Operation condition calculation system, control device for internal combustion engine, and adaptation device for internal combustion engine

Abstract

An operation condition calculation system is provided with: a first learning device that uses, as an input, an operation condition of an in-vehicle device mounted on a vehicle, and uses, as an output, a physical state generated by the in-vehicle device operating on the basis of the operation condition; a second learner that outputs a performance index using the output of the first learner as an input; a gradient calculation unit that calculates a gradient between the first learner and the second learner; and an operation changing unit that changes the operation condition using the gradient.

Description

Technical Field

[0001] The present invention relates to an operating condition calculation system, a control device for an internal combustion engine, and an adaptation device for an internal combustion engine. Background Art

[0002] In recent years, in the development of environmentally friendly engines, reducing the emission of PN (Particulate Number) or THC (Total Hydrocarbon) generated when the engine is started or running at low temperatures has become a problem. These are particularly common when the external atmosphere is low temperature or when the engine is started, and it is required to suppress them. In addition, it is necessary to predict their occurrence in advance through real-time control and adjust the operating conditions. In Patent Document 1, a method for monitoring an engine using a cascaded neural network including a plurality of neural networks is disclosed, the method comprising the following steps: storing data corresponding to the cascaded neural network in a memory; inputting signals generated by a plurality of engine sensors into the cascaded neural network; updating a second neural network at a first speed together with the output of the first neural network, the output being based on the input signal; and outputting at least one engine control signal from the second neural network at a second speed, wherein the second speed is faster than the first speed. Prior art literature Patent Literature

[0003] Patent Document 1: Japanese Patent Application Laid-Open No. 2003-328837 Summary of the Invention Problems to be solved by the invention

[0004] In the invention described in Patent Document 1, there is room for improvement in the setting of operating conditions. Technical means to solve the problem

[0005] The operating condition calculation system of the first embodiment of the present invention comprises: a first learner that takes the operating conditions of an on-board device mounted on a vehicle as input and outputs a physical state resulting from the operation of the on-board device based on the operating conditions; a second learner that takes the output of the first learner as input and outputs a performance indicator; a gradient calculation unit that calculates the gradients of the first learner and the second learner; and an operation change unit that uses the gradients to change the operating conditions. A control device for an internal combustion engine according to a second aspect of the present invention includes the above-mentioned operating condition calculation system. An adaptation device for an internal combustion engine according to a third aspect of the present invention includes the above-mentioned operating condition calculation system. Effects of the Invention

[0006] According to the present invention, optimal operating conditions can be calculated from the viewpoint of performance indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 is a schematic diagram of an internal combustion engine. Figure 2 is a diagram showing an example of a failure of an internal combustion engine. Figure 3 is a functional configuration diagram of the arithmetic device of the first embodiment. Figure 4 is a hardware configuration diagram of the arithmetic device. Figure 5 is a flowchart showing the processing of the arithmetic device. Figure 6 is a functional configuration diagram of the arithmetic device of the modified example 1. Figure 7 is a functional configuration diagram of the arithmetic device of the second embodiment. Figure 8 is a functional configuration diagram of the arithmetic device of the third embodiment. Figure 9 is a functional configuration diagram of the arithmetic device of the fourth embodiment. Figure 10 is a functional configuration diagram of the arithmetic device of the fifth embodiment. Figure 11 is a functional configuration diagram of the arithmetic device of the sixth embodiment. DETAILED DESCRIPTION

[0008] First Embodiment Hereinafter, with reference to Figures 1-5 The first embodiment of the operation condition arithmetic system will be described.

[0009] Figure 1 is a schematic diagram of an internal combustion engine 1 that is an arithmetic target of the operation condition arithmetic system. The internal combustion engine 1 mounted on a vehicle 9 includes a piston 101, a cylinder 102, and a cylinder head 104. The cylinder head 104 forms a combustion chamber 103 in cooperation with a piston top surface 101P of the piston 101 and a combustion chamber inner wall 109 of the cylinder 102. In addition, the internal combustion engine 1 has at least one of a direct injection injector 111 and an intake port injection injector 112 as an injector for fuel injection.

[0010] A spark plug 105 is disposed directly above the combustion chamber 103, and has an electrode 106 that ignites the mixture. During the intake stroke, air flows into the combustion chamber 103 through an intake port 107 opened by an intake valve 113 that communicates with the main combustion chamber. Fuel is supplied to the combustion chamber 103 in the form of a mist spray from the direct injection injector 111 or the intake port injection injector 112, and vaporizes to become a mixture. Then, the mixture is compressed by the piston 101, and at an appropriate timing, a main ignition signal is sent to the ignition coil 110, and the mixture in the combustion chamber 103 is ignited by the spark plug 105, causing the mixture in the combustion chamber 103 to burn. As a result, the pressure in the combustion chamber 103 rises, pushing the piston 101, which rotates the crankshaft 115 via the connecting rod 113A to generate power.

[0011] Figure 2 FIG. 1 is a diagram showing an example of a malfunction of the internal combustion engine 1. When liquid fuel is injected from the injector, the fuel sometimes adheres to the opposing wall in the liquid state. For example, as shown in FIG. 2, when the spray 201 is injected into the combustion chamber 103 by the intake port injection injector 112, a fuel liquid film 202 is formed on the piston top surface 101P and the combustion chamber inner wall 109. In addition, since the spray 201 is high speed, when it collides with the piston top surface 101P or the combustion chamber inner wall 109, etc., the impact can cause liquid droplets 203 to splash into the combustion chamber 103. In addition, sometimes a tip wetting 204 occurs in which fuel droplets remain and adhere around the spray port of the direct injection injector 111. Figure 2

[0012] When the spark plug 105 is used to ignite the mixture in the combustion chamber 103, if the fuel liquid film 202 or the liquid droplets 203 remain, they burn in the liquid state. At this time, the fuel liquid film 202, the liquid droplets 203, and the tip wetting 204 that remain at the moment the flame reaches do not contribute to the engine output, but rather, diffuse combustion occurs. As a result, PN (Particulate Number) and THC (Total Hydro Carbon) are generated. A similar problem occurs when the intake port injection injector 112 is used. Fuel that does not contribute to the engine output, such as the fuel liquid film 202, the liquid droplets 203, and the tip wetting 204, is referred to as unburned fuel 205. In addition, if the fuel concentration of the mixture in the combustion chamber 103 decreases due to the generation of the unburned fuel 205, and the fuel present around the spark plug 105 becomes insufficient, when ignition occurs by the spark plug 105, problems such as misfire in which the flame does not form sufficiently, or unstable combustion, etc. can occur.

[0013] ​In addition, under cold engine conditions, such as starting at low temperatures, not only is unburned fuel 205 easily generated, but the following problems also exist. That is, since the temperature inside the combustion chamber 103 is mostly low, the number of misfires increases, more PN and THC are generated, and combustion stability may deteriorate. Furthermore, during the operation of the internal combustion engine, the internal state of the combustion chamber 103, headed by the amount of unburned fuel 205, is complexly related to multiple factors such as external factors such as the flow in the cylinder, the outside air temperature, the cooling water temperature, the fuel temperature, the engine speed, the fuel injection timing, the number of fuel injection distributions, and the ignition timing and other control variables. In addition, since the amount of PN and THC generated is determined by complex physical phenomena such as combustion and chemical reactions, it is not easy to predict the amount of PN and THC generated based on operating conditions.

[0014] Several techniques have been proposed to predict the final exhaust gas composition based on operating conditions, using machine learning techniques, such as neural networks. However, as mentioned above, these techniques rely on a plethora of operating conditions, and conditions unrelated to the final output are also input, resulting in noise and other factors, which can reduce prediction accuracy. Therefore, in this embodiment, the composition of the exhaust gas from the internal combustion engine 1 and the combustion performance within the internal combustion engine 1 are largely dependent on the state of the air-fuel mixture ignited by the spark plug 105 within the combustion chamber 103. As described below, optimal operating conditions are calculated from the perspective of performance indicators.

[0015] Figure 3 This is a functional structure diagram of the computing device 800. The computing device 800 is a general-purpose computing machine or electronic control unit. The computing device 800 may be mounted on the vehicle 9 having the internal combustion engine 1, or may not be mounted on the vehicle 9. When the computing device 800 is mounted on the vehicle 9, the calculation results of the computing device 800 can be used to control the internal combustion engine 1. In this case, the computing device 800 can be called the "control device" of the internal combustion engine 1. When the computing device 800 is not mounted on the vehicle 9, for example, during the design stage of the internal combustion engine 1, the optimal control conditions corresponding to the design specifications of the internal combustion engine 1 are determined. In this case, the computing device 800 is also called the "adaptation device" of the internal combustion engine 1. The computing device 800 includes a first learner 302, a second learner 306, a first gradient calculation unit 307, a second gradient calculation unit 308, and an operation change unit 309.

[0016] The first learner 302 is a learner that has been previously learned. The input of the first learner 302 is the operating condition 301 of the internal combustion engine 1, and the output of the first learner 302 is a physical quantity 303 representing the internal state of the ignition timing of the combustion chamber 103. The second learner 306 is a learner that has been previously learned. The input of the second learner 306 is the physical quantity 303 representing the internal state of the ignition timing of the combustion chamber 103, and the output of the second learner 306 is a performance indicator 305. However, although Figure 3 Although not shown, part or all of operating conditions 301 may be input to second learner 306. The data used in learning by first learner 302 and second learner 306 may be data obtained through numerical analysis or actual measurement values. Operating conditions 301 are data actually acquired from sensors mounted on vehicle 9 or data that can be acquired from sensors mounted on vehicle 9.

[0017] The operating conditions 301 include a plurality of specific conditions as detailed operating conditions. The specific conditions include intake air temperature, cooling water temperature, load, number of fuel injections, rotational speed, injection amount in each injection, injection start time, injection time, fuel pressure, ignition timing, etc. More specific conditions may include the number of distributions, injection time, injection time, and injection distribution ratio when the fuel is distributed. In addition, the specific conditions include: conditions that cannot be easily changed by the internal combustion engine 1, such as the outside air temperature (hereinafter also referred to as "external factors"), and conditions related to the control of the internal combustion engine 1 (hereinafter also referred to as "controlled variables"). External factors include outside air temperature, cooling water temperature, and load. Controlled variables include the number of fuel injections, injection amount in each injection, injection start time, injection time, fuel pressure, ignition timing, etc. The operating conditions 301 are conceived to be various combinations, for example, rotational speed, load, intake air pressure, intake air temperature, water temperature, fuel pressure, number of fuel injections, fuel injection timing, distribution ratio, and ignition timing may be included as essential components.

[0018] Physical quantity 303 includes multiple specific physical quantities. Examples of these specific physical quantities include the amount of the mixture, the temperature within the combustion chamber at the time of ignition, and the pressure within the combustion chamber at the time of ignition. Physical quantity 303 can be composed of various combinations, but may also include the in-cylinder temperature, in-cylinder pressure, amount of fuel deposited, and amount of overly rich mixture at the time of ignition as essential components.

[0019] The performance index 305 includes one or more specific indexes, i.e., specific indexes. The specific indexes are, for example, the amount of PN generation, the amount of THC generation, the combustion variation rate, and the like. In addition, among the specific indexes, indexes indicating that some kind of abnormality has occurred in the internal combustion engine, such as the exhaust gas temperature, the catalyst temperature, and the like, can also be included. In addition, the generation of PN is strongly affected by at least the amount of unburned fuel 205 or the fuel-rich mixture, which is generated in the combustion chamber, represented by fuel adhesion or fuel droplet floating, the temperature and the pressure in the combustion chamber at the time of ignition. THC is strongly affected by the unburned fuel or the temperature of the combustion chamber 103, and the combustion variation rate is affected by the concentration of fuel near the electrode 106 of the spark plug 105, the pressure and the temperature of the combustion chamber 103, and the uniformity of the mixture in the combustion chamber 103.

[0020] The neural network constituting the first learner 302 has the same number of nodes as the specific conditions in the input layer and the same number of nodes as the specific physical quantities in the output layer. The neural network constituting the second learner 306 has the same number of nodes as the specific physical quantities in the input layer and the same number of nodes as the specific indexes in the output layer.

[0021] In addition, for example, a single learner that takes the operating conditions 301 as input and outputs the performance indexes 305 can also be made, but by combining the first learner 302 and the second learner 306, the size of each learner can be reduced. In addition, the number of operations in the learner required for the estimation of the performance indexes 305 can be reduced, and an improvement in estimation accuracy can be achieved.

[0022] The first gradient operation section 307 operates the relationship between the input and the output of the first learner 302 as the first gradient 321. Specifically, the first gradient operation section 307 outputs the control variable having the largest gradient with respect to each specific physical quantity that is the output of the first learner 302 and the gradient value thereof. The second gradient operation section 308 operates the relationship between the input and the output of the second learner 306 as the second gradient 322. More specifically, the second gradient operation section 308 outputs the specific physical quantity having the largest gradient with respect to each performance index that is the output of the second learner 306 and the gradient value thereof. Thus, the operating change section 309 can acquire the degree of influence of the physical quantities 303 on each performance index 305 and the degree of influence of the operating conditions 301 on each physical quantity 303 in the form of the gradient value.

[0023] The operation change section 309 outputs a new operation condition 310 as an input of the operation condition 301, the first gradient 321, and the second gradient 322. However, the new operation condition 310 is distinguished from the initially read operation condition 301 only in the name and the symbol, and the first learner 302 reads the new operation condition 310 in the same manner as the operation condition 301. In the operation of the new operation condition 310, the operation change section 309 uses, for example, the gradient vector obtained from the first gradient operation section 307, and adopts an algorithm such as the Newton method.

[0024] The operation change section 309 can specify an index to be improved in advance or can set a priority order in a case where the performance index 305 includes a plurality of indexes. The operation change section 309 operates the new operation condition 310 in which the control variable is changed in order to improve the performance index 305. For example, the new operation condition 310 increases the injection time from the current "100 ms" by "10 ms" and sets it to "110 ms".

[0025] Figure 4 is a hardware configuration diagram of the operation device 800. The operation device 800 is provided with a CPU 81 as a central operation device, a ROM 82 as a read-only storage device, a RAM 83 as a readable and writable storage device, and an input and output device 84 as a user interface. The CPU 81 loads a program stored in the ROM 82 into the RAM 83 and executes it, whereby various operations such as the first learner 302 and the operation change section 309 are performed. The operation device 800 can also be realized by a rewritable logic circuit FPGA (Field Programmable Gate Array) or an application specific integrated circuit ASIC (Application Specific Integrated Circuit) instead of the combination of the CPU 81, the ROM 82, and the RAM 83. In addition, the operation device 800 can also be realized by a combination of different configurations, such as a combination of the CPU 81, the ROM 82, the RAM 83, and the FPGA, instead of the combination of the CPU 81, the ROM 82, and the RAM 83.

[0026] Figure 5 is a flowchart showing the process of the operation device 800. In step S301, the first learner 302 reads the initial value of the operation condition 301. In the next step S302, the first learner 302 and the second learner 306 operate the performance index 305 from the operation condition 301. Specifically, the first learner 302 operates the physical quantity 303 from the operation condition 301, and the second learner 306 operates the performance index 305 from the physical quantity 303.

[0027] In the next step S303, the operation change unit 309 determines whether the performance index 305 calculated in step S302 is an optimal value. The operation change unit 309 makes the determination in this step by, for example, comparing a predetermined threshold value with the performance index 305. If the operation change unit 309 determines that the performance index 305 is an optimal value, the operation ends. Figure 5 The process shown proceeds to step S304 when it is determined that the performance indicator 305 is not an optimal value.

[0028] In step S304, second gradient calculation unit 308 identifies the physical quantity that strongly influences performance index 305, namely, the index-influencing physical quantity. In step S305, first gradient calculation unit 307 identifies the operating condition that strongly influences the index-influencing physical quantity, namely, the index-influencing operating condition. In the following step S306, operation change unit 309 updates operating condition 301, and the process returns to step S302. When returning from step S306 to step S302, operating condition 301 updated in step S306 is used in step S302.

[0029] According to the first embodiment described above, the following effects can be obtained. (1) The computing device 800, which can also be referred to as an operating condition computing system, includes: a first learner 302 that receives as input the operating condition 301 of the internal combustion engine 1 as an onboard device and outputs as output a physical quantity 303 representing a physical state resulting from the operation of the internal combustion engine 1 based on the operating condition 301; a second learner 306 that receives as input the output of the first learner 302 and outputs a performance indicator 305; a first gradient computing unit 307 and a second gradient computing unit 308 that serve as gradient computing units for computing the gradients of the first learner 302 and the second learner 306; and an operation changing unit 309 that changes the operating condition 301 using the first gradient 321 and the second gradient 322. Therefore, it is possible to calculate the optimal operating condition 301 from the perspective of the performance indicator 305.

[0030] (2) The performance index 305 is at least one of the exhaust gas components and the combustion fluctuation rate from the internal combustion engine 1. The operation changing unit 309 changes at least one of the injection timing, injection distribution ratio, fuel pressure, and ignition timing during the distributed injection of the fuel.

[0031] (3) At least the rotation, load, intake pressure, intake temperature, water temperature, fuel pressure, number of fuel injections, fuel injection timing, distribution ratio, and ignition timing are input to the first learner 302, and at least the cylinder temperature, cylinder pressure, fuel adhesion amount, and over-rich mixture amount at the ignition timing are output.

[0032] (4) At least the in-cylinder temperature just before ignition, the in-cylinder pressure, the fuel adhesion amount, the over-rich mixture amount, and a part or all of the inputs to the first learner are input to the second learner 306. The second learner 306 outputs at least the exhaust gas composition after combustion and the combustion variation rate.

[0033] (Modified example 1) Figure 6 is a functional configuration diagram of the arithmetic device 800 of the modified example 1. In this diagram, a fuel variable 301C is added to the operation condition 301. This is different from the first embodiment. Figure 3 The fuel variable 301C is, for example, information on the properties of the fuel supplied to the internal combustion engine 1, the kind of the fuel, the timing of the fuel injection into the fuel tank, the replacement timing of the fuel injection device, and the like. In the combustion of the internal combustion engine 1, the combustion mode changes depending on the fuel used, and the amount of the PN or THC or the like emitted also changes. Furthermore, since the properties of the fuel change depending on the deterioration state of the fuel, by including the state of the fuel in the operation condition 301, it is possible to cope with the change in the fuel supply or the kind of the fuel.

[0034] According to this modified example 1, the following effects can be obtained. (5) The properties of the fuel injected into the internal combustion engine 1 are input to the first learner 302. Therefore, it is possible to calculate the operation condition 301 corresponding to the properties of the fuel.

[0035] (6) The kind of the fuel injected into the internal combustion engine 1 and the timing of the fuel injection into the fuel tank are input to the first learner 302. Therefore, it is possible to calculate the operation condition 301 corresponding to the kind of the fuel and the more detailed properties of the fuel such as the injection timing into the fuel tank.

[0036] (Modified example 2) The operation based on the arithmetic device 800 is not only applicable to the internal combustion engine 1 but also to various components mounted on a plurality of vehicles 9. For example, as explained below, it can also be applied to an electric motor. For example, in the electric motor, the current value applied to the electric motor corresponds to the operation condition 301, the generated magnetic flux and the interlinked magnetic flux to the magnetic core correspond to the physical quantity 303, and the torque, the speed, and the efficiency correspond to the performance index 305.

[0037] (Modified example 3) In the above-described first embodiment, the learning of the neural networks included in the first learner 302 and the second learner 306 is completed in advance. However, the first learner 302 and the second learner 306 can perform additional learning in parallel with the measurement and the inference.

[0038] Second embodiment Referring to Figure 7The second embodiment of the operating condition calculation system will be described. In the following description, the same reference numerals are assigned to the same components as in the first embodiment, and the main differences will be discussed. Parts not specifically described are the same as in the first embodiment. The main difference between this embodiment and the first embodiment is that the first learner 302 is composed of multiple learners.

[0039] Figure 7 This is a functional configuration diagram of the arithmetic device 800A according to the second embodiment. Figure 7 Compared with the first embodiment Figure 3 The difference is that the first learner 302 is replaced by a first learner 302A. The first learner 302A includes one or more learners that estimate only one specific physical quantity. Figure 7 , the figure shows a case where first learner 302A estimates three specific physical quantities. First learner 302A includes first-first learner 302-1, first-second learner 302-2, and first-third learner 302-3. The neural networks constituting first-first learner 302-1, first-second learner 302-2, and first-third learner 302-3 each have a node in the output layer that represents only one specific physical quantity.

[0040] The number of nodes in the input layer of the neural networks constituting the first learner 302-1, the second learner 302-2, and the third learner 302-3 may be different. Specifically, each neural network may input only specific conditions that are strongly correlated with the estimated specific physical quantity, or may input conditions obtained by excluding specific conditions that are weakly correlated with the estimated specific physical quantity from all specific conditions.

[0041] According to the second embodiment described above, the following effects can be obtained. (7) First learner 302A includes a machine learning engine for each specific physical quantity. All or part of the operating conditions are input to first-first learner 302-1, first-second learner 302-2, and so on. This allows for a reduction in the size of each neural network constituting first learner 302A, and improves the accuracy of the output results.

[0042] - Third embodiment - Reference Figure 8 The third embodiment of the operating condition calculation system will be described. In the following description, the same reference numerals are assigned to the same components as in the first embodiment, and the main differences will be discussed. Parts not specifically described are the same as in the first embodiment. The main difference between this embodiment and the first embodiment is that the second learner 306 is composed of multiple learners.

[0043] Figure 8is a functional configuration diagram of the arithmetic device 800B of the third embodiment. The difference from the first embodiment Figure 3 is that the second learner 306 is replaced with a second learner 306A. The second learner 306A includes one or more learners that estimate only one specific index. In Figure 8 , a case where the second learner 306A estimates three specific indexes is shown, and the second learner 306A includes a second-1 learner 306-1, a second-2 learner 306-2, and a second-3 learner 306-3.

[0044] The neural networks that constitute the second-1 learner 306-1, the second-2 learner 306-2, and the second-3 learner 306-3, respectively, have only nodes that represent one specific index in the output layer. Specifically, the second-1 learner 306-1, the second-2 learner 306-2, and the second-3 learner 306-3 output the first index 305-1, the second index 305-2, and the third index 305-3, respectively.

[0045] The number of nodes of the input layer of the neural networks that constitute the second-1 learner 306-1, the second-2 learner 306-2, and the second-3 learner 306-3, respectively, can also be different. That is, each neural network can input only a specific physical quantity that is strongly associated with the estimated specific index, or can input a condition in which a specific physical quantity that is weakly associated with the estimated specific index is excluded from all the specific conditions.

[0046] According to the third embodiment described above, the following effects can be obtained. (8) The second learner 306A has one machine learner for each performance index. All or a part of the output of the first learner 302 is input to the second-1 learner 306-1, the second-2 learner 306-2, and the like, respectively. Therefore, it is possible to reduce the size of each neural network that constitutes the second learner 306A, and an improvement in the accuracy of the output result can be expected.

[0047] - Fourth Embodiment - Referring to Figure 9 A fourth embodiment of the operation condition arithmetic system will be described. In the following description, the same reference signs are given to the same constituent elements as those of the first embodiment, and the differences will mainly be described. The parts not specifically described are the same as those of the first embodiment. In this embodiment, the first learner 302 and the second learner 306 are mainly different from those of the first embodiment in that each of them is constituted by a plurality of learners.

[0048] Figure 9 is a functional configuration diagram of the arithmetic device 800C of the fourth embodiment. The difference from the first embodiment Figure 3The difference is that a first learner 302A is provided instead of the first learner 302, and a second learner 306A is provided instead of the second learner 306. The first learner 302A is as described in the second embodiment, and the second learner 306A is as described in the third embodiment.

[0049] The first learner 302A includes one or more learners that estimate only one specific physical quantity. Figure 7 , the figure shows a case where first learner 302A estimates three specific physical quantities. First learner 302A includes first-first learner 302-1, first-second learner 302-2, and first-third learner 302-3. The neural networks constituting first-first learner 302-1, first-second learner 302-2, and first-third learner 302-3 each have a node in the output layer that represents only one specific physical quantity.

[0050] -Fifth embodiment- Reference Figure 10 The fifth embodiment of the operating condition calculation system will be described. In the following description, the same reference numerals are assigned to the same components as in the first embodiment, and the main differences will be described. Parts not specifically described are the same as in the first embodiment. This embodiment differs from the first embodiment primarily in detecting the operating state.

[0051] Figure 10 This is a functional block diagram of the computing device 800D of the fifth embodiment. Figure 3 800D is different in that it includes an operating state detection unit 331 and a sensor 332. Sensor 332 measures a value related to performance index 305 of the vehicle equipped with computing device 800D. Sensor 332 is, for example, a PM sensor or a THC measuring device. The measured value is input from sensor 332 to operating state detection unit 331, and performance index 305 is input from second learner 306 to operating state detection unit 331.

[0052] Performance indicator 305 is estimated based on the output of sensor 332. Operating state detection unit 331 compares estimated performance indicator 305 with performance indicator 305 output by second learner 306. If a deviation between the two exceeds a predetermined threshold, abnormality detection signal 333 is output. Abnormality detection signal 333 can be reported to the occupants of vehicle 9 or read by a device (not shown) mounted on vehicle 9. For example, if the THC calculated by operating state detection unit 331 deviates from the THC calculated by second learner 306, a fault such as an increase in the amount of fuel deposited due to an abnormality in the fuel injection system or ignition failure due to wear of spark plug 105 is considered. In other words, the present invention can be used as an indicator to confirm whether internal combustion engine 1 is operating normally.

[0053] (9) The operation state detection section 331 is capable of detecting the operation state of the internal combustion engine 1 based on the output of the sensor 332 mounted on the vehicle 9 and the performance index calculated by the second learner 306. Therefore, it is possible to easily determine whether the internal combustion engine 1 is operating normally.

[0054] - 6th Embodiment - Reference Figure 11 The 6th embodiment of the operation condition calculation system will be described. In the following description, the same symbols are assigned to the same constituent elements as those of the 1st embodiment, and the differences will mainly be described. The portions not specifically described are the same as those of the 1st embodiment. In the present embodiment, the main difference from the 1st embodiment is that the display section for displaying the calculation result is provided. In addition, in the present embodiment, the calculation device is not mounted on the vehicle.

[0055] Figure 11 is a functional configuration diagram of the calculation device 800E of the 6th embodiment. Figure 11 The difference from Figure 3 is that the user interface 350 is provided. The user interface 350 functions as an input section and an output section. The user interface 350 outputs the calculation results of the first learner 302, the second learner 306, and the operation change section 309. The user interface 350 accepts the operation condition 301 input by the user.

[0056] The user interface 350 has an initial operation condition input field 351, an internal combustion engine state display field 352, a performance index display field 353, and a next operation condition display field 354. The user inputs the operation condition 301 in the initial operation condition input field 351 using the mouse or the keyboard included in the input / output device 84. The value input to the initial operation condition input field 351 is input to the first learner 302. The first learner 302 in the present embodiment also outputs the calculated physical quantity 303 to the user interface 350, and the user interface 350 displays the physical quantity 303 in the internal combustion engine state display field 352.

[0057] The second learner 306 in the present embodiment also outputs the calculated performance index 305 to the user interface 350, and the user interface 350 displays the performance index 305 in the performance index display field 353. The operation change section 309 in the present embodiment also outputs the calculated new operation condition 310 to the user interface 350, and the user interface 350 displays the new operation condition 310 in the next operation condition display field 354.

[0058] When the user inputs the operation condition 301, the estimation results of the first learner 302 and the second learner 306 are displayed on the user interface 350. The user can judge whether to adopt the input operation condition by confirming these. In addition, in the case where the user desires to further improve the performance index 305, the user can refer to the next operation condition display field 354.

[0059] According to the above-described sixth embodiment, the following advantageous effects can be obtained. (10) The user interface 350 that functions as an input section of the input operation condition 301, and the user interface 350 that functions as a display section of the output of the first learner 302, the output of the second learner 306, and the output of the operation change section 309. Therefore, the appropriate operation condition 301 can be easily calculated.

[0060] In each of the above-described embodiments and modified examples, the configuration of the functional block is only one example. Several functions indicated as different functional blocks can be integrally configured, or the configuration indicated by one functional block diagram can be divided into two or more functions. In addition, a part of the function possessed by each functional block can be provided as a configuration of another functional block.

[0061] In each of the above-described embodiments and modified examples, the program is stored in the ROM 82, but the program can also be stored in a nonvolatile storage device not shown. In addition, the arithmetic device 800 can be provided with an input-output interface not shown, and can read the program from another device through the input-output interface and a medium available to the arithmetic device 800 when necessary. The medium here refers to, for example, a storage medium that can be attached to and detached from the input-output interface, or a communication medium, that is, a wired, wireless, optical, or the like network, or a carrier wave or a digital signal propagated in the network. In addition, a part or all of the functions realized by the program can be realized by a hardware circuit or an FPGA.

[0062] Each of the above-described embodiments and modified examples can be combined. In the above, various embodiments and modified examples are described, but the present application is not limited to these. Other modes considered within the scope of the technical idea of the present application are also included in the scope of the present application. Explanation of Symbols

[0063] 1: Internal combustion engine 9: Vehicle 84: Input-output device 301: Operation condition 302: First learner 303: Physical quantity 305: Performance index 306: Second learner 307: First gradient calculation section 308: 2nd gradient calculation section 309: operation change section 310: new operation condition 321: 1st gradient 322: 2nd gradient 331: operation state detection section 332: sensor 350: user interface 800, 800A-800E: arithmetic device

Claims

1. An operating condition calculation system, characterized in that: have: a first learner that receives as input an operating condition of an on-board device mounted on a vehicle and outputs as output a physical state resulting from the on-board device operating based on the operating condition; a second learner that takes the output of the first learner as input and outputs a performance indicator; a gradient calculation unit that calculates the gradients of the first learner and the second learner; and An operation changing unit changes the operating condition using the gradient.

2. The operating condition calculation system according to claim 1, characterized in that: The on-vehicle device is an internal combustion engine.

3. The operating condition calculation system according to claim 2, characterized in that: The performance indicator is at least one of exhaust gas components and combustion fluctuation rate from the internal combustion engine, The operation changing portion changes at least one of an injection timing, an injection distribution ratio, a fuel pressure, and an ignition timing during distributed injection of the fuel.

4. The operating condition calculation system according to claim 2, characterized in that: At least the speed, load, intake pressure, intake temperature, water temperature, fuel pressure, number of fuel injections, fuel injection timing, distribution ratio, and ignition timing are input to the first learner, and at least the cylinder temperature, cylinder pressure, fuel adhesion amount, and over-rich mixture amount at the ignition timing are output.

5. The operating condition calculation system according to claim 2, wherein: The second learner receives at least the cylinder temperature, cylinder pressure, fuel adhesion amount, and rich mixture amount immediately before ignition, as well as part or all of the input to the first learner, and outputs at least the exhaust gas components generated after combustion and the combustion variation rate.

6. The operating condition calculation system according to claim 2, characterized in that: The first learner has a machine learning machine for each physical quantity. All or part of the operating conditions are input to the machine learning machine.

7. The operating condition calculation system according to claim 2, characterized in that: The second learner has a machine learning machine for each performance indicator, All or part of the output of the first learner is input to each of the machine learning machines.

8. The operating condition calculation system according to claim 4, characterized in that: The properties of the fuel fed into the internal combustion engine are input to the first learner.

9. The operating condition calculation system according to claim 4, characterized in that: The first learner receives as input the type of fuel fed into the internal combustion engine or the timing at which fuel is fed into the fuel tank.

10. The operating condition calculation system according to claim 2, wherein: An operating state detection unit is provided for calculating the performance index using outputs of sensors mounted on the vehicle and detecting abnormality of the vehicle by comparing the performance index with the performance index calculated by the second learner.

11. The operating condition calculation system according to claim 2, wherein: have: an input unit that inputs the operating conditions; and A display unit displays the output of the first learner, the output of the second learner, and the output of the operation changing unit.

12. The operating condition calculation system according to claim 2, wherein: The performance index is an index indicating the status of the components and devices of the vehicle. The operating condition is sensor information acquired from a sensor mounted on the vehicle.

13. A control device for an internal combustion engine, characterized in that: A system comprising the operating condition calculation system according to claim 2.

14. An adapter device for an internal combustion engine, characterized in that: A system comprising the operating condition calculation system according to claim 2.

Images