Engine torque estimation device

Abstract

To improve estimation accuracy of engine torque.SOLUTION: This engine torque estimating device 10 is provided with a CPU11 and a second memory 13. The second memory 13 stores mapping data DM that defines a mapping learned by machine learning. The mapping outputs an output variable corresponding to the engine torque when an input variable is input. The input variables include the engine speed NE, the intake air amount GA, the fuel flow rate FFR, and the ignition timing AIG. The CPU11 acquires an input variable and derives an engine torque estimated value TQ based on an output variable output from a mapping by inputting the acquired input variable to the mapping.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an engine torque estimation device. [Background technology]

[0002] Patent Document 1 discloses an example of an estimation device equipped with a neural network. The estimation device acquires crankshaft rotation data obtained using a crankshaft rotation sensor. The estimation device estimates engine torque based on an output variable of a neural network when the crankshaft rotation data is input as an input variable to the neural network. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2008-151118 Summary of the Invention [Problem to be solved by the invention]

[0004] In recent years, there has been a demand for improved accuracy in estimating engine torque. [Means for solving the problem]

[0005] An engine torque estimation device for solving the above problem estimates engine torque, which is the output torque of a spark-ignition engine. The engine torque estimation device includes an execution device and a storage device. The storage device stores mapping data that defines a mapping learned through machine learning. The mapping outputs an output variable corresponding to engine torque when input variables are input. The input variables include engine speed, intake air amount, fuel flow rate, and ignition timing. The execution device executes the following steps: acquiring the input variables; and inputting the acquired input variables into the mapping to derive an estimated value of engine torque based on the output variable output from the mapping. [Effects of the Invention]

[0006] To improve the accuracy of estimating engine torque. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a block diagram showing the configuration of an engine torque estimation device according to an embodiment. [Figure 2] FIG. 2 is a flowchart showing the flow of processing executed by the engine torque estimation device of FIG. DETAILED DESCRIPTION OF THE INVENTION

[0008] An embodiment of the engine torque estimation device will be described with reference to FIGS. <Configuration of engine torque estimation device> 1 shows an example of an engine torque estimation device 10. The engine torque estimation device 10 estimates engine torque, which is the output torque of a spark ignition engine.

[0009] The engine torque estimation device 10 includes a CPU 11, a first memory 12, and a second memory 13. The first memory 12 stores various control programs executed by the CPU 11. The second memory 13 stores mapping data DM that defines a mapping learned by machine learning. An example of a mapping is a neural network. In this embodiment, the CPU 11 corresponds to an "execution device," and the second memory 13 corresponds to a "storage device."

[0010] When input variables are input to the mapping, the mapping outputs an output variable corresponding to engine torque. The input variables of the mapping include engine speed NE, intake air amount GA, fuel flow rate FFR, and ignition timing AIG. The input variables include command torque TQ*, which is an engine torque command value, intake air temperature TPA, which is the temperature of air introduced into the engine cylinders, and fuel injection amount QF. Further, the input variables include engine air-fuel ratio AF and alternator power generation EGC.

[0011] <How to learn mapping> The learning device acquires a plurality of data pairs each consisting of training data and teacher data. The training data includes the input variables described above. The teacher data is an actual value of the engine torque.

[0012] The learning device acquires the output variables of the mapping when one piece of training data is input to the mapping as an input variable. The learning device compares the estimated engine torque value TQ derived based on the acquired output variables with the actual engine torque value, which is the training data. The learning device then updates various parameters in the mapping so that the estimated engine torque value TQ approaches the actual engine torque value.

[0013] The learning device generates mapping data DM that defines the learned mapping by performing this type of machine learning a predetermined number of times or more. <Process for estimating engine torque> A series of processes executed by the CPU 11 to estimate the engine torque will be described with reference to Fig. 2. The CPU 11 repeatedly executes the series of processes shown in Fig. 2 at every predetermined control period.

[0014] In step S11, the CPU 11 acquires various engine state values, including the engine speed NE, intake air amount GA, fuel flow rate FFR, ignition timing AIG, command torque TQ*, intake air temperature TPA, fuel injection amount QF, engine air-fuel ratio AF, and alternator power generation EGC.

[0015] In the next step S13, the CPU 11 inputs the various state values ​​acquired in step S11 as input variables to the mapping. In the following step S15, the CPU 11 acquires output variables of the mapping.

[0016] Then, in step S17, the CPU 11 derives the engine torque estimated value TQ based on the output variables of the mapping obtained in step S15, and then the CPU 11 ends the series of processes.

[0017] <Actions and Effects of This Embodiment> In the engine torque estimation device 10, when estimating engine torque, the CPU 11 inputs various state values, including the engine speed NE, as input variables to the mapping. The various state values ​​include the engine speed NE, the intake air amount GA, the fuel flow rate FFR, and the ignition timing AIG. These state values ​​are parameters that affect the engine torque.

[0018] The CPU 11 inputs these various state values ​​as input variables to the mapping, and derives the engine torque estimated value TQ based on the output variables output from the mapping. Therefore, the engine torque estimation device 10 can improve the estimation accuracy of the engine torque by increasing the types of state values ​​input to the mapping as input variables.

[0019] <Example of change> The above embodiment can be modified as follows: The above embodiment and the following modifications can be combined with each other within the scope of technical compatibility.

[0020] The input variables of the mapping do not need to include other parameters as long as they include the engine speed NE, the intake air amount GA, the fuel flow rate FFR, and the ignition timing AIG. The mapping may be a trained model other than a neural network. [Explanation of symbols]

[0021] 10...engine torque estimation device, 11...CPU, 13...second memory, DM...mapping data.

Claims

[Claim 1] An engine torque estimation device that estimates engine torque, which is an output torque of a spark ignition engine, an execution device and a storage device; The storage device stores mapping data that defines a mapping that has been learned by machine learning, The mapping outputs an output variable corresponding to an engine torque when an input variable is input, The input variables include engine speed, intake air flow, fuel flow, and ignition timing; The execution device obtaining the input variables; and deriving an estimated value of engine torque based on the output variables output from the mapping by inputting the acquired input variables into the mapping. Engine torque estimator.

Images