¹ , θaL, VL, tL, a parameter representing a gear ratio determined from the shift diagram in FIG. 3, and shift command values sft1, sft2, sft3, sft4. The shift is determined, and the final shift command value calculating means 40 outputs the gear ratio command value of the automatic transmission 10 to the gear changeover electromagnetic valve device 20 based on the determination of the neural network 32, so that the automatic transmission 10 Is controlled. Therefore, a fine shift command can be issued even between shift positions determined by the shift diagram without newly adding a shift diagram, so that a large amount of storage capacity of the control device is not required and high-precision speed ratio control is performed. Is obtained. The coupling coefficients (weights) in the neural network 32 and the functions shown in FIG. 7 are set in advance to values determined off-line so that the automatic transmission 10 and the lock-up clutch 12 are appropriately controlled. .

Further, according to the present embodiment, the hierarchical neural network 32 in which the state is transmitted from the input layer to the output layer is used. There is an advantage that the number is reduced and the calculation time is shortened.

Further, according to the present embodiment, the time series data θa (0), θa (1), θa (2)... Θa ( N
Since a), V (0), V (1), V (2),..., V (Nv) are used, the state of the vehicle can be accurately grasped, and control can be performed more in line with human senses. There are advantages.

Next, another embodiment of the present invention will be described. In the following description, the same parts as those in the above-described embodiment are denoted by the same reference numerals, and description thereof will be omitted.

In place of the neural network 32, a neural network 50 shown in FIG. 8 may be used. Neuronal elements X _n-5 belonging to the input layer of the neural network _{50, X n-4, X} n-3, X n-2, X n-1, nerve cell elements belonging to X _n and the output layer Z _k That Between Z ₁ , Z ₂ , Z ₃ , Z ₄ , Z ₅ , and Z ₆ , as shown by a two-dot chain line, they are connected via a coupling body having a coupling coefficient WZ _k to form an input layer. Neuron element X
transmission element DZ ₁ ~DZ ₆ to directly transmit the state of the _n-5 to X _n of the output layer to the neuron element Z ₁ to Z ₆ are provided. A number indicating the contents of the nerve cell elements Z _k is by the (3) the following equation (5) and Equation (4) and be used to replace the equation,
The state of the neuron elements Z _k is calculated.

In the present embodiment, as described above, the neuron element X
_{n-5, X n-4} , X n-3, X n-2, X n-1, X n and neuronal elements Z _1, Z
_{2, Z 3, Z 4,} Z 5, since transmission element DZ ₁ ~DZ ₆ between Z ₆ are provided, a shift instruction value obtained from the shift diagram and engagement diagram sft1, SFT2 , Sft3, sft4, the engagement command value l _on , and the release command value l _off can be given predetermined weights, so that there is an advantage that the values determined by the shift diagram and the engagement diagram are given some priority.

As mentioned above, although one Example of this invention was described based on drawing, this invention is applied also in another aspect.

For example, in the above-described embodiment, the lock-up clutch
Neural networks 32, 50 for 12 engagement controls
Of the final lock-up command value calculating means 42
And step S7 are provided, but they may be deleted.

The time series data representing the running state of the vehicle includes the steering angle θs in addition to the aforementioned throttle valve opening θa and vehicle speed V.
And vehicle acceleration may be used.

Further, the automatic transmission 10 described above is a stepped gear type in which the speed ratio is changed stepwise by being composed of a planetary gear device and a plurality of brakes and clutches. It may be a continuously variable transmission that can be used. In this case, a shift diagram for determining a target gear ratio or a target input shaft rotation speed from the throttle valve opening θa and the vehicle speed V is used, and the neural network for the transmission in the output layer in the neural networks 32 and 50 is used. The cell element is composed of one, and the nerve cell element is, for example, +1
A state represented by continuous values from to -1 is taken.

Further, in the above-described embodiment, the throttle valve opening degree θa is used as one of the parameters representing the vehicle running state.
The intake pipe negative pressure of 16 and the fuel injection amount of the engine 16 may be used. In short, it is sufficient that the amount corresponds to the required output of the engine 16.

Further, in the above-described embodiment, the hierarchical neural network 32 is used, but an interconnected neural network may be used.

The above is merely an example of the present invention, and the present invention can be variously modified without departing from the gist thereof.

[Brief description of the drawings]

FIG. 1 is a functional block diagram for explaining the functions of the embodiment of FIG. FIG. 2 is a block diagram illustrating a configuration of a gear ratio control device for an automatic transmission according to an embodiment of the present invention. FIG. 3 is a shift diagram used in the embodiment of FIG. FIG. 4 is an engagement diagram used in the embodiment of FIG. FIG. 5 is a diagram for explaining the configuration of the neural network of FIG. FIG.
2 is a flowchart illustrating the operation of the embodiment of FIG. FIG. 7 is a diagram illustrating functions used in the flowchart of FIG. FIG. 8 is a view corresponding to FIG. 5 for explaining another embodiment of the present invention. 10: Automatic transmission 24b: ROM (storage means) 32: Hierarchical neural network 40: Final shift command value calculation means (command value calculation means)

 ────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-2-21058 (JP, A) JP-A-2-138558 (JP, A) JP-A-1-116868 (JP, A) JP-A-2-210 165256 (JP, A) JP-A-1-14897 (JP, A)

Claims (1)

(57) [Claims] An automatic transmission gear ratio control device for controlling a gear ratio of an automatic transmission based on a plurality of types of traveling parameters representing a traveling state of a vehicle, wherein an engine request among the traveling parameters is provided. Storage means for storing in advance a shift diagram for determining the speed ratio of the automatic transmission based on the output and the vehicle speed; and based on the travel parameters and a parameter representing the speed ratio determined from the speed diagram. A neural network for determining a gear ratio of the automatic transmission; and command value calculating means for outputting a gear ratio command value for the automatic transmission based on the determination of the neural network. Gear ratio control device for automatic transmission.