Verification-cycle generation method, verification-cycle generation system, and verification-cycle generation program

Abstract

The present invention is a verification-cycle generation method for generating a verification cycle for automobile development, the verification cycle being realistic and satisfying boundary conditions for a plurality of automobiles, wherein a plurality of data sets concerning a driving cycle are input, the plurality of data sets are individually clustered and classified into a plurality of data groups, a plurality of verification cycles concerning automobile development are generated on the basis of the plurality of clustered data groups, each of the plurality of verification cycles is executed through simulation or with a real automobile, and a representative cycle, which best satisfies a criterion defined by a user, is selected from among the plurality of verification cycles.

Description

Verification cycle generation method, verification cycle generation system, and verification cycle generation program

[0001] The present invention relates to a verification cycle generation method, a verification cycle generation device, and a verification cycle generation program used in automobile development.

[0002] In conventional automobile development, as shown in Non-Patent Document 1, for example, confirmation tests are performed using a transient test cycle (verification cycle) under conditions (boundary conditions) that are expected to be unfavorable for the automobile on actual roads. The quality of automobile development is then judged based on the results of this confirmation test. Typically, this verification cycle is often created based on data acquired by driving a specific automobile on actual roads.

[0003] However, during actual road driving, the conditions (boundary conditions) that are expected to be unfavorable for a vehicle on the actual road change depending on the driving environment, traffic conditions, route, driving style, vehicle specifications such as the powertrain, etc. For example, in an RDE (Real Driving Emission) test, the worst-case (boundary conditions) for exhaust gases change depending on the driving environment, traffic conditions, route, driving style, vehicle specifications, etc. Therefore, it is virtually impossible to cover the boundary conditions for many vehicles in a single verification cycle based on past data and experience.

[0004] Michael Gorgen et al. “Holistic Hybrid RDE Calibration Methodology for EU7” FEV Engine Congress 2021 paper 14

[0005] The present invention has been made to solve the above problems, and its main object is to generate a realistic verification cycle that satisfies the boundary conditions for a plurality of automobiles.

[0006] In other words, the verification cycle generation method of the present invention is a verification cycle generation method for generating a verification cycle for automobile development, which is characterized by inputting a plurality of data sets regarding driving cycles, clustering each of the plurality of data sets to classify them into a plurality of data sets, generating a plurality of verification cycles related to automobile development based on the clustered plurality of data sets, executing each of the plurality of verification cycles on a simulation device, an actual device, or a combination of these, and selecting from the plurality of verification cycles a representative cycle that best satisfies criteria defined by a user.

[0007] In this way, a plurality of data sets relating to driving cycles are input, each of the plurality of data sets is clustered and classified into a plurality of data sets, and a verification cycle for automobile development is generated based on the plurality of clustered data sets, so that a realistic verification cycle that satisfies the boundary conditions for a plurality of vehicles can be generated. In particular, in the present invention, a plurality of verification cycles for automobile development are generated, each of the plurality of verification cycles is executed on a simulation device, an actual machine, or a combination of these, and a representative cycle that best satisfies user-defined criteria is selected from the plurality of verification cycles, so that a more realistic verification cycle (representative cycle) that satisfies the boundary conditions for a plurality of vehicles can be generated.

[0008] As a specific embodiment of the data sets of the driving cycle, the plurality of data sets may include vehicle speed information relating to the vehicle speed and load information relating to the load.

[0009] In a specific embodiment of the data set of the driving cycle, the plurality of data sets may further include brake information relating to braking or environmental factor information relating to environmental factors.

[0010] A specific embodiment of generating multiple verification cycles is to generate multiple verification cycles by performing a series of steps multiple times: clustering each of the multiple data sets into multiple data sets; and generating one verification cycle for automobile development based on the clustered multiple data sets.

[0011] As a specific embodiment for generating a plausible verification cycle, it is conceivable to generate the verification cycle from the clustered plurality of data sets using a Markov chain.

[0012] In order to increase the number of data sets and generate more realistic validation cycles, it is desirable to adjust the input data sets by applying a predetermined adjustment factor to the data sets.

[0013] The user-defined criteria are preferably criteria for selecting the worst verification cycle from the plurality of verification cycles, criteria for selecting the best verification cycle, or criteria for selecting the verification cycle most frequently used by the user. In this configuration, the verification cycle that satisfies the criteria for selecting the worst verification cycle is the worst-case verification cycle. For example, in the case of an engine vehicle, the verification cycle may be the worst-case verification cycle, such as fuel consumption or CO ２ It is the cycle with the highest emissions, and in the case of an electric vehicle, it is the cycle with the worst power consumption (hydrogen consumption in the case of an FCV). Also, the cycle that meets the criteria for selecting the best verification cycle is the best case verification cycle. For example, in the case of an engine vehicle, it is the cycle with the worst fuel consumption or CO ２ This is the cycle that produces the lowest emissions, and in the case of electric vehicles, it is the cycle that produces the best power consumption (hydrogen consumption in the case of FCVs).

[0014] In order to be suitable for use in road emission tests such as RDE, it is desirable that the verification cycle is related to the road emission test, and that the multiple data sets to be input include the vehicle speed information and the load information in the road emission test.

[0015] In order to be suitable for use in road battery evaluation tests of vehicles having batteries, it is desirable that the verification cycle relates to road battery evaluation tests, and that the multiple input data sets include the vehicle speed information, the load information, and deceleration information related to deceleration in the road battery evaluation tests.

[0016] In order to be suitable for use in road battery evaluation tests, the deceleration information preferably includes the amount of regeneration.

[0017] In addition, the verification cycle generation device of the present invention is a verification cycle generation device that generates a verification cycle for automobile development, and is characterized by comprising: a dataset receiving unit that receives multiple datasets for driving cycles; a clustering unit that clusters each of the multiple datasets and classifies them into multiple data sets; a verification cycle generation unit that generates multiple verification cycles for automobile development based on the clustered multiple data sets; and a simulation device, an actual machine (test facility), or a combination of these that executes each of the multiple verification cycles.

[0018] Furthermore, the verification cycle generation program of the present invention is a verification cycle generation program for generating a verification cycle for automobile development, and is characterized in that it causes a computer to perform the functions of a dataset receiving unit that receives multiple datasets for driving cycles, a clustering unit that clusters each of the multiple datasets and classifies them into multiple data sets, a verification cycle generation unit that generates multiple verification cycles for automobile development based on the clustered multiple data sets, and a simulation unit that executes each of the multiple verification cycles.

[0019] According to the present invention as described above, it is possible to generate realistic verification cycles that satisfy boundary conditions for a plurality of automobiles.

[0020] FIG. 1 is a diagram schematically showing the configuration of a verification cycle generation device according to an embodiment of the present invention. FIG. 2 is a diagram showing an example of clustering according to the embodiment. FIG. 3 is a diagram showing a verification cycle generation method using a Markov chain according to the embodiment. FIG. 4 is a diagram showing a verification cycle generation method according to the embodiment. FIG. 5 is a diagram showing a verification cycle generation method according to a modified embodiment. ２ FIG. 1 is a distribution diagram showing the distribution of emissions.

[0021] A verification cycle generation method and a verification cycle generation system according to an embodiment of the present invention will be described below with reference to the drawings.

[0022] <Device Configuration of Verification Cycle Generation System> The verification cycle generation system 100 of this embodiment generates verification cycles used in automobile development (including adaptation). Here, automobiles include, for example, internal combustion engine automobiles, hybrid automobiles, electric automobiles, and fuel cell automobiles.

[0023] Specifically, the verification cycle generation system 100 includes a dataset receiving unit 2 that receives multiple datasets for driving cycles, a clustering unit 3 that clusters each of the multiple datasets and classifies them into multiple data sets, a verification cycle generation unit 4 that generates multiple verification cycles for automobile development based on the clustered multiple data sets, and a simulation device 5 that executes each of the multiple verification cycles.

[0024] In this embodiment, the dataset receiving unit 2, the clustering unit 3, and the verification cycle generating unit 4 are configured by a cycle generating device 10 (a computer having a CPU, an internal memory, an input / output interface, etc.) that is separate from the simulation device 5. The dataset receiving unit 2, the clustering unit 3, and the verification cycle generating unit 4 may be configured by a computer common to the simulation device, or the dataset receiving unit 2, the clustering unit 3, and the verification cycle generating unit 4 may be configured by separate computers.

[0025] The dataset receiving unit 2 receives a plurality of datasets for a driving cycle. Here, the plurality of datasets include datasets acquired in the past and / or datasets of simulation results obtained by a simulation device such as a digital twin. Furthermore, the plurality of datasets satisfy the standards for a driving cycle determined by laws and regulations, etc. For example, if the driving cycle is determined by an RDE, the plurality of datasets satisfy the RDE test standards.

[0026] Each data set also includes vehicle speed information related to the vehicle speed and load information related to the load. Here, the vehicle speed information is, for example, vehicle speed, engine speed, or motor speed. The load information is factors that determine the load, such as accelerator depression amount (accelerator position), throttle opening, fuel injection amount, torque requirement amount, power requirement amount, hydrogen requirement amount, or air (oxygen) requirement amount. Furthermore, each data set may further include brake information related to the brakes or environmental factor information related to environmental factors. Here, the brake information is, for example, brake depression amount (brake position). The environmental factor information is, for example, air pressure, pressure, temperature, humidity, road surface conditions (road surface μ, road surface roughness), or road conditions (traffic congestion amount).

[0027] The clustering unit 3 clusters each of the multiple data sets received by the data set receiving unit 2 and classifies the data into multiple data sets. The clustering method used by the clustering unit 3 can be, for example, a non-hierarchical clustering method such as the k-means method or the Fussy clustering method. FIG. 2 shows a graph resulting from clustering using the k-means method. In this graph, the vertical axis represents engine speed (vehicle speed) and the horizontal axis represents accelerator opening. The clustering unit 3 also clusters the multiple data sets into multiple categories, such as "Hard Acceleration," "Mild Acceleration," "Constant Speed," "Mild Acceleration," "Hard Acceleration," and / or "Stop."

[0028] The verification cycle generation unit 4 generates a verification cycle for automobile development based on the plurality of data sets clustered by the clustering unit. Specifically, the verification cycle generation unit 4 generates a verification cycle from the plurality of clustered data sets using a Markov chain.

[0029] As shown in FIG. 3, the verification cycle generation unit 4 calculates the probability that a node will occur from a given node to the next node, and creates the most likely driving pattern by connecting the nodes with the highest probability.

[0030] Here, each node is a single piece of data indicating "engine speed - accelerator opening" and is classified into a plurality of clustered categories ("Hard Acceleration," "Mild Acceleration," "Constant Speed," "Mild Acceleration," "Hard Acceleration," and / or "Stop"). Transition probabilities between the plurality of nodes are calculated, and the nodes are connected based on these transition probabilities to generate the most likely driving pattern (verification cycle). Note that Figure 3 shows a transition probability matrix for transitions from (i) to (j) for each node included in the six categories, as well as a numerical example for transitioning from the "Hard Acceleration" node to the "Hard Acceleration" node.

[0031] The verification cycle generated by the verification cycle generation unit 4 is verified by a simulation device 5. For example, a simulation device called a digital twin can be used as the simulation device 5. Note that data such as the driving environment, traffic conditions, route, driving style, and vehicle specifications are input in advance into the simulation device 5 in order to enable the verification cycle to be executed.

[0032] <Verification Cycle Generation Method> Next, a verification cycle generation method using the verification cycle generation system of this embodiment will be described with reference to Fig. 4. ２ An example of generating a verification cycle with the largest amount of emissions (a worst-case verification cycle) will be described.

[0033] A plurality of data sets of various driving cycles in road exhaust gas tests are input to the data set receiving unit 2. Here, the plurality of data sets include actual data sets acquired in the past and / or virtual data sets of simulation results obtained by a simulation device such as a digital twin. In addition, the plurality of data sets in this embodiment include data sets of CO ２ The actual data set for the cycle with the highest emissions (the actual data set for the bad case), and / or CO ２ It includes a hypothetical dataset for the cycle with the highest emissions (a bad-case hypothetical dataset).

[0034] Then, the clustering unit 3 clusters each of the plurality of data sets and classifies them into a plurality of data sets. In this example, the clustering is performed into six categories: "Hard Acceleration," "Mild Acceleration," "Constant Speed," "Mild Acceleration," "Hard Acceleration," and "Stop."

[0035] Next, the verification cycle generation unit 4 generates one verification cycle for on-road exhaust gas testing related to automobile development using a Markov chain based on the clustered data sets. ２ This is the verification cycle that is expected to produce the highest emissions.

[0036] In this embodiment, as described above, multiple verification cycles are generated by performing a series of steps multiple times, in which each of multiple data sets is clustered and classified into multiple data sets, and one verification cycle for automobile development is generated based on the clustered multiple data sets.

[0037] Then, each of the multiple verification cycles is executed by the simulation device 5, and a representative cycle that best satisfies the user-defined criteria is selected from the multiple verification cycles. The user-defined criteria in this embodiment are criteria for selecting the worst verification cycle from the multiple verification cycles. In other words, multiple verification cycles are executed in a simulation, and the CO ２ The verification cycle with the highest emissions is selected as the representative cycle.

[0038] <Effects of this embodiment> According to the verification cycle generation method of this embodiment, multiple data sets regarding driving cycles are input, each of the multiple data sets is clustered and classified into multiple data sets, and a verification cycle for automobile development is generated based on the multiple clustered data sets, so that a realistic verification cycle that satisfies the boundary conditions for multiple vehicles can be generated.

[0039] In particular, in this embodiment, multiple verification cycles related to automobile development are generated, each of the multiple verification cycles is executed by the simulation device 5, and a representative cycle that best satisfies the criteria defined by the user is selected from the multiple verification cycles, making it possible to generate a more realistic verification cycle (representative cycle) that satisfies the boundary conditions for multiple vehicles.

[0040] Furthermore, in this embodiment, by using simulation data obtained from a digital twin or the like in addition to previously acquired data sets, a more realistic verification cycle that satisfies any boundary condition that matches the vehicle being evaluated can be created using clustering and Markov chain probabilistic techniques. The vehicle evaluated using this verification cycle can be more robust than evaluations performed using conventional empirical verification cycles. Furthermore, according to this embodiment, driving under environmental conditions and driving styles that are not realistic can be performed using a simulation device, and these can be incorporated into the input data set.

[0041] Other Modified Embodiments The present invention is not limited to the above-described embodiments.

[0042] For example, in the above embodiment, the generated verification cycle is executed by the simulation device 5 to determine whether it is the cycle that best satisfies the criteria defined by the user, but it may also be possible to execute the verification cycle using, for example, an actual machine (test facility) or a combination of the actual machine and the simulation device 5 to determine whether it is the cycle that best satisfies the criteria defined by the user. Here, the actual machine (test facility) includes a chassis dynamometer, an engine dynamometer, a motor dynamometer, a driveline dynamometer, or the like.

[0043] 5, a data set may be adjusted by applying a predetermined adjustment factor to the input data set. This configuration not only allows the number of data sets to be input to be increased, but also makes it easier to generate a cycle that best satisfies user-defined criteria (e.g., a cycle that produces worse results) by adjusting the adjustment factor.

[0044] 6, the verification cycles generated by the verification cycle generation unit 4 may be input as multiple data sets to the data set reception unit 2, or the representative cycle may be input as multiple data sets to the data set reception unit 2. By inputting the verification cycles generated in this way again, the representative cycle can be made to more closely match the above criteria.

[0045] In the above embodiment, the worst verification cycle is selected from among a plurality of verification cycles, but the verification cycle with the best results may be selected, or the verification cycle most frequently used by the user, or the verification cycle with the average results may be selected. Here, the verification cycle most frequently used by the user is selected based on the CO ２ The most frequent CO in the distribution of emissions ２ Included in the scope of emissions.

[0046] Furthermore, the exhaust gas test of the embodiment is carried out using carbon dioxide (CO ２), but the exhaust gas test was to measure carbon monoxide (CO), nitrogen oxides (NO Ｘ ), hydrocarbons (CH), total hydrocarbons (THC), ammonium (NH ３ ), formaldehyde (HCHO), nitrous oxide (N ２ The exhaust gas analyzer may measure components contained in the exhaust gas, such as particulate matter (PM), particulate matter (PM), and / or PM particle number (PN).

[0047] Furthermore, although the above embodiment generates a verification cycle related to an exhaust gas test, a verification cycle related to an on-road battery evaluation test may also be generated. In this case, the multiple input data sets include vehicle speed information, load information, and / or deceleration information related to deceleration in the on-road battery evaluation test. The deceleration information includes the brake depression amount or regeneration amount, etc. Furthermore, a verification cycle related to a fuel consumption evaluation test, a verification cycle related to a power consumption evaluation test, a verification cycle related to a hydrogen consumption evaluation test, a verification cycle related to a brake dust evaluation test, or a verification cycle related to a tire dust evaluation test may also be generated.

[0048] Although various verification cycles that satisfy user-defined criteria can be created for these evaluation tests, a verification cycle that can be considered as an example of a worst-case scenario will be described below. For the verification cycle for the on-road battery evaluation test, for example, a verification cycle with the highest power consumption can be generated. For the verification cycle for the fuel consumption evaluation test, for example, a verification cycle with the highest fuel consumption can be generated. For the verification cycle for the power consumption evaluation test, for example, a verification cycle with the highest power consumption can be generated. For the verification cycle for the hydrogen consumption evaluation test, for example, a verification cycle with the highest hydrogen consumption can be generated. For the verification cycle for the brake dust or tire dust evaluation test, for example, a verification cycle with the highest brake dust or tire dust amount can be generated.

[0049] In addition, various modifications and combinations of the embodiments may be made as long as they do not go against the spirit of the present invention.

[0050] According to the present invention, it is possible to generate realistic verification cycles that satisfy boundary conditions for multiple vehicles.

[0051] REFERENCE SIGNS LIST 100: Verification cycle generation system 10: Cycle generation device 2: Data set reception unit 3: Clustering unit 4: Verification cycle generation unit 5: Simulation device

Claims

1. A test specimen testing system for testing a vehicle having an automated driving system (hereinafter, AD) or an advanced driver assistance system (hereinafter, ADAS) or a test specimen that is a part thereof, A dynamometer for applying a load to the specimen; An automatic driving robot that performs braking, accelerating or steering of the test specimen, A specimen testing system in which the autonomous driving robot links an active driving state in which the specimen actively drives using AD or ADAS, and a passive driving state in which the specimen passively drives using the brakes, accelerator, or steering wheel.

2. A control unit for controlling the autonomous driving robot is further provided. The specimen testing system according to claim 1 , wherein the control unit controls the autonomous driving robot to link the active driving state with the passive driving state.

3. The test piece further includes an ambient environment input device for inputting a simulated signal simulating an ambient environment to the test piece, The specimen testing system according to claim 2 , wherein the control unit controls the autonomous driving robot in response to an input from the surrounding environment input device to link the active driving state with the passive driving state.

4. The control unit controls the dynamometer, The specimen testing system according to claim 2 or 3, wherein the control unit controls the autonomous driving robot and the dynamometer to link the active driving state with the passive driving state.

5. The autonomous driving robot is An actuator for starting or ending active travel by the AD or ADAS of the test specimen; 2. The specimen testing system according to claim 1, wherein at least one of the brake operation, accelerator operation, and steering operation of the specimen is stopped at the timing when active driving by the AD or ADAS of the specimen is started.

6. The autonomous driving robot is 6. The specimen testing system according to claim 5, wherein at least one of braking operation, accelerator operation, and steering operation of the specimen is resumed when active driving by the AD or ADAS of the specimen is terminated.

7. Further comprising a camera for capturing an image of the instrument panel or head-up display of the test specimen, and an analysis device for reading and analyzing the captured image of the camera; the actuation actuator has a function of setting an adaptive cruise control of the test specimen, The specimen testing system according to claim 5 or 6, wherein the autonomous driving robot sets at least one of the target vehicle speed of the specimen or the distance between the specimen and the vehicle in front by the actuation actuator based on the results of reading and analyzing the image captured by the camera.

8. 7. The system for testing a specimen according to claim 5 or 6, wherein the actuation actuator is fixed to a steering wheel of the specimen.

9. 9. The specimen testing system according to claim 8, further comprising a hands-on simulation section for simulating a hands-on state, the hands-on simulation section being provided on a fixing member for fixing the operating actuator to the steering wheel.

10. 4. The specimen testing system according to claim 1, wherein the autonomous driving robot has a stop actuator that cancels active driving by the AD or ADAS of the specimen by stepping on the brake pedal of the specimen or by pressing a cancel button of the AD or ADAS function.

11. A test specimen testing method for testing a vehicle having an automated driving system (hereinafter, referred to as AD) or an advanced driver assistance system (hereinafter, referred to as ADAS) or a part thereof, comprising: A dynamometer is used to apply a load to the test specimen, and an automatic driving robot is used to perform braking, accelerating, or steering of the test specimen. A specimen testing method in which the autonomous driving robot links an active driving state in which the specimen actively drives using AD or ADAS with a passive driving state in which the specimen passively drives using the brakes, accelerator or steering wheel.

12. A test specimen test program used in a test specimen test system that tests a vehicle having an automated driving system (hereinafter, AD) or an advanced driver assistance system (hereinafter, ADAS) or a test specimen that is a part thereof, The specimen testing system includes: A dynamometer for applying a load to the specimen; An automatic driving robot that operates the brakes, accelerators, or steering wheels of the test specimen, The specimen testing program controls the autonomous driving robot to cause a computer to perform the function of linking an active driving state in which the specimen actively drives using AD or ADAS with a passive driving state in which the specimen passively drives using the brakes, accelerator or steering wheel.