JP2020041505A - Controller of internal combustion engine, learned model for controlling internal combustion engine, control method of internal combustion engine, control program of internal combustion engine, and control system of internal combustion engine - Google Patents

Abstract

To provide a controller of an internal combustion engine capable of highly accurately executing a determination as to whether to stop idling.SOLUTION: A controller of an internal combustion engine that controls operation of an internal combustion engine of a vehicle, comprises: a determination unit that when the value of an input parameter is input, using a first learned model using a neural network, outputs a determination as to whether to stop the internal combustion engine; and an engine control unit that controls the internal combustion engine based on the determination output from the determination unit. The input parameter of the determination unit includes at least one of an elapsed time since the last change of lubricant of the internal combustion engine and the total operating time of the internal combustion engine or the total mileage of the vehicle.SELECTED DRAWING: Figure 4

Description

  The present disclosure relates to a control device for an internal combustion engine, a learned model for controlling the internal combustion engine, a control method for the internal combustion engine, a control program for the internal combustion engine, and a control system for the internal combustion engine.

  2. Description of the Related Art Conventionally, an idling stop control for automatically stopping idling of an internal combustion engine of a vehicle and automatically restarting the internal combustion engine in accordance with the state of the vehicle has been known for the purpose of improving fuel efficiency of the vehicle and the like. . In the idling stop control, idling of the internal combustion engine is stopped when a predetermined stop condition is satisfied, such as when the vehicle is stopped due to a signal waiting, and when a predetermined restart condition such as release of the depression of a brake pedal is satisfied, the internal combustion engine is stopped. Control for restarting the engine is performed.

  In such an idling stop control technique, whether or not the internal combustion engine can be restarted is one of the conditions for stopping the idling of the internal combustion engine. For example, Patent Literature 1 discloses a technique for determining whether to stop idling based on the temperature of lubricating oil and the temperature of cooling water that affect the success or failure of restarting the internal combustion engine.

JP 2001-263123 A

  By the way, the success or failure of the restart of the internal combustion engine largely depends on the viscosity of the lubricating oil, and the viscosity of the lubricating oil changes due to various factors other than the temperature of the lubricating oil and the temperature of the cooling water. Therefore, in the related art as described in Patent Literature 1, it is not always possible to appropriately determine whether to stop idling, and even when idling is stopped based on the temperature of lubricating oil or the temperature of cooling water, A situation can occur where the internal combustion engine cannot be restarted. In order to avoid this, it is necessary to make the idling stop condition strict, but if the stop condition is made strict, the stop condition becomes difficult to be satisfied. If the stop condition is hardly satisfied, the number of times of idling stop cannot be sufficiently secured, and as a result, the improvement of fuel efficiency cannot be sufficiently realized.

  The present disclosure has been made in view of the above problems, and has as its object to provide a control device for an internal combustion engine capable of appropriately determining whether to stop idling, a trained model for controlling the internal combustion engine, an internal combustion engine, It is an object to provide an engine control method, an internal combustion engine control program, and an internal combustion engine control system.

  The gist of the present disclosure is as follows.

  (1) A control device for an internal combustion engine that controls the operation of an internal combustion engine of a vehicle. When a value of an input parameter is input, the control device controls the internal combustion engine using a first learned model using a neural network. A determination unit that outputs a determination as to whether or not to stop, and an engine control unit that controls the internal combustion engine based on the determination output from the determination unit, wherein an input parameter of the determination unit is the internal combustion engine A control device for an internal combustion engine, including at least one of an elapsed time since the last change of lubricant of the engine, a total operation time of the internal combustion engine, and a total travel distance of the vehicle.

  (2) The control device for an internal combustion engine according to (1), wherein the input parameter of the determination unit further includes a viscosity grade of a lubricating oil of the internal combustion engine.

  (3) An input unit for inputting a user input indicating a viscosity grade of the lubricating oil of the internal combustion engine is further provided, and the user input input to the input unit is input to the determination unit as a value of an input parameter of the determination unit. The control device for an internal combustion engine according to the above (2).

  (4) The input parameter of the determination unit further includes a remaining capacity of the battery of the vehicle, and the output parameter of the first learned model is any of the above (1) to (3), which is the determination. 2. The control device for an internal combustion engine according to claim 1.

  (5) The output parameter of the first learned model is a power consumption amount required for restarting the internal combustion engine, and the determination unit determines that the power consumption amount output from the first learned model is The control device for an internal combustion engine according to any one of (1) to (3), wherein the control device outputs the determination that the internal combustion engine is stopped when the internal combustion engine is equal to or less than a first threshold value.

  (6) The apparatus further includes a learning unit that updates the first learned model by learning weights of the neural network using teacher data including an actually measured value of power consumption used for restarting the internal combustion engine. The control device for an internal combustion engine according to (5).

  (7) The determining unit determines a difference between a power consumption output from the first learned model and an actually measured value of the power consumption used for restarting the internal combustion engine in a second predetermined value. The control device for an internal combustion engine according to the above (5) or (6), further comprising determining that the lubricating oil of the internal combustion engine has been replaced when the difference is equal to or greater than the threshold value.

  (8) a viscosity output unit that outputs a viscosity grade of a lubricating oil of an internal combustion engine using a second learned model using a neural network when a value of the input parameter is input; Further includes the viscosity grade, the input parameter of the second learned model includes an actually measured value of power consumption used for restarting the internal combustion engine, and the output parameter of the second learned model Contains the viscosity grade, and the determination unit receives the viscosity grade output from the viscosity output unit as the viscosity grade which is an input parameter of the determination unit. The control device for an internal combustion engine according to any one of the above items.

  (9) The determination unit is configured to determine a difference between a power consumption output from the first learned model and an actually measured power consumption used for restarting the internal combustion engine in a second predetermined value. If the value is equal to or greater than the threshold value, further, it is determined that the lubricating oil of the internal combustion engine has been replaced, and if the determination has been made that the replacement has been performed, the determination unit includes an input parameter of the determination unit. The control device for an internal combustion engine according to (8), wherein the viscosity grade output from the viscosity output unit is input as the viscosity grade.

  (10) The input parameter includes at least one of an elapsed time since the last change of the lubricating oil of the internal combustion engine of the vehicle and a total operation time of the internal combustion engine, and determines whether to stop the internal combustion engine. Using the neural network in which the weight is learned using the determination as an output parameter and using whether the internal combustion engine is actually restarted as teacher data or not, and causing the processor to output the determination, controlling the internal combustion engine. Trained model for.

  (11) The consumption required for restarting the internal combustion engine is included in the input parameter, which includes at least one of an elapsed time since the last change of the lubricating oil of the internal combustion engine of the vehicle and a total operation time of the internal combustion engine. The power consumption required for restarting the internal combustion engine using a neural network in which weights are learned using the measured value of the power consumption used for restarting the internal combustion engine as teacher data using the power amount as an output parameter. A trained model for controlling the internal combustion engine, causing the processor to function to output a.

  (12) A control method of an internal combustion engine for controlling operation of an internal combustion engine of a vehicle, wherein when a value of an input parameter is input, the internal combustion engine is controlled using a first learned model using a neural network. Outputting a determination of whether or not to stop the engine, and controlling the internal combustion engine based on the output determination, causing the processor to execute, the input parameter comprising: A method for controlling an internal combustion engine, comprising at least one of an elapsed time since the last replacement and a total operation time of the internal combustion engine.

  (13) A control program for an internal combustion engine for controlling the operation of the internal combustion engine of a vehicle, wherein, when a value of an input parameter is input, the internal combustion engine uses a first learned model using a neural network to execute the internal combustion engine. Outputting a determination of whether or not to stop the engine, and controlling the internal combustion engine based on the output determination, causing the processor to execute, the input parameter comprising: A control program for an internal combustion engine, including at least one of an elapsed time since the last replacement and a total operation time of the internal combustion engine.

  (14) When the value of the input parameter is input, a determination as to whether or not to stop the internal combustion engine is output using a learned model using a neural network, and the internal combustion engine is determined based on the output determination. A control system for an internal combustion engine, comprising: an electronic control unit, and a server configured to communicate with the electronic control unit, wherein the electronic control unit includes a value of the input parameter and a value of the learned parameter. Obtaining data including measured values of output parameters of the model, transmitting the data to a server, the server receiving the data, and obtaining the values of the input parameters and the measured values of the output parameters in the data Then, a data set indicating the relationship between the acquired input parameter value and the actually measured value of the output parameter is created, and the created data set is created. Therefore, the weight of the neural network in the trained model is learned using the measured value of the output parameter as teacher data, and the trained model that has performed the learning is transmitted to the electronic control unit. Receiving the trained model from which the learning has been performed, updating the weight of the neural network with the trained model having performed the learning, and input parameters of the electronic control unit, the last change of the lubricating oil of the internal combustion engine. A control system for an internal combustion engine, including at least one of an elapsed time from time and a total operation time of the internal combustion engine.

  According to the present disclosure, it is possible to appropriately determine whether to stop idling.

FIG. 1 is a schematic configuration diagram of an internal combustion engine and an ECU that controls the internal combustion engine according to the first embodiment. FIG. 2 is a diagram illustrating an example of a neural network. 3 (a) to 3 (e) show the viscosity of the lubricating oil, the elapsed time after replacing the lubricating oil, the total operating time of the internal combustion engine, the fuel amount and oil temperature in the lubricating oil, and the friction of the internal combustion engine. It is a figure each showing a relationship. FIG. 4 is an example of a neural network in the first trained model used in the first embodiment. FIG. 5 is a flowchart for explaining a control routine of idling stop control in the control device for the internal combustion engine according to the first embodiment. FIG. 6 is a flowchart for explaining a control routine of idling stop control in the control device for an internal combustion engine according to the second embodiment. FIG. 7 is a flowchart for explaining a control routine for determining whether or not the lubricating oil of the internal combustion engine has been changed in the control device for the internal combustion engine according to the third embodiment. FIG. 8 is a functional block diagram of a control unit according to the fourth embodiment. FIG. 9 is a schematic configuration diagram of a machine learning system according to the fifth embodiment.

  Hereinafter, embodiments will be described in detail with reference to the drawings. In the following description, similar components are denoted by the same reference numerals.

<First embodiment>
≫Configuration of internal combustion engine≫
FIG. 1 is a schematic configuration diagram of an internal combustion engine 100 and a control device (electronic control unit (ECU) 200) that controls the internal combustion engine 100 according to the first embodiment. As shown in FIG. 1, the internal combustion engine 100 includes an engine body 1, an intake device 20, and an exhaust device 30. In the present embodiment, the internal combustion engine 100 is a multi-cylinder internal combustion engine (only one cylinder is shown in FIG. 1).

  The engine body 1 includes a cylinder block 2 and a cylinder head 3 fixed to an upper surface of the cylinder block 2.

  A plurality of cylinders 4 are formed in the cylinder block 2. Inside each cylinder 4 is housed a piston 5 which reciprocates inside the cylinder 4 by receiving a combustion pressure. The space defined by the inner wall surface of the cylinder head 3, the inner wall surface of the cylinder 4, and the crown surface of the piston 5 is a combustion chamber 6.

  The cylinder head 3 has an intake port 7 and an exhaust port 8. The cylinder head 3 is provided with an intake valve 9, an exhaust valve 10, a fuel injection valve 11, and a spark plug 12.

  The piston 5 is connected to a crankshaft 14 via a connecting rod 13. The reciprocating motion of the piston 5 is converted into a rotational motion of the crankshaft 14.

  The crank angle sensor 14a generates an output pulse each time the crankshaft 14 rotates, for example, by 15 °, as a signal for calculating the engine speed or the like.

  The starter 15 is a starter motor that starts the internal combustion engine 100 by driving the crankshaft 14. The rotor shaft of the starter 15 is connected to the crankshaft 14 via a well-known power transmission mechanism (not shown) such as a belt mechanism so that power can be transmitted.

  The battery 16 supplies power to the starter 15 and the like. The battery 16 is charged by an alternator (not shown) that converts the power of the crankshaft 14 of the internal combustion engine 100 into electric energy when the internal combustion engine 100 is operating. The battery 16 is composed of, for example, a lead storage battery.

  The battery sensor 16a detects the voltage, current, temperature, internal resistance, capacity, charge acceptability, remaining capacity (or SOC: State of charge) of the battery 16, and the like. Water temperature sensor 17 detects the temperature of cooling water for cooling engine body 1. The oil temperature sensor 18 detects the temperature of the lubricating oil that lubricates the friction sliding portion of the engine body 1.

  The intake device 20 detects an intake pipe 21, a throttle valve 22, an air flow meter 23 that detects a flow rate of air flowing through the intake pipe 21 and drawn into the cylinder 4, and detects an intake air temperature passing through the intake pipe 21. And an intake air temperature sensor 24. The throttle valve 22 is opened and closed by a throttle actuator 22a, and its opening (throttle opening) is detected by a throttle sensor 22b.

  The exhaust device 30 detects an air-fuel ratio of exhaust gas passing through the exhaust pipe 31 and a catalyst 32 that is provided in the middle of the exhaust pipe 31 and includes a three-way catalyst, an oxidation catalyst, a NOx catalyst, and the like. An air-fuel ratio sensor 33.

  The ECU 200 includes a storage unit 210, a control unit 220, an input unit 230, and an output unit 240 which are connected to each other by the bi-directional bus 201.

  The storage unit 210 can include, for example, a ROM (Read Only Memory) and a RAM (Random Access Memory). The storage unit 210 stores various programs and learned models used for processing by the control unit 220, and various data (for example, various parameters, teacher data, various thresholds, and the like).

  The control unit 220 can be, for example, a processor having a CPU and its peripheral circuits. The control unit 220 can execute various controls of the vehicle by executing various programs stored in the storage unit 210.

  Further, as shown in FIG. 1, control unit 220 includes a determination unit 221, an engine control unit 222, and a learning unit 223. Each of these units included in the control unit 220 is a functional module implemented by a program executed on a processor included in the control unit 220.

  The input unit 230 includes a crank angle sensor 14a, a battery sensor 16a, a water temperature sensor 17, an oil temperature sensor 18, a throttle sensor 22b, an air flow meter 23, an intake air temperature sensor 24, an air-fuel ratio sensor 33, and an atmospheric pressure sensor for detecting atmospheric pressure. Output signals such as 252 are input via the corresponding AD converters 231. The input unit 230 receives, via a corresponding AD converter 231, an output voltage of a load sensor 251 a that generates an output voltage proportional to the amount of depression of an accelerator pedal 251 as a signal for detecting an engine load. Is done. The input unit 230 receives a signal related to a user input input to an arbitrary input device. The input unit 230 receives, as a signal for calculating an engine rotation speed and the like, an output signal of a crank angle sensor 14a that generates an output pulse every time the crankshaft 14 rotates, for example, by 15 °. As described above, the input signals of the various sensors necessary for controlling the internal combustion engine 100 are input to the input unit 230.

  The output unit 240 is electrically connected to each control component such as the fuel injection valve 11, the ignition plug 12, the starter 15, and the throttle actuator 22a via the corresponding drive circuit 241.

  The ECU 200 controls the internal combustion engine 100 by outputting a control signal for controlling each control component from the output unit 240 based on output signals of various sensors input to the input unit 230.

≪Idling stop control≫
In the present embodiment, the engine control unit 222 of the ECU 200 performs an idling stop control that automatically stops idling of the internal combustion engine 100 and then automatically restarts the idling when the power from the internal combustion engine 100 is not required. Execute. As described above, by stopping the internal combustion engine 100 when the power from the internal combustion engine 100 is unnecessary, it is possible to improve the fuel efficiency of the vehicle on which the internal combustion engine 100 is mounted.

  The motive power of the internal combustion engine 100 becomes unnecessary when, for example, the vehicle is stopped and the brake pedal is depressed. In such a state, it is not necessary to drive the vehicle, so the engine control unit 222 automatically stops idling of the internal combustion engine 100.

  On the other hand, when the depression of the brake pedal is released while the internal combustion engine 100 is stopped by the idling stop control, the vehicle needs to be driven. Therefore, in such a case, the engine control unit 222 automatically restarts the internal combustion engine 100 to drive the vehicle.

問題 Problems in idling stop control≫
By the way, in the idling stop control, in determining whether or not to stop idling, it is necessary to consider the remaining capacity of the battery in addition to the running state of the vehicle and the amount of depression of the brake pedal as described above. That is, when the internal combustion engine is restarted by the starter, it is necessary to supply relatively large electric power to the starter. Therefore, if the idling is stopped by the idling stop control even though the remaining capacity of the battery is small, the internal combustion engine cannot be restarted.

  In order to prevent the internal combustion engine from being unable to be restarted as described above, the amount of power consumption required for restarting the internal combustion engine before idling is stopped by idling stop control, and the predicted value is set to a predetermined value. It is conceivable to permit the stop of idling when the threshold value is not exceeded. However, if the prediction accuracy of the power consumption is poor, it is necessary to increase the safety margin at the above-described threshold. If the safety margin is made large in this way, it is difficult to stop idling, so that the number of times of idling stop cannot be sufficiently secured, and as a result, improvement in fuel efficiency cannot be sufficiently realized.

  Therefore, in the present embodiment, as described in detail below, by using a learned model using a neural network, it is possible to determine whether to stop idling based on the amount of power consumption required for restarting the internal combustion engine. It is possible to judge appropriately. Hereinafter, the present embodiment will be described in detail.

の Overview of Neural Network≫
First, a neural network used in the determination unit 221 will be described with reference to FIG. FIG. 2 shows an example of a neural network. The circles in FIG. 2 represent artificial neurons. In a neural network, this artificial neuron is usually referred to as a node or unit (referred to herein as a node). In FIG. 2, L = 1 indicates an input layer, L = 2 and L = 3 indicate a hidden layer, and L = 4 indicates an output layer. In FIG. 2, x ₁ and x ₂ indicate nodes in the input layer (L = 1) and output values from the nodes, and y indicates each node in the output layer (L = 4) and the output values from the nodes. Is shown. Similarly, z ₁ , z ₂ and z ₃ of the hidden layer (L = 2) indicate output values from each node of the hidden layer (L = 2), and z ₁ and z _{2 of the} hidden layer (L = 3). Indicates an output value from each node of the hidden layer (L = 3). The number of hidden layers can be one or any number, and the number of nodes in the input layer and the number of nodes in the hidden layer can also be any number. In this embodiment, the number of nodes in the output layer is one.

At each node of the input layer, the input is output as it is. On the other hand, the output values x ₁ and x ₂ of each node of the input layer are input to each node of the hidden layer (L = 2). At each node of the hidden layer (L = 2), the total input value u is calculated using the corresponding weight w and bias b. For example, the total input value u _k calculated at the node indicated by z _k (k = 1, 2, 3) of the hidden layer (L = 2) in FIG. Number of nodes).

Then, the total input value u _k is converted by activation function f, is output from the node indicated by z _k of the hidden layer (L = 2), as the output value _{z k (= f (u k} )) . On the other hand, the output values z ₁ , z ₂ and z ₃ of each node of the hidden layer (L = 2) are input to each node of the hidden layer (L = 3). At each node of the hidden layer (L = 3), the total input value u (Σz · w + b) is calculated using the corresponding weight w and bias b. The total input value u is similarly converted by the activation function f, and output as output values z ₁ and z ₂ from each node of the hidden layer (L = 3). In the present embodiment, a sigmoid function σ is used as the activation function.

On the other hand, the output values z ₁ and z ₂ of each node of the hidden layer (L = 3) are input to the nodes of the output layer (L = 4). At the output layer node, the total input value u (ｕz · w + b) is calculated using the corresponding weight w and the bias b, or the total input value u (Σz · w) is calculated. In the present embodiment, the identity function is used as the activation function at the output layer node. Therefore, the total input value u calculated at the output layer node is directly output from the output layer node as the output value y Is output as

≫Learning in neural networks≫
In the present embodiment, the value of each weight w and the value of the bias b in the neural network are learned using the backpropagation method. This backpropagation method is well known, and therefore, the outline of the backpropagation method will be briefly described below. Since the bias b is a kind of the weight w, the bias b is assumed to be one of the weights w in the following description.

Now, in a neural network as shown in FIG. 2, if the weight at the input value u ^(L) to the node of each layer of L = 2, L = 3 or L = 4 is represented by w ^(L) , the error function E The differentiation by the weight w ^(L) , that is, the gradient ∂E / ∂w ^(L) is expressed by the following equation (1).

Here, since z ^(L-1) · ∂w ^(L) = ∂u ^(L) , if (∂E / ∂u ^(L) ) = δ ^(L) , the above equation (1) becomes: It can be expressed by the following equation (2).

ここで、z^(L)＝ｆ（ｕ^(L)）と表すと、上記（３）式の右辺に現れる入力値ｕ_k ^(L+1)は、次の（４）式で表すことができる。

Here, if u ^(L) fluctuates, the error function E fluctuates through a change in the total input value u ^{(L + 1)} of the next layer, so δ ^(L) is expressed by the following equation (3). (K is the number of nodes in the L + 1 layer).

Here, assuming that z ^(L) = f (u ^(L) ), the input value u _k ^{(L + 1)} appearing on the right side of the above equation (3) can be expressed by the following equation (4). .

Here, the first term on the right side (∂E / ∂u ^{(L + 1)} ) of the above equation (3) is δ ^{(L + 1)} . Equation (3) of the second term on the right side ^{_{(∂u k (L + 1)}} / ∂u (L)) , from equation (4) can be expressed by the following equation (5).

Therefore, δ ^(L) can be expressed by the following equation (6) from the above equations (3) to (5).

That is, ^once δ ^{(L + 1)} is obtained, δ ^(L) can be obtained.

By the way, teacher data including a certain input value x and correct data t for the input value x has been obtained. If the output value from the output layer for this input value x is y, the square error becomes an error function. when being used, the square error E is calculated by ^{E = (y-t) 2} /2. At the node of the output layer (L = 4) shown in FIG. 2, the output value is y = f (u ^(L) ). In this case, δ ^{(L) at} the node of the output layer (L = 4 ⁾ Is given by the following equation (7).

By the way, in the present embodiment, as described above, since f (u ^(L) ) is an identity function, f ′ (u ^(Ll) ) = 1. Therefore, δ ^(L) = y−t, and δ ^(L) can be obtained.

Once δ ^(L) is obtained, δ ^(L-1) of the preceding layer can be obtained using the above equation (6). In this way, the δ of the previous layer is sequentially obtained, and using the values of δ, the differential of the error function E for each weight w, that is, the gradient ∂E / ∂w ^(L) Can be requested.

When the gradient ∂E / ∂w ^(L) is obtained, the value of the weight w is updated using the gradient ∂E / ∂w ^{(L) so} that the value of the error function E decreases. That is, learning of the value of the weight w is performed. When a batch or mini-batch is used as teacher data, a sum-of-squares error E expressed by the following equation (8) is used as the error function E. Here, N is the total number of teacher data, i is a natural number equal to or less than N (i = 1, 2,..., N), and y _i and t _i are the output value and the correct answer data for the input value x _ki , respectively. Show.

On the other hand, if the learning sequentially calculates the square error is carried out, as error function E, the square error E = (y-t) of the above ^2/2 is used.

≪Specific configuration example≫
Next, a specific configuration example of the control device for an internal combustion engine according to the first embodiment will be described. When the value of the input parameter is input, the determination unit 221 of the control device for the internal combustion engine according to the present embodiment determines whether to stop the internal combustion engine using the learned model using the neural network, that is, It outputs a determination of whether or not to stop idling. In particular, in the present embodiment, the power consumption required for restarting the internal combustion engine 100 is output as an output parameter from the learned model. The determination unit 221 determines whether or not to stop idling based on the power consumption output from the learned model.

  As described above, when the power consumption required for restarting the internal combustion engine 100 is output using the neural network, the power consumption required for restarting the internal combustion engine 100 is input to the neural network (that is, the input of the learned model). By using a large number of parameters that affect the amount of power consumption, the estimation accuracy of the power consumption can be improved.

Table 1 below shows an example of input parameters input to the neural network used in the determination unit 221 of the present embodiment. In the example shown in Table 1, the parameters from No. 1 to No. 8, that is, the elapsed time since the last change of the lubricating oil of the internal combustion engine 100 (hereinafter referred to as “elapsed time after lubricating oil change”) ), The total operation time of the internal combustion engine, the viscosity grade of the lubricating oil, the oil temperature, the amount of fuel in the lubricating oil, the cooling water temperature, the intake air temperature, and the atmospheric pressure are used as input parameters. No. 9 indicates an output parameter of the neural network. In the example shown in Table 1, the value y1 of the output parameter is the power consumption required for restarting the internal combustion engine 100. When diesel fuel is used, a cetane number may be further included as an input parameter.

  As shown in Table 1, the values of the input parameters No. 1 to No. 8 are indicated by x1 to x8, and the values of the output parameters are indicated by y1. In the following description, each input parameter may be represented by x1 to x8.

  The values x1 to x8 of these input parameters are all parameters that affect the power consumption required for restarting the internal combustion engine, and the reason will be briefly described below.

  First, among the input parameter values x1 to x8, the input parameter values x1 to x5 affect the friction of the internal combustion engine as described below with reference to FIGS. 3A to 3E. . Since the friction of the internal combustion engine affects the torque required for restarting the internal combustion engine, it also affects the amount of power consumed by the starter 15 to generate the torque. As a result, the input parameter values x1 to x5 affect the power consumption required for restarting the internal combustion engine.

  3 (a) to 3 (e) show the viscosity of the lubricating oil, the elapsed time after replacing the lubricating oil, the total operating time of the internal combustion engine, the fuel amount and oil temperature in the lubricating oil, and the friction of the internal combustion engine. The relationships are shown below.

  As shown in FIG. 3A, the magnitude of the friction of the internal combustion engine is proportional to the magnitude of the viscosity of the lubricating oil. Therefore, the viscosity grade of the lubricating oil affects the friction of the internal combustion engine.

  As shown in FIG. 3B, as the lubricating oil deteriorates with time, the friction of the internal combustion engine also increases. Further, as shown in FIG. 3B, when the lubricating oil is replaced, the lubricating oil of the internal combustion engine changes from a deteriorated state to a new state, so that the friction of the internal combustion engine sharply decreases. Thereafter, as the time elapses, the lubricating oil deteriorates and the friction of the internal combustion engine gradually increases. When the lubricating oil is replaced again, the friction of the internal combustion engine also sharply decreases. Therefore, the elapsed time after the replacement of the lubricating oil affects the friction of the internal combustion engine.

  As shown in FIG. 3C, the total operation time of the internal combustion engine affects the wear state of the internal combustion engine. Immediately after the start of use of the internal combustion engine, the internal combustion engine is in a state of large friction due to burrs and the like at the time of its manufacture, but the operation reduces the contact due to sliding of the burrs. Friction gradually decreases. Thereafter, as the operation time increases, for example, a groove (not shown in FIG. 1) formed inside the cylinder block 2 for retaining lubricating oil wears out, and an oil film is formed on the inner surface of the cylinder block 2. For example, the friction of the internal combustion engine increases due to the inability to maintain the pressure. Therefore, the operation time of the internal combustion engine affects the friction of the internal combustion engine.

  Further, as shown in FIG. 3 (d), when the lubricating oil is diluted with fuel and the amount of fuel in the lubricating oil increases, the viscosity also decreases, so that the amount of fuel in the lubricating oil is reduced by the friction of the internal combustion engine. Affect. Further, as shown in FIG. 3E, as the oil temperature increases, the viscosity of the lubricating oil decreases, so that the oil temperature affects the friction of the internal combustion engine.

  As described above, since the input parameter values x1 to x5 affect the friction of the internal combustion engine, they affect the power consumption required for restarting the internal combustion engine.

  Next, the reason why the input parameter values x6 to x8 among the input parameter values x1 to x8 affect the power consumption required for restarting the internal combustion engine will be described. The cooling water temperature x6 affects the cylinder wall temperature, the intake air temperature x7 affects the cylinder gas temperature, and the atmospheric pressure x8 affects the compression end temperature in the combustion chamber 6. Therefore, the values x6 to x8 of these input parameters all affect the temperature of the engine body, and thus affect the ignitability of the fuel. Note that the cetane number indicates the ignitability of diesel fuel and thus affects the ignitability of fuel.

  In order to restart the internal combustion engine, it is necessary to drive the starter 15 until a complete explosion occurs in the internal combustion engine. However, if the ignitability of the fuel in the combustion chamber 6 is poor, the starter 15 may be driven until a complete explosion occurs in the internal combustion engine. Since the time is longer, the amount of power consumed by the starter 15 from the stop of the internal combustion engine to the complete explosion of the internal combustion engine also increases. Therefore, the ignitability of the fuel in the combustion chamber 6 affects the power consumption required for restarting the internal combustion engine.

  Note that the input parameters to the neural network (that is, the input parameters to the determination unit 221) do not necessarily need to include all the parameters x1 to x8 described above, and may include only some of these parameters. good. However, even in such a case, the input parameters to the neural network include at least one of the elapsed time after the replacement of the lubricating oil of the internal combustion engine 100 and the total operation time of the internal combustion engine 100. By using input parameters that are not conventionally used while affecting the power consumption required for restarting the internal combustion engine, the power consumption required for restarting the internal combustion engine, which is an output parameter with high accuracy, It is possible to output the quantity. In addition, even in the case described above, since the influence on the power consumption required for restarting the internal combustion engine of the viscosity grade is large, the input parameter to the neural network may include the viscosity great. preferable.

  In addition, the values x1 to x8 of the input parameters are obtained by, for example, the obtaining methods shown in Table 1. That is, as shown in Table 1, the elapsed time x1 after the replacement of the lubricating oil is obtained from a calculated value calculated by counting the elapsed time since the replacement of the lubricating oil in the ECU 200. At the time of replacing the lubricating oil, which is the time to start counting the elapsed time after replacing the lubricating oil, for example, it can be determined by a user input to the input unit 230 via an arbitrary input device. The total operation time x2 of the internal combustion engine is obtained from a value calculated by counting the operation time of the internal combustion engine 100 in the ECU 200. The viscosity grade x3 of the lubricating oil is obtained from a user input to the input unit 230 via an arbitrary input device. The oil temperature x4 is obtained from the output value of the oil temperature sensor 18. The fuel amount x5 in the lubricating oil is obtained from a value calculated by the ECU 200 from the output value of the air-fuel ratio sensor 33 using a known method, for example. The cooling water temperature x6 is obtained from the output value of the water temperature sensor 17, the intake air temperature x7 is obtained from the output value of the intake air temperature sensor 24, and the atmospheric pressure x8 is obtained from the output value of the atmospheric pressure sensor 252. When a cetane number is used as an input parameter, the cetane number is obtained from a user input to the input unit 230 via an arbitrary input device or a value calculated by the ECU 200 using a known calculation method. Note that the method of acquiring the values of the parameters shown in Table 1 is merely an example, and the values of the parameters shown in Table 1 can also be acquired by other methods.

  FIG. 4 is an example of a neural network in the first learned model used in the determination unit 221 in the present embodiment. Hereinafter, the learned model used in the determination unit 221 is referred to as a first learned model. The neural network in the first learned model shown in FIG. 4 outputs the power consumption y1 required for restarting the internal combustion engine, which is the output value of the neural network, using the above-mentioned x1 to x8 as input parameters. . In this embodiment, the neural network is composed of P layers (P is an arbitrary integer of 3 or more), and the number of nodes in each hidden layer can be an arbitrary number. In the neural network in the first trained model shown in FIG. 4, the input layer (L = 1) has eight nodes corresponding to eight input parameters. Any number of nodes other than eight may be provided according to the number.

  Next, control performed by the ECU 200 in the present embodiment will be described. FIG. 5 is a flowchart illustrating a control routine for determining whether or not to stop idling in the control device for an internal combustion engine according to the first embodiment. The flow of this control routine is realized by the processor of the control unit 220 executing a program stored in the storage unit 210.

  First, in step S501, the determination unit 221 determines whether a predetermined idling stop condition is satisfied. The predetermined idling stop condition is satisfied, for example, when the vehicle speed is 0 km / h, the engine speed is less than 1200 rpm, and the depression amount of the brake pedal is a certain amount or more. However, the predetermined idling stop condition does not include a condition relating to a parameter input to the neural network in the present embodiment, such as a condition of a cooling water temperature or an atmospheric pressure.

  If it is determined in step S501 that the predetermined idling stop condition is not satisfied, the control routine proceeds to step S502. In step S502, the engine control unit 222 controls the internal combustion engine 100 so as to continue operation without stopping the internal combustion engine 100.

  If it is determined in step S501 that the predetermined idling stop condition is satisfied, the control routine proceeds to step S503. In step S <b> 503, the values of various input parameters are input to the determination unit 221, and the power consumption required for restarting the internal combustion engine 100 is output from the first learned model of the determination unit 221. The values of the various input parameters are obtained by the method shown in Table 1 described above.

  Next, in step S504, the determination unit 221 determines whether the power consumption output in step S503 is equal to or less than a first threshold. Here, the first threshold value is calculated based on the remaining capacity of the battery 16 output from the battery sensor 16a by the determination unit 221. The first threshold value is set to a value equal to or less than the remaining capacity of the battery 16 at that time, for example, in order to reliably restart the internal combustion engine 100.

  If it is determined in step S504 that the power consumption is not equal to or smaller than the first threshold, the control routine proceeds to step S505. In step S505, the determination unit 221 outputs a determination that the internal combustion engine 100 is not to be stopped. Thereafter, in step S502, the engine control unit 222 controls the internal combustion engine 100 so as to continue operation without stopping the internal combustion engine 100 based on the determination that the internal combustion engine 100 is not to be stopped.

  On the other hand, when it is determined in step S504 that the power consumption is equal to or less than the first threshold, in step S506, the determination unit 221 outputs a determination to stop the internal combustion engine 100. Thereafter, in step S507, the engine control unit 222 controls the internal combustion engine 100 to stop based on the determination to stop.

  Thus, the present control routine ends.

Here, the learning of the weight of the neural network used for the first learned model can be performed before shipping the vehicle. For example, before shipment of the vehicle, an external computer changes the input parameters x1 to x8 described above to various values, and repeats the restart of the internal combustion engine from the stop of idling, so that each time of each restart. N teacher data including the input parameter values and the correct answer data t _i of the output parameters at the time of restart for each input parameter value is created. Here, as described above, N is the total number of teacher data and i = 1, 2,..., N, and the correct answer data t in the present embodiment is the actual measurement of the power consumption used for restarting the internal combustion engine. Value. Then, as described above, the external computer sets the neural network so that the value of the square error E between the output parameter value y1 _i of the neural network and the corresponding correct data t _i decreases for each teacher data. The first learned model is created by learning the value of the weight w. In the vehicle ECU 200, the first learned model (weight and bias data) learned in this way is stored in the storage unit 210.

Further, the learning unit 223 uses the value y1 of each output parameter output during driving of the vehicle and the N pieces of teacher data including the correct answer data t _K actually obtained after the output, to calculate a square error. The first learned model may be updated in the storage unit 210 by calculating E and learning the value of the weight w of the neural network so that the value of E decreases. At this time, the actual measurement value (correct data) of the power consumption amount y1 actually used for restarting the internal combustion engine is obtained from the output value of the battery sensor 16a as shown in Table 1. If learning is not performed during driving of the vehicle, ECU 200 may not include learning unit 223.

  As described above, according to the present embodiment, it is possible to estimate the amount of power consumption required for restarting the internal combustion engine with high accuracy, and thus it is possible to execute the determination whether to stop idling with high accuracy. As a result, the idling stop condition can be relaxed, and the stop condition can be easily satisfied. Therefore, a sufficient number of times of idling stop can be secured, and therefore, an improvement in fuel efficiency can be realized.

  Note that instead of the total operating time of the internal combustion engine, which is an input parameter, another parameter (for example, the total traveling distance of the vehicle) that increases as the total operating time of the internal combustion engine increases may be used. The same applies to the following embodiments.

<Second embodiment>
Next, a control device for an internal combustion engine according to a second embodiment will be described. The configuration of the control device for the internal combustion engine according to the second embodiment is basically the same as the configuration of the control device for the internal combustion engine according to the first embodiment. In the following, a description will be given mainly of parts different from the configuration of the control device for the internal combustion engine according to the first embodiment.

  In the present embodiment, a determination as to whether or not to stop the internal combustion engine is output using a first learned model using a neural network. In the present embodiment, a sigmoid function is used as an activation function of the output layer of the neural network. In the present embodiment, the ECU 200 may not include the learning unit 223.

  Table 2 below shows an example of input parameters input to the neural network used in the determination unit 221 of the present embodiment. In the example shown in Table 2, in addition to the input parameters described in the first embodiment, the remaining capacity of the battery 16 is used as an input parameter. In addition, No. 10 indicates an output parameter, and in the example shown in Table 2, the value y2 of the output parameter is a numerical value related to determination of whether or not to stop the internal combustion engine. As described above, in the present embodiment, since the sigmoid function is used as the activation function of the output layer of the neural network, for example, the output parameter output from the first learned model is 0.5 or more (y2 ≧ 0. In the case of 5), the determination unit 221 outputs a determination to stop the internal combustion engine. Conversely, when the output parameter output from the first learned model is less than 0.5 (y2 <0.5), the determination unit 221 outputs a determination that the internal combustion engine is not stopped. Become.

  The remaining battery capacity x9 is obtained from the output value of the battery sensor 16a as shown in Table 2. The remaining capacity x9 of the battery may be obtained by another method. The values x1 to x8 of the other input parameters are obtained by the method described above in the first embodiment.

  The reason why the values x1 to x9 of the input parameters affect the determination y2 of whether or not to stop the internal combustion engine will be described. As described above, the values x1 to x8 of the input parameters affect the power consumption required for restarting the internal combustion engine. In addition, the determination of whether or not to stop the internal combustion engine is affected by the amount of power consumed and the remaining battery capacity x9 required to restart the internal combustion engine. Therefore, these input parameters x1 to x9 affect the determination y2 of whether or not to stop the internal combustion engine.

  Here, the input parameter includes at least one of an elapsed time after the replacement of the lubricating oil of the internal combustion engine 100 and a total operation time of the internal combustion engine 100. By using an input parameter that has not been conventionally used while affecting the determination of whether or not to stop the internal combustion engine, it is possible to output the determination of whether or not to stop the internal combustion engine, which is an output parameter, with high accuracy. .

  FIG. 6 is a flowchart for explaining a control routine of idling stop control in the control device for an internal combustion engine according to the second embodiment. The flow of this control routine is realized when the processor of the control unit 220 executes a program stored in the storage unit 210.

  First, in step S601, the determination unit 221 determines whether a predetermined idling stop condition is satisfied. If it is determined in step S601 that the predetermined idling stop condition is not satisfied, in step S602, the engine control unit 222 controls the internal combustion engine 100 so as to continue operation without stopping the internal combustion engine 100.

  If it is determined in step S601 that the predetermined idling stop condition is satisfied, the control routine proceeds to step S603. In step S603, the values of various input parameters are input to the determination unit 221 and a determination as to whether to stop the internal combustion engine 100 is output from the first learned model of the determination unit 221. If it is determined in step S603 that the internal combustion engine 100 is not to be stopped, the control routine proceeds to step S602, and the engine control unit 222 causes the internal combustion engine 100 to continue operating without stopping the internal combustion engine 100. Control.

  When a determination to stop the internal combustion engine 100 is output in step S602, the control routine proceeds to step S604. In step S604, the engine control unit 222 controls the internal combustion engine 100 to stop the internal combustion engine 100.

  Thus, the present control routine ends.

In the present embodiment, the learning of the weight of the neural network used for the first learned model can be performed before the vehicle is shipped. For example, before shipment of the vehicle, an external computer changes the values x1 to x9 of the input parameters described above to various values, and repeats the restart of the internal combustion engine from the stop of idling, so that at each restart, , And N pieces of teacher data including the correct answer data t _i for each input parameter value at each restart. Here, as described above, N is the total number of teacher data and i = 1, 2,..., N, and the correct answer data t in the present embodiment is whether or not the internal combustion engine was actually restarted. Is a label indicating In the present embodiment, a cross entropy error represented by the following equation (9) is used as an error function.

The external computer calculates the weight of the neural network such that the value of the cross entropy error E calculated from the value y2 _i of the output parameter of the neural network and the corresponding correct data t _i decreases for each teacher data. By learning the value of w, a first learned model is created. In the vehicle ECU 200, the first learned model (weight and bias data) learned in this way is stored in the storage unit 210.

  According to the present embodiment, the determination as to whether or not to stop the internal combustion engine is output using the first learned model using the neural network, so that the determination as to whether or not to stop idling can be performed with high accuracy. As a result, the idling stop condition can be relaxed, and the stop condition can be easily satisfied. Therefore, a sufficient number of times of idling stop can be secured, and therefore, an improvement in fuel efficiency can be realized.

<Third embodiment>
Next, a control device for an internal combustion engine according to a third embodiment will be described. The configuration of the control device for the internal combustion engine according to the third embodiment is basically the same as the configuration of the control device for the internal combustion engine according to the first embodiment. In the following, a description will be given mainly of parts different from the configuration of the control device for the internal combustion engine according to the first embodiment.

  In the present embodiment, as described below, the determination unit 221 determines whether or not the lubricating oil of the internal combustion engine has been changed based on the value of the output parameter of the first learned model according to the first embodiment. Can be determined. However, in the present embodiment, when the lubricating oil is exchanged when calculating the elapsed time after the lubricating oil exchange, which is an input parameter of the first learned model, a response to a user input is made as in the first embodiment. Instead, it is determined that the determination unit 221 previously determined that the lubricant of the internal combustion engine has been replaced.

  Here, when the lubricating oil is changed, the friction value of the internal combustion engine sharply decreases as described above with reference to FIG. 3B, so that the actually measured value of the power consumption used for restarting the internal combustion engine becomes sharp. To decrease. On the other hand, even when the lubricating oil has been changed, the elapsed time after the change of the lubricating oil, which is an input parameter, is not reset to 0 and continues counting unless it is determined that the lubricating oil of the internal combustion engine has been changed. For this reason, the first learned model does not reflect the reduction in friction of the internal combustion engine due to the replacement of the lubricating oil, and therefore the power consumption (hereinafter referred to as the power consumption required for restarting the internal combustion engine output from the first learned model). , "The predicted value of power consumption") hardly changes. As a result, immediately after the replacement of the lubricating oil, the difference between the predicted value and the measured value of the power consumption becomes large. In the present embodiment, this is used to determine whether or not the lubricating oil of the internal combustion engine has been replaced.

  FIG. 7 is a flowchart for explaining a control routine for determining whether or not the lubricating oil of the internal combustion engine has been changed in the control device for the internal combustion engine according to the third embodiment. The flow of this control routine is realized when the processor of the control unit 220 executes a program stored in the storage unit 210.

  In step S701, the determination unit 221 determines whether the difference between the measured value of the power consumption used for restarting the internal combustion engine and the predicted value of the power consumption is equal to or greater than a predetermined second threshold. judge. Here, the second threshold value is appropriately set, for example, so that there is no erroneous detection of replacement of the lubricating oil due to noise or the like.

  If it is determined in step S701 that the difference between the predicted value and the measured value of the power consumption is not greater than or equal to the second threshold, the control routine proceeds to step S702, and the determination unit 221 determines whether the internal combustion engine has been lubricated. Judge that the oil has not been changed. If it is determined in step S701 that the difference between the predicted value and the measured value of the power consumption is equal to or greater than the second threshold, the control routine proceeds to step S703, and the determination unit 221 determines whether the lubricating oil of the internal combustion engine The exchange is determined. Next, in step S704, the elapsed time after replacing the lubricating oil, which is an input parameter of the first learned model, is reset to 0, and counting is started from 0.

  Thus, the present control routine ends.

  Further, in the present embodiment, when it is determined that the lubricating oil of the internal combustion engine has been changed, the learning unit 223 preferably resets the elapsed time after changing the lubricating oil to 0 as described above. When the counting is started from 0, the first learned model is updated by learning the weight of the neural network using the teacher data including the actually measured value of the power consumption used for restarting the internal combustion engine. Good.

  According to the present embodiment, there is no need to input the time of lubrication oil replacement, which is an input parameter, which is the time to start counting the elapsed time after lubrication oil replacement, by user input, so that serviceability to the user can be improved.

<Fourth embodiment>
Next, a control device for an internal combustion engine according to a fourth embodiment will be described. The configuration of the control device for an internal combustion engine according to the fourth embodiment is basically the same as the configuration of the control device for an internal combustion engine according to the fourth embodiment. In the following, a description will be given mainly of parts different from the configuration of the control device for the internal combustion engine according to the first embodiment.

  FIG. 8 is a functional block diagram of a control unit according to the fourth embodiment. As illustrated in FIG. 8, the control unit 220 includes a determination unit 221, an engine control unit 222, a learning unit 223, and a viscosity output unit 224.

  In the present embodiment, when the value of the input parameter is input, the viscosity output unit 224 outputs the viscosity grade of the lubricating oil of the internal combustion engine using the second learned model using the neural network. Hereinafter, the learned model used in the viscosity output unit 224 is referred to as a second learned model.

  Table 3 below shows an example of input parameters input to the neural network used in the viscosity output unit 224. In the example shown in Table 3, m (m = 8) parameters from No. 1 to No. 8, that is, the elapsed time after replacing the lubricating oil, the total operating time of the internal combustion engine, the oil temperature, the lubricating oil The fuel amount, cooling water temperature, intake air temperature, atmospheric pressure, and power consumption used for restarting the internal combustion engine are used as input parameters. No. 9 indicates an output parameter, and in the example shown in Table 3, the value y3 of the output parameter is the viscosity grade of the lubricating oil of the internal combustion engine.

  The reason why the values x1, x2, x4 to x8, x10 of the input parameters affect the viscosity grade y3 of the lubricating oil will be described. As described above, the elapsed time x1, the oil temperature x4, and the fuel amount x5 in the lubricating oil after changing the lubricating oil affect the viscosity grade of the lubricating oil. Since the cooling water temperature x6, the intake air temperature x7, and the atmospheric pressure x8 affect the power consumption required for restarting the internal combustion engine as described above, as shown in FIG. Also affect. Further, as described above, since the total operation time x2 of the internal combustion engine and the power consumption x10 used for restarting the internal combustion engine affect the friction of the internal combustion engine, as shown in FIG. It also affects the viscosity grade of the lubricating oil. Therefore, the values x1, x2, x4 to x8, x10 of the input parameters affect the viscosity grade y3 of the lubricating oil.

  Here, the input parameters to the second learning model include at least the power consumption used for restarting the internal combustion engine. By using an input parameter that has not been conventionally used while affecting the viscosity grade of the lubricating oil, it becomes possible to output the viscosity grade of the lubricating oil of the internal combustion engine, which is an output parameter, with high accuracy.

  In the present embodiment, the determination unit 221 uses the viscosity output unit instead of the viscosity grade input by the user input as in the first embodiment, as the viscosity grade that is the input parameter of the first learned model. The viscosity grade output from 224 may be input. Further, the viscosity output unit 224 may be configured to calculate the viscosity grade when it is determined in step S703 shown in FIG. 7 that the lubricant has been changed in the third embodiment.

  According to the present embodiment, it is not necessary to input the viscosity grade of the lubricating oil, which is an input parameter, by a user input, so that the serviceability to the user can be improved.

<Fifth embodiment>
FIG. 9 is a schematic configuration diagram of an internal combustion engine control system according to the fifth embodiment. FIG. 9 shows a vehicle 310 and a server 320 installed outside the vehicle 310. As shown in FIG. 9, vehicle 310 includes a communication unit 311 and an ECU 312. The ECU 312 has the same configuration as the ECU 200 according to the first embodiment, except that the ECU 312 does not include the learning unit 223.

  When the values of various input parameters are input, the determination unit 221 of the ECU 312 outputs the values of the output parameters using a first learned model using a neural network. The determination unit 221 of the ECU 312 can use the input parameter and the output parameter used in the first embodiment or the second embodiment, respectively, as the input parameter and the output parameter.

  Also, the ECU 312 obtains the value of the input parameter and the measured value of the output parameter using various sensors, and transmits the data including the obtained value of the input parameter and the measured value of the output parameter via the communication unit 311. Send it to server 320.

  9, the server 320 includes a communication unit 321, a parameter value acquisition unit 322, a data set creation unit 323, and a learning unit 324. The communication unit 321 of the server 320 is configured to be able to communicate with the communication unit 311 of the vehicle 310.

  In the machine learning system according to the present embodiment, the ECU 312 of the vehicle 310 transmits data including the values of the input parameters and the actually measured values of the output parameters acquired using various sensors to the server 320 via the communication unit 311.

  The parameter value acquisition unit 322 of the server 320 receives the data transmitted from the vehicle 310 via the communication unit 321 and acquires the values of the input parameters and the actually measured values of the output parameters in the data. The data set creation unit 323 of the server 320 creates a data set indicating the relationship between the value of the input parameter acquired by the parameter value acquisition unit 322 and the corresponding measured value of the output parameter. The learning unit 324 learns the weight of the neural network in the first trained model using the measured values of the output parameters as the teacher data according to the data set created by the data set creating unit 323. Next, the learning unit 324 of the server 320 transmits the first learned model that has performed the learning to the vehicle 310 via the communication unit 321.

  The ECU 312 of the vehicle 310 receives the first learned model transmitted from the server 320 via the communication unit 311 and stores the received first learned model in the storage unit 210 of the ECU 312. The weight of the neural network in one trained model is updated. When the value of the input parameter is input, the determination unit 221 of the ECU 312 of the vehicle 310 outputs the value of the output parameter using the weight of the neural network in the updated first learned model, and outputs the value of the output parameter. And outputs a determination as to whether or not to stop the internal combustion engine based on. Engine control unit 222 of ECU 312 of vehicle 310 controls internal combustion engine 100 based on the determination output from determination unit 221.

  According to the present embodiment, the first learned model can be updated by the ECU 312 of the vehicle 310 receiving the first learned model in which the weight of the neural network has been learned in the server 320. This eliminates the need to provide the ECU 312 of the vehicle 310 with a high-performance arithmetic processing unit for learning the weights of the neural network, so that the ECU 312 is updated based on the acquired parameter values while reducing the cost of the vehicle. By using the first learned model, it is possible to realize control relating to determination of whether or not to stop idling.

DESCRIPTION OF SYMBOLS 1 Engine main body 2 Cylinder block 3 Cylinder head 4 Cylinder 5 Piston 6 Combustion chamber 7 Intake port 8 Exhaust port 9 Intake valve 10 Exhaust valve 11 Fuel injection valve 12 Spark plug 13 Connecting rod 14 Crankshaft 14a Crank angle sensor 15 Starter 16 Battery 16a Battery Sensor 17 Water temperature sensor 18 Oil temperature sensor 20 Intake device 21 Intake pipe 22 Throttle valve 22a Throttle actuator 22b Throttle sensor 23 Air flow meter 24 Intake temperature sensor 30 Exhaust device 31 Exhaust pipe 32 Catalyst 33 Air-fuel ratio sensor 100 Internal combustion engine 200 ECU
Reference Signs List 201 Bipolar bus 210 Storage unit 220 Control unit 221 Judgment unit 222 Engine control unit 223 Learning unit 224 Viscosity output unit 230 Input unit 231 AD converter 240 Output unit 241 Drive circuit 252 Atmospheric pressure sensor

Claims (14)

A control device for an internal combustion engine that controls operation of an internal combustion engine of a vehicle,
A determination unit configured to output a determination as to whether to stop the internal combustion engine by using a first learned model using a neural network when a value of the input parameter is input;
An engine control unit that controls the internal combustion engine based on the determination output from the determination unit,
The input parameter of the determination unit includes at least one of an elapsed time since the last change of the lubricating oil of the internal combustion engine and a total operation time of the internal combustion engine or a total mileage of the vehicle. Control device.   The control device for an internal combustion engine according to claim 1, wherein the input parameter of the determination unit further includes a viscosity grade of a lubricating oil of the internal combustion engine.   An input unit for inputting a user input indicating a viscosity grade of a lubricating oil of the internal combustion engine is further provided, wherein the user input input to the input unit is input to the determination unit as a value of an input parameter of the determination unit. Item 3. The control device for an internal combustion engine according to Item 2.   4. The internal combustion engine according to claim 1, wherein the input parameter of the determination unit further includes a remaining capacity of a battery of the vehicle, and an output parameter of the first learned model is the determination. 5. Engine control device. The output parameter of the first learned model is power consumption required for restarting the internal combustion engine,
4. The determination unit according to claim 1, wherein when the power consumption output from the first learned model is equal to or less than a first threshold, the determination unit outputs the determination to stop the internal combustion engine. 5. The control device for an internal combustion engine according to claim 1.   The learning unit that updates the first learned model by learning weights of a neural network using teacher data including an actually measured value of power consumption used for restarting the internal combustion engine. 6. The control device for an internal combustion engine according to claim 5.   The determination unit determines that a difference between the power consumption output from the first learned model and the actually measured power consumption used for restarting the internal combustion engine is equal to or greater than a predetermined second threshold. 7. The control device for an internal combustion engine according to claim 5, wherein in some cases, it is further determined that the lubricating oil of the internal combustion engine has been replaced. When a value of the input parameter is input, the apparatus further includes a viscosity output unit that outputs a viscosity grade of lubricating oil of the internal combustion engine using a second learned model using a neural network,
The input parameter further includes the viscosity grade,
The input parameter of the second learned model includes a measured value of the power consumption used for restarting the internal combustion engine,
Output parameters of the second trained model include the viscosity grade;
The internal combustion engine according to any one of claims 1 to 7, wherein the determination unit receives the viscosity grade output from the viscosity output unit as the viscosity grade, which is an input parameter of the determination unit. Control device. The determination unit determines that a difference between the power consumption output from the first learned model and the actually measured power consumption used for restarting the internal combustion engine is equal to or greater than a predetermined second threshold. In some cases, further, it is determined that the lubricating oil of the internal combustion engine has been replaced,
9. When the replacement is determined, the determination unit receives the viscosity grade output from the viscosity output unit as the viscosity grade, which is an input parameter of the determination unit. 3. The control device for an internal combustion engine according to claim 1.   The input parameter includes at least one of an elapsed time since the last change of the lubricating oil of the internal combustion engine of the vehicle and a total operating time of the internal combustion engine, and outputs a determination whether to stop the internal combustion engine. Using a neural network in which the weight is learned as a parameter and whether or not the internal combustion engine has actually restarted as teacher data, causing the processor to output the determination, and learning for controlling the internal combustion engine. Model.   The input parameter includes at least one of an elapsed time since the last change of the lubricating oil of the internal combustion engine of the vehicle and the total operation time of the internal combustion engine, and an amount of power consumption required for restarting the internal combustion engine. Outputting the power consumption required for restarting the internal combustion engine by using a neural network whose weight is learned using the measured value of the power consumption used for restarting the internal combustion engine as teacher data as an output parameter. A trained model for controlling the internal combustion engine, so that the processor functions. A control method of an internal combustion engine for controlling operation of an internal combustion engine of a vehicle,
When the value of the input parameter is input, outputting a determination as to whether to stop the internal combustion engine using a first learned model using a neural network,
Controlling the internal combustion engine based on the output determination, and causing the processor to execute,
The control method for an internal combustion engine, wherein the input parameter includes at least one of an elapsed time since a last change of lubricating oil of the internal combustion engine and a total operation time of the internal combustion engine. A control program for an internal combustion engine for controlling operation of the internal combustion engine of the vehicle,
When the value of the input parameter is input, outputting a determination as to whether to stop the internal combustion engine using a first learned model using a neural network,
Controlling the internal combustion engine based on the output determination, and causing the processor to execute,
A control program for an internal combustion engine, wherein the input parameter includes at least one of an elapsed time since a last change of lubricating oil of the internal combustion engine and a total operation time of the internal combustion engine. When the value of the input parameter is input, a determination as to whether or not to stop the internal combustion engine is output using a learned model using a neural network, and the internal combustion engine is controlled based on the output determination. , An electronic control unit,
A server configured to be able to communicate with the electronic control unit;
A control system for an internal combustion engine, comprising:
The electronic control unit obtains data including a value of the input parameter and an actually measured value of an output parameter of the learned model, and transmits the data to a server.
The server receives the data, acquires the value of the input parameter and the measured value of the output parameter in the data, and sets a data set indicating the relationship between the acquired value of the input parameter and the measured value of the output parameter. Creates, learns the weight of the neural network in the trained model using the measured values of the output parameters as teacher data according to the created data set, and transmits the learned trained model to the electronic control unit. ,
The electronic control unit receives the learned model that has performed the learning from the server, updates the weight of the neural network with the learned model that has performed the learning,
The control system for an internal combustion engine, wherein the input parameters of the electronic control unit include at least one of an elapsed time since a last change of lubricating oil of the internal combustion engine and a total operation time of the internal combustion engine.

Images