Method for creating a judgment space, method for updating a judgment space, method for determining road surface conditions, judgment space creating device, and road surface condition determining device

Abstract

A method according to the present invention is for, by using information on the road surface state of a road surface on which a vehicle is traveling and output waveform data obtained by measuring a change in physical quantity of tires in the vehicle during traveling, creating a determination space for determining the road surface state from the output waveform data. The method involves: acquiring the output waveform data for a plurality of different road surface states and traveling conditions at a time of measuring the output waveform data (S11); classifying the output waveform data into a plurality of groups on the basis of the traveling conditions (S12); performing machine learning for each of the groups using training data in which information on the road surface states are associated with the output waveform data (S13, 14, 15); and thereby creating a determination space corresponding to each of the groups (S16). Accordingly, this method makes it possible to determine the road surface state with high accuracy.

Description

[Technical field]

[0001] The present invention relates to a method and device for subjecting output waveform data indicating changes in physical quantities of tires measured using a sensor for different road surface conditions to machine learning, and creating a decision space for determining the road surface condition from the output waveform data, a method for updating the decision space, and a method and device for determining the road surface condition when data to be determined is obtained using the created decision space. [Background technology]

[0002] Conventionally, a vibration detection means attached to a tire detects a time-varying waveform of tire vibration during running, and the time-varying waveform is used to determine the road surface condition on which the tire is in contact. For example, Patent Document 1 describes a road surface condition determination method that uses a feature amount of the time-varying waveform of the front wheel to determine a dry road surface from a wet road surface, and uses a feature amount of the time-varying waveform of the rear wheel to determine a dry road surface from an icy road surface, a dry road surface from a snowy road surface, or a dry road surface from an icy and snowy road surface. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2019-123293 A DISCLOSURE OF THEINVENTION [Problem to be solved by the invention]

[0004] However, the waveform of tire vibrations over time while the vehicle is moving contains noise influenced by factors other than the road surface conditions, making it difficult to accurately determine the road surface conditions by reading only the influence of the road surface conditions. Therefore, an object of the present invention is to provide a method for creating a judgment space for determining road surface conditions, a judgment space updating method, a road surface condition determination method, a judgment space creating device, and a road surface condition determination device, which are capable of determining road surface conditions with high accuracy. [Means for solving the problem]

[0005] The present invention has the following configuration as a means for solving the above-mentioned problems. A method for creating a judgment space for determining road surface conditions from output waveform data, using information on the road surface conditions of a road surface on which a vehicle is traveling and output waveform data measuring changes in physical quantities of tires of a traveling vehicle, the method comprising the steps of: acquiring the output waveform data for a plurality of different road surface conditions and the driving conditions at the time of measuring the output waveform data; classifying the output waveform data into a plurality of groups based on the driving conditions; and performing machine learning for each group using teacher data in which information on the road surface conditions is linked to the output waveform data, thereby creating the judgment space corresponding to each group. The output waveform data is classified into multiple groups based on the driving conditions, and machine learning is performed for each group to create a corresponding judgment space, resulting in a judgment space that can accurately judge the road surface conditions.

[0006] The output waveform data may be normalized for each group, and the teacher data may be generated by linking the normalized output waveform data with information on the road surface condition. By normalizing the output waveform data for each group, the influence of factors other than the driving conditions used to classify the output waveform data can be suppressed, and the output waveform data belonging to the group can be easily compared.

[0007] The driving conditions may be the driving speed of the vehicle and / or the amount of wear of the tires when the output waveform data was measured, and the groups may be groups classified based on the driving speed and / or the amount of wear of the tires. Since the driving speed and / or wear amount are driving conditions that have a large effect on the output waveform data, by classifying the output waveform data using one or both of these as indicators, a judgment space that can accurately judge the road surface condition can be created.

[0008] Instead of performing machine learning for each group using teacher data in which information on the road surface condition is linked to the output waveform data belonging to that group, machine learning may be performed using first teacher data in which the road surface condition is linked to the output waveform data, and for each group, important areas in the output waveform data that are likely to differ depending on the road surface condition may be determined, important portion waveform data that is the output waveform data of the important areas is extracted from the output waveform data, and the judgment space corresponding to each group may be created by performing machine learning for each group using second teacher data in which the important portion waveform data is linked to information on the road surface condition. An important region is determined for each group, and second teacher data in which important portion waveform data in the important region is linked to the road surface condition information is used for machine learning, thereby making it possible to efficiently create a judgment space.

[0009] The output waveform data may be the output of a piezoelectric sensor attached to an inner surface of the tire. By using the output of the piezoelectric sensor as output waveform data, it becomes possible to calculate driving conditions such as the state of tire wear and the vehicle's driving speed from the output waveform data.

[0010] A method for judging a road surface condition from a measured value of output waveform data measuring a change in a physical quantity of a tire of a traveling vehicle, using one of a plurality of judgment spaces created by the above-mentioned method for creating a judgment space, characterized in that the road surface condition judgment method obtains judgment target data, which is the output waveform data measured to judge the road surface condition, and judgment driving conditions, which are the driving conditions at the time the judgment target data was measured, selects all of the judgment spaces that include the acquired judgment driving conditions as applicable judgment spaces from the plurality of judgment spaces, applies the applicable judgment space to the judgment target data, and judges the road surface condition of the road surface during traveling in the applicable judgment space based on the road surface condition to which the judgment target data is linked. The road surface condition can be determined with high accuracy by applying the judgment space created for each group classified according to the driving conditions to the judgment target data.

[0011] A method for updating a judgment space, characterized in that new teacher data is created by linking the judgment target data acquired in the above-mentioned road surface condition judgment method with information on the road surface condition judged by the road surface condition judgment method, and the judgment space judged to include the judgment target data is updated using the new teacher data. By updating the judgment space using the result of the road surface condition judgment, the accuracy of judging the road surface condition is improved.

[0012] 1. A judgment space creation device that uses information on road surface conditions on a road surface on which a vehicle is traveling and output waveform data measuring changes in physical quantities of tires of a traveling vehicle to create a judgment space for determining road surface conditions from the output waveform data, the judgment space creation device comprising: a classification processing unit that acquires the output waveform data for a plurality of different road surface conditions and the driving conditions at the time of measuring the output waveform data, and classifies the output waveform data into a plurality of groups based on the driving conditions; and a judgment space creation unit that performs machine learning for each group using teacher data in which information on the road surface conditions is linked to the output waveform data, and creates the judgment space corresponding to each group. By classifying the output waveform data into a plurality of groups according to the driving conditions and creating a decision space for each group, a decision space with high accuracy in determining the road surface condition can be created.

[0013] The judgment space creation device may include, instead of the judgment space creation unit, an important area determination unit that performs machine learning using first teacher data in which the road surface condition is linked to the output waveform data and determines, for each group, an important area in the output waveform data that is likely to differ depending on the road surface condition, an important part extraction unit that extracts important part waveform data that is the output waveform data of the important area from the output waveform data, and a second judgment space creation unit that creates the judgment space for each group by performing machine learning for each group using second teacher data in which the important part waveform data is linked to information on the road surface condition. By determining important areas for each group and using the second teacher data in which the important portion waveform data is linked to information on road surface conditions for machine learning, a judgment space can be efficiently created using data from portions of the output waveform data that make it easier to distinguish differences in road surface conditions.

[0014] A road surface condition judgment device that judges the road surface condition from measured values ​​of output waveform data that measures changes in physical quantities of tires of a traveling vehicle, the road surface condition judgment device comprising: a measuring device that measures the changes in the physical quantities of the tires; a recording unit that stores a plurality of judgment spaces created by the judgment space creation method of the present invention; a judgment space selection unit that acquires judgment target data, which is the output waveform data measured in order to judge the road surface condition, and judgment driving conditions, which are the driving conditions at the time the judgment target data was measured, and selects, from the plurality of judgment spaces, all of the judgment spaces that include the acquired judgment driving conditions for judgment as applicable judgment spaces; and a judgment unit that judges whether or not the judgment target data is included for each applicable judgment space, and judges the road surface condition of the road surface during traveling from information on the road surface condition linked to the applicable judgment space that is determined to include the judgment target data. The determination unit can, for example, obtain the driving conditions from the vehicle device or obtain the driving conditions by calculating them from the output waveform data. By creating multiple judgment spaces, selecting a judgment space that includes the driving conditions for judgment as the applicable judgment space, and judging the road surface condition during driving from the road surface condition information linked to the applicable judgment space, the road surface condition during driving can be judged with high accuracy.

[0015] The physical quantities of the tire may include the vehicle's traveling speed at the time of measurement and the degree of tire wear, the measuring device may be a piezoelectric sensor attached to an inner surface of the tire and output time series data of the tire deformation speed as the output waveform data, and the determination unit may calculate the vehicle's traveling speed from the periodicity of the output waveform data, and calculate the degree of tire wear from the magnitude of the output value. By using a piezoelectric sensor as a measuring device, it is possible to accurately measure the change in physical quantity of a tire over time. In addition, the driving speed and degree of tire wear can be calculated from the output waveform data from the piezoelectric sensor, which makes it possible to simplify the road surface condition judgment device.

[0016] The present invention may further include an update unit capable of updating the judgment space stored in the recording unit, wherein the update unit generates additional teacher data, which is new teacher data linking the output waveform data used to determine the road surface condition, the judgment result of the road surface condition by the judgment, and the driving conditions at the time the road surface condition was judged, creates an update group by adding the additional teacher data to a group of teacher data used to create the judgment space used as the applicable judgment space during the judgment, and performs machine learning on the update group to update the judgment space. In this case, a validity evaluation unit may be provided which obtains information on whether the road surface condition judged by the judgment unit is valid or not, and if valid, sets the information on the road surface condition judged by the judgment unit as information on the road surface condition corresponding to the measured value. By creating an update group and performing machine learning on the update group to update the judgment space, a judgment space with good judgment accuracy of the road surface condition is obtained. In addition, by evaluating the validity of the road surface condition determined by the validity evaluation unit, appropriate teacher data can be added and machine learning can be performed to improve the judgment accuracy of the judgment space. Effect of the Invention

[0017] By performing machine learning for each group classified based on the driving conditions to create a corresponding judgment space, the judgment space can be used to accurately judge the road surface condition. Therefore, the judgment space can be applied to the judgment target data to accurately judge the road surface condition. [Brief description of the drawings]

[0018] [Figure 1] 1 is a flowchart of a method for creating a decision space according to the first embodiment. [Diagram 2] Functional block diagram of a decision space creation device according to the first embodiment. [Diagram 3] Functional block diagram of a decision space creation device according to a second embodiment. [Figure 4] 1 is a flowchart of a method for creating a decision space according to a second embodiment. [Diagram 5] Graph showing schematic diagram of important areas in output waveform data [Figure 6] Functional block diagram of a road surface condition determination device [Figure 7] 1 is a flowchart of a method for determining road surface conditions according to a third embodiment. [Figure 8] 1 is a flowchart of a method for updating a decision space according to a fourth embodiment. BEST MODE FOR CARRYING OUT THEINVENTION

[0019] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will now be described with reference to the accompanying drawings, in which the same components are designated by the same reference numerals throughout the drawings, and the description thereof will be omitted where appropriate.

[0020] <Embodiment 1> As shown in FIG. 2, the judgment space creation device 10 of this embodiment is a device that uses information on the road surface condition on which a vehicle is traveling and output waveform data obtained by measuring changes in physical quantities of tires of a traveling vehicle, and creates a judgment space for determining the road surface condition from the output waveform data. The decision space creation device 10 includes a measurement device 11 and a machine learning device 12 .

[0021] The measuring device 11 measures changes in physical quantities of tires of a vehicle, and measures output waveform data under a plurality of different road surface conditions. For example, a piezoelectric sensor or an acceleration sensor can be used as the measuring device 11. Examples of piezoelectric sensors include film-like composite piezoelectric elements using potassium niobate, potassium sodium niobate, barium titanate, or lead zirconate titanate powder, and polymeric piezoelectric elements such as PVDF and PVDF-TrFE.

[0022] The machine learning device 12 includes a recording device 13, a calculation unit 14, and a judgment space creation unit 15, and creates a judgment space based on the output waveform data acquired by the measurement device 11 and information on the road surface state and the driving conditions when the output waveform data was measured. Information on the driving conditions and road surface state is acquired from equipment or other input devices provided in the vehicle (not shown), or is calculated and acquired from the output of the measurement device 11, and stored in the recording device 13.

[0023] The recording device 13 stores the output waveform data acquired by the measuring device 11, information on road surface conditions, and driving conditions in association with each other, and is implemented by a memory such as a RAM. The calculation unit 14 classifies the output waveform data into a plurality of groups based on the information on the road surface condition and the driving conditions. The driving conditions refer to elements such as the vehicle and the environment outside the vehicle while driving. The vehicle includes the vehicle body and components attached to the vehicle body such as tires. The driving conditions used for classification include the driving speed, the degree of tire wear, and the temperature. These are classified according to a predetermined numerical range. One driving condition or a combination of multiple driving conditions is used to classify the output waveform data.

[0024] The decision space creation unit 15 performs machine learning for each group using teacher data in which the output waveform data is linked to information on the road surface condition, and creates a decision space corresponding to each group. The decision space creation unit 15 is configured as computer hardware, software (program), etc.

[0025] 1 is a flowchart of the method for creating a decision space according to the present embodiment. The method for creating a decision space according to the present embodiment uses information on the road surface condition on which a vehicle is traveling and output waveform data obtained by measuring changes in the physical quantities of the tires of a traveling vehicle, and creates a decision space for determining the road surface condition from the output waveform data.

[0026] The judgment space creation device 10 acquires output waveform data for each of a plurality of different road surface conditions (DRY, WET, ICE, etc.) and the driving conditions at the time of measuring the output waveform data (S11). The judgment space creation device 10 acquires driving conditions such as the vehicle's driving speed and the degree of tire wear from the measurement device 11 or other devices. The output waveform data acquired in S11 is used for processing by the calculation unit 14 after being recorded in the recording device 13. Note that the output waveform data may be used for processing by the calculation unit 14 without being recorded in the recording device 13.

[0027] The calculation unit 14 classifies the output waveform data into a plurality of groups based on the vehicle driving conditions acquired in S11 (S12). Then, the calculation unit 14 links the output waveform data to group information (S13), links the output waveform data linked to the group information to information on the road surface condition (road surface information data), and creates teacher data (S14). In the above explanation of the flowchart, the output waveform data is linked to the information on the road surface condition in S14, but the linking may be performed at any stage from S11 to S13. Since the measurement is performed in S11 when the road surface condition is known and the information on the road surface condition is clear at the stage of S11, the above linking can be performed at any stage from S11 to S13.

[0028] The judgment space creation unit 15 performs machine learning for each group using the teacher data (S15). Then, a judgment space corresponding to each group is created (S16). By creating a judgment space corresponding to each group, that is, by creating a plurality of judgment spaces corresponding to the number of groups, a judgment space with good accuracy in judging the road surface condition can be created. When the judgment space is created by classifying the road surface conditions into groups including the plurality of road surface conditions using the driving conditions, each judgment space is classified into a space corresponding to the plurality of road surface conditions to be judged. S15 and S16 are the so-called "learning" phase of machine learning, and the judgment space creation unit 15 creates a "judgment line" for classifying the road surface condition using each judgment space classified using the driving conditions.

[0029] The output waveform data of S11 is preferably output from a piezoelectric sensor attached to the inner surface of the tire. The output waveform of the piezoelectric sensor has a portion where the output value differs between a new tire and a tire with advanced wear. Because the output waveform of the piezoelectric sensor has such characteristics, the wear state of the tire can be estimated based on the output waveform data of the piezoelectric sensor.

[0030] In addition, the output of the piezoelectric sensor increases when the surface of the tire opposite to the part where the sensor is attached comes into contact with the ground. In other words, if the vehicle is traveling at a constant speed for a certain period of time, the output value periodically increases, so the traveling speed of the vehicle can be calculated from the tire size and the period in which the output increases. When the judgment space creation unit 15 estimates the wear state, information such as the traveling speed may be obtained from vehicle equipment. On a sunny day when the road surface is dry (DRY), the wear state of the tire can be accurately estimated based on the output waveform data of the piezoelectric sensor.

[0031] The driving conditions in S11 include, for example, the vehicle's running speed and / or the amount of tire wear when the output waveform data is measured. When these are acquired as the driving conditions, machine learning in S15 and generation of a judgment space in S16 are performed for each group classified based on either the vehicle's running speed or the amount of tire wear, or the vehicle's running speed and the amount of tire wear. The running speed and / or the amount of wear, which are the driving conditions on the vehicle side, are factors that have a large effect on the output waveform data. Therefore, by using one or both of these as an index for classifying the output waveform data into groups, a judgment space capable of accurately determining the road surface condition can be created.

[0032] The following table shows a model of how to create a judgment space. In creating the judgment space, output waveform data is measured in each range classified by a predetermined degree of wear and running speed when the road surface condition is known in advance. Table 1 shows an example in which output waveform data is measured under predetermined running conditions for each road surface condition, with the degree of wear being 0 to 50% and below as low degree of wear, the degree of tire wear being over 50% and below 100% as high degree of wear, the range of the running speed of the vehicle being over 0 km / h and below 40 km / h as low running speed, and the range of the running speed being over 40 km / h and below 100 km / h as high running speed. In Table 1, the road surface condition is classified into two types, DRY (dry state) and WET (wet state), but the classification of the road surface condition is not limited to this. For example, WET may be divided into two or more levels depending on the wetness of the road surface, or ICE (frozen state) may be added. In addition, the degree of wear and the running speed as running conditions may each be classified into three or more ranges. [Table 1]

[0033] As shown in Table 1, output waveform data (1) DSL1, DSL2, ... DSLn are measured under driving conditions of low wear and low driving speed on a DRY road surface condition. Similarly, output waveform data (2) to (8) are measured for other combinations of measurement conditions on a DRY road surface condition and for each driving condition on a WET road surface condition.

[0034] The output waveform data measured as described above is divided into multiple groups based on the driving conditions, and machine learning is performed for each group to create a decision space for determining whether the road surface condition is DRY or WET.

[0035] For example, when classifying the output waveform data into groups containing a plurality of road surface conditions using driving conditions, the output waveform data is divided into two groups, (1), (2), (5), and (6), and (3), (4), (7), and (8), when classifying based on the degree of wear only. The output waveform data is divided into two groups, (1), (3), (5), and (7), and (2), (4), (6), and (8), when classifying based on the driving speed. The output waveform data is divided into four groups, (1) and (5), (2) and (6), (3) and (7), and (4) and (8), when classifying based on the degree of wear and the driving speed.

[0036] In the method of creating a judgment space in this embodiment, instead of creating a judgment space using all of the acquired output waveform data as it is as teacher data, the acquired output waveform data is classified based on the driving conditions, and machine learning is performed for each classified group to create a judgment space, as shown in Fig. 1. This results in a judgment space with good judgment accuracy in which the influence of noise caused by the driving conditions on the output waveform data is suppressed.

[0037] For example, in the case of classifying into groups based on wear and running speed, when data with high wear and low running speed are obtained as the data to be judged, first, from the four types of judgment spaces, a group consisting of (3) and (7) with high wear and low running speed is selected. By making a judgment using the judgment space corresponding to the running conditions when the data to be judged was measured, the road surface condition can be judged with high accuracy.

[0038] <Embodiment 2> The method of creating a judgment space in this embodiment differs from the method of creating a judgment space in embodiment 1 in that the output waveform data is normalized for each group, and the teacher data is generated by linking the normalized output waveform data to information on the road surface condition, and instead of performing machine learning for each group using teacher data in which information on the road surface condition is linked to the output waveform data belonging to that group, machine learning is performed using first teacher data in which the road surface condition is linked to the output waveform data, and for each group, important areas in the output waveform data that are likely to differ depending on the road surface condition are determined, important portion waveform data that is the output waveform data of the important area is extracted from the output waveform data, and machine learning is performed for each group using second teacher data in which the important portion waveform data is linked to information on the road surface condition, thereby creating a judgment space corresponding to each group.

[0039] 3 is a functional block diagram of the judgment space creating device 20 of this embodiment. The judgment space creating device 20 differs from the judgment space creating device 10 in that the judgment space creating unit 25 of the machine learning device 22 in the judgment space creating device 20 includes an important region determining unit 251, an important part extracting unit 252, and a second judgment space creating unit 253. The function of each unit will be described with reference to the flowchart of FIG.

[0040] FIG. 4 is a flowchart of a method for creating a decision space according to this embodiment. The calculation unit 14 acquires a group of output waveform data belonging to a predetermined group (S21). Then, each output waveform data in the output waveform data group acquired in S21 is normalized for each group (S22). The normalization of each output waveform data converts the output waveform data so that the maximum value in each output waveform data is set to 1, the minimum value is set to 0, and the values ​​in between are proportionally distributed. The output waveform data group is acquired from the measurement device 11 or the recording device 13.

[0041] The judgment space creation unit 25 acquires road surface information data for each normalized output waveform data (S23), generates first teacher data in which the normalized output waveform data is linked to information on the road surface condition (S24), and performs machine learning using the first teacher data created in S24 (S25).

[0042] The judgment space creation unit 25 uses the important area determination unit 251 to determine, for each group, an important area, which is an area (part, range) in the output waveform data that is likely to differ depending on the road surface condition, based on the results of machine learning, and the important part extraction unit 252 extracts important part waveform data, which is the output waveform data of the important area, from the output waveform data (S26).

[0043] FIG. 5 is a graph showing a schematic diagram of an important region in the output waveform data. The figure shows the difference and importance of the output waveform data depending on the road surface condition (DRY, WET). The importance indicates the degree of influence that the road surface condition has on the output waveform data. The important region determination unit 251 determines the important region in the output waveform data based on the importance, and the important portion extraction unit 252 extracts the important portion waveform data. That is, the determination space creation unit 25 creates an important region extractor having a function of extracting an important region in a group, and applies the important region extractor to the output waveform data to extract the important portion waveform data.

[0044] The second judgment space creation unit 253 recreates second teacher data that links the important part waveform data extracted by the important part extraction unit 252 with information on the road surface condition (road surface information data) (S27). Then, machine learning using the second teacher data is performed for each group (S28), and a judgment space corresponding to each group is created (S29).

[0045] As described above, the method of creating a judgment space according to the present embodiment applies machine learning to data of important regions extracted from output waveform data normalized for each group, and creates a nonlinear space (judgment space) used to judge road surface conditions. Normalizing the output waveform data makes it easier to compare relative changes in waveform, rather than absolute values. For example, normalization makes it easier to compare a plurality of output waveform data measured under different tire temperature conditions. Then, waveform data of a portion of the output waveform data where differences in road surface conditions are easier to discern is extracted, and machine learning is performed using this important portion waveform data. This allows the judgment space to be created efficiently. In addition, using the important portion waveform data for machine learning makes it possible to prevent the judgment space from being created as a locally optimal portion.

[0046] <Embodiment 3> FIG. 6 is a functional block diagram of the road surface condition determining device 30 of the present embodiment. The road surface condition determination device 30 is a device that determines the road surface condition from measured values ​​of output waveform data that measures changes in physical quantities of tires of a traveling vehicle, and is equipped with a measurement device 11 and a machine learning device 32.

[0047] The machine learning device 32 includes a recording device 33 , a calculation unit 34 , a judgment unit 35 , vehicle equipment 36 , and a judgment space update unit 37 . The recording device 33 stores a plurality of decision spaces created by the decision space creating method of the present invention. The calculation unit 34 acquires the judgment target data, which is output waveform data measured to judge the road surface condition, and the judgment driving conditions, which are the driving conditions at the time the judgment target data was measured, and selects, from among the multiple judgment spaces, all judgment spaces that include the acquired judgment driving conditions as applicable judgment spaces. The judgment unit 35 judges whether or not each applicable judgment space contains judgment target data, and judges the road surface condition of the road surface during traveling from the road surface condition information linked to the applicable judgment space that is judged to contain the judgment target data.

[0048] The determination unit 35 can obtain the driving conditions from the vehicle device 36, or can obtain them by calculating them from the output waveform data of the recording device 33. For example, the vehicle driving speed is obtained from the vehicle device 36, and information on tire deformation over time is obtained from the output of a piezoelectric sensor provided on the inner surface of the tire as the measuring device 11. The vehicle device 36 is, for example, an on-board meter that measures the vehicle driving speed, or an on-board or portable device that can connect to the Internet and obtain various information.

[0049] When the physical quantities of the tire are the vehicle's traveling speed and the degree of tire wear at the time of measuring the physical quantities by the measuring device 11, it is preferable to use a piezoelectric sensor attached to the inner surface of the tire as the measuring device 11. The determining unit 35 can use the time series data of the tire's deformation speed outputted as output waveform data by the piezoelectric sensor, calculate the vehicle's traveling speed from the periodicity of the output waveform data, and calculate the degree of tire wear from the magnitude of the output value. Therefore, in this case, the vehicle's traveling speed and the degree of tire wear can be obtained based on the output from the measuring device 11 without using the vehicle equipment 36.

[0050] The judgment space update unit 37 is capable of updating the judgment space stored in the recording device 33, and is composed of computer hardware and software (programs). The judgment space update unit 37 generates additional teacher data, which is new teacher data that links the judgment target data, which is output waveform data used to judge the road surface condition, the judgment result of the road surface condition by the judgment, and the driving conditions when the road surface condition was judged. An update group is created by adding the additional teacher data to a group of teacher data linked to a driving condition similar to the driving condition linked to the additional teacher data. The group of teacher data to which the additional teacher data is added is the one used to create the judgment space used to judge the road surface condition based on the judgment target data when the additional teacher data was generated. In this way, the additional teacher data is added to the teacher data linked to the similar driving conditions. Then, machine learning is performed on the update group to update the judgment space.

[0051] FIG. 7 is a flowchart showing the road surface condition determination method of this embodiment. The road surface condition determination method of this embodiment uses any one of a plurality of judgment spaces created by the judgment space creation method described in the above-mentioned embodiments 1 and 2 to determine the road surface condition from the measured values ​​of output waveform data that measures changes in physical quantities of tires of a traveling vehicle.

[0052] The calculation unit 34 acquires the judgment target data and the judgment driving conditions, which are the driving conditions when the judgment target data was measured (S31). The judgment target data of S31 is the output waveform data measured by the measuring device 11 to judge the road surface condition, and the judgment driving conditions of S31 are the driving conditions when the judgment target data was measured.

[0053] The determination unit 35 selects all determination spaces including the acquired determination driving conditions as applicable determination spaces from among the multiple determination spaces created by the determination space creation method (S32). Then, the application determination space selected in S32 is applied to the determination target data (S33). In the application determination space applied in S33, the road surface condition of the road surface during travel is determined based on the road surface condition to which the determination target data is linked (S34). Then, the road surface condition estimation value determined in S34 is output (S35). The output road surface condition estimation value is notified to, for example, a display device on the vehicle side or a mobile information terminal such as a smartphone other than the vehicle.

[0054] When creating a judgment space by classifying the road surface conditions into groups containing multiple road surface conditions, the applicable judgment space selected in S32 is applied to the data to be judged (S33), and the road surface condition is judged based on which area in the judgment space the data to be judged is included in (S34).

[0055] <Embodiment 4> The update method of this embodiment creates new teacher data by linking the judgment target data obtained in the road surface condition judgment method of embodiment 3 with information on the road surface condition judged by the road surface condition judgment method, and updates the judgment space judged to include the judgment target data using the new teacher data.

[0056] FIG. 8 is a flowchart of the method for updating the judgment space according to this embodiment. As shown in the figure, first, it is determined whether or not the estimated value acquired in the road surface condition judgment method is valid (S41). For example, when the judgment target data is acquired and the road surface condition is estimated in a state where the road surface condition is known in advance, an evaluator can judge the validity of the road surface condition estimated value. Alternatively, the judgment space update unit 37 in the road surface condition judgment device 30 may make the judgment. In this case, the judgment space update unit 37 may acquire the external conditions (temperature, etc.) at the time of judgment from the vehicle equipment 36, and make a judgment by comparing the acquired external conditions with the estimated value.

[0057] If the data is evaluated as valid in S41 (YES), the decision space update unit 37 associates the road surface condition estimation value with the decision target data to create additional second teacher data (S42). Then, machine learning is performed with the additional second teacher data (S43). Based on the result of the machine learning in S43, an updated decision space is generated (S44). The generated updated decision space is recorded in the recording device 33.

[0058] If it is evaluated as inappropriate in S41 (NO), the judgment space update unit 37 judges whether or not appropriate road surface information data can be obtained (S45). For example, if the evaluator can input appropriate road surface conditions (YES in S45), the appropriate road surface information data is linked to the acquired data to generate additional second teacher data (S46). Then, the machine learning (S43) described above is performed to generate an updated judgment space (S44). If the evaluator is unable to input a valid road surface condition (NO in S45), it is determined that the updated decision space is not to be updated (S47).

[0059] In the road surface condition determination device 30, the recording device 33, the calculation unit 34, the determination unit 35, and the vehicle equipment 36 are the areas used to determine the road surface condition to which the determination target data corresponds. In contrast, the recording device 33, the vehicle equipment 36, and the determination space update unit 37 are the areas that use the determination target data to update the teacher data and learn (update) using the teacher data.

[0060] As in the updating method of this embodiment, the measured data to be judged is fed back to update the teacher data, and the teacher data is used for learning, thereby improving the judgment accuracy of the judgment space. EXAMPLES

[0061] [Example] <Creating a decision space> A vehicle was driven on a test course under the following road surface conditions and driving conditions, and output waveform data, which is time-series data on the deformation speed of the tire, was measured using a piezoelectric sensor attached to the inner surface of the tire. Based on the vehicle speed and tire wear when each output waveform data was measured, the output waveform data was divided into six groups, each of which included multiple road surface conditions, and the output waveform data was normalized for each group to create a decision space. In addition, important portion waveform data was extracted from each output waveform data, and based on the vehicle speed and tire wear, the output waveform data was divided into six groups, each of which included multiple road surface conditions, and the important portion waveform data was normalized for each group to create a decision space.

[0062] Road Condition DRY: Dry road surface (reproduces the road surface on a sunny day) WET0: Wet but no running water (reproduces the road surface after rain) WET1: Road surface with low water flow (reproduces the road surface in rainy weather) WET2: Road surface with large amount of water flow (Reproduces the road surface during heavy rain) Driving conditions Wear level: 0, 50, 100% Vehicle speed: 31~40kpH(km / h), 45~60kpH(km / h)

[0063] The road surface conditions WET0, WET1, and WET2 correspond to the three-stage water volume switch button for controlling the sprinkler system, which can be set when driving on the test course. WET1 is equivalent to road surface conditions during light rain with an hourly rainfall of about 10 mm, while WET2 is estimated to correspond to road surface conditions during heavy rain with an hourly rainfall of more than 10 mm, up to about 50 mm.

[0064] <Determination of road surface conditions based on data to be determined> (Grouping) The data to be judged of a vehicle traveling on a test course was measured, the driving conditions for judgment at that time were obtained, and from the judgment spaces created and divided into six groups, the judgment space containing the driving conditions for judgment was selected as the applicable judgment space, and the road surface condition was judged based on the data to be judged.

[0065] (Important part) Instead of the output waveform data, important portion waveform data extracted from the output waveform data was used, and the road surface condition was judged based on the important portion waveform data in the same manner as for the judgment target data described above.

[0066] <Comparative example, as is> The same output waveform data as in the embodiment was not grouped according to the driving conditions, but was used as is to create a judgment space, and the road surface condition was judged based on the same judgment target data as in the embodiment.

[0067] <Result> The results for the Examples (grouped, critical parts) and the Comparative Examples (as is) are shown in the table below. [Table 2] [Table 3] [Table 4]

[0068] In the examples, in order to show the effect of grouping, a classification method using a linear function (support vector machine (linear kernel)) was used. If a different method such as a neural network is used for the classification method, the accuracy rate will vary. For this reason, it is important that the accuracy rate is improved by grouping.

[0069] From Tables 2 and 4, it was found that the accuracy rate of road surface conditions was improved by using a judgment space grouped based on driving conditions. The working example shown in Table 2 was able to judge the WET classification to a certain extent. In contrast, the comparative example shown in Table 4 was unable to judge WET0 and WET2 at all, and it can be said that they were simply put into DRY, which has a large amount of data.

[0070] As shown in Tables 2 and 3, when important portion waveform data was used, the same judgment results were obtained as when output waveform data was used. This shows that even when important portion waveform data is used to reduce the amount of calculations, the road surface condition can be judged with high accuracy by using the grouped judgment space. The use of important portion waveform data may contribute to improving the generalization performance of judging road surface conditions with high accuracy on places other than test courses such as public roads.

[0071] <Application example for tire wear detection> A decision space used for purposes other than judging (determining) the road surface condition may be created using the method for creating a decision space according to the embodiment 1. For example, the output waveform data may be classified into a plurality of groups based on the driving conditions, and machine learning may be performed for each group using teacher data in which the output waveform data is linked with information on the tire wear state, to create a decision space for judging the tire wear state for each group.

[0072] An example of a tire wear state determination method for determining (discriminating) the tire wear state from output waveform data, which is data to be determined, using a determination space for determining the tire wear state will be described below. On a clear, sunny day or other such condition when the road surface is clearly dry and information on the driving speed is available from the vehicle's gauges, measurements are taken with a measuring device to obtain data to be judged. All judgment spaces that include the judgment driving conditions for judgment of the judgment target data (here, road surface condition and traveling speed) are regarded as applicable judgment spaces, and it is determined which applicable judgment space the judgment target data is included in. The tire wear state is judged from the information on the tire wear state in the applicable judgment space selected by the above judgment. By dividing the tire wear state into smaller sections when creating the judgment space, it is possible to judge the wear state in more detail. For example, when the road surface is clearly dry and the driver of the vehicle presses the judgment start button, the wear condition judgment method of this application example can be implemented to judge the wear condition. [Industrial Applicability]

[0073] INDUSTRIAL APPLICABILITY The present invention is useful as an apparatus or method for judging road surface conditions based on the results of measuring the condition of tires while a vehicle is running. [Explanation of symbols]

[0074] 10: Judgment space creation device 11: Measuring equipment 12: Machine learning device 13: Recording device 14: Calculation unit (classification processing unit) 15: Judgment space creation section 20: Judgment space creation device 22: Machine learning device 25: Judgment space creation section 251: Important area determination section 252: Important part extraction part 253: Second judgment space creation unit 30: Road surface condition determination device 32: Machine learning device 33: Recording device (recording section) 34: Calculation unit (classification processing unit, judgment space selection unit) 35: Judgment section 36: Vehicle equipment 37: Decision space update unit (validity evaluation unit)

Claims

1. A method for creating a judgment space for judging the road surface condition from output waveform data by using information on the road surface condition on which a vehicle is traveling and output waveform data obtained by measuring a change in a physical quantity of a tire of a traveling vehicle, the method comprising the steps of: Acquire the output waveform data for a plurality of different road surface conditions and the driving conditions at the time of measuring the output waveform data; The output waveform data is classified into a plurality of groups based on the driving conditions. performing machine learning for each of the groups using teacher data in which the output waveform data is linked to information on the road surface condition, thereby creating the judgment space corresponding to each of the groups; the driving condition is a driving speed of the vehicle and / or an amount of wear of the tires when the output waveform data is measured, The groups are classified based on the traveling speed and / or the amount of wear of the tires. How to create a decision space.

2. normalizing the output waveform data for each group; The teacher data is generated by linking the normalized output waveform data with information on the road surface condition. The method for creating a judgment space according to claim 1 .

3. Instead of performing machine learning for each group using teacher data in which information on the road surface condition is linked to the output waveform data belonging to the group, performing machine learning using first teacher data in which the output waveform data is associated with the road surface condition, and determining, for each group, an important area in the output waveform data that is likely to differ depending on the road surface condition; extracting important portion waveform data, which is the output waveform data of the important region, from the output waveform data; performing machine learning for each of the groups using second teacher data in which the important portion waveform data is linked to information on the road surface condition, thereby creating the judgment space corresponding to each of the groups; The method for creating a judgment space according to claim 1 or 2.

4. The output waveform data is an output of a piezoelectric sensor attached to an inner surface of the tire. The method for creating a judgment space according to claim 1 or 2.

5. A method for determining a road surface condition from measured values ​​of output waveform data obtained by measuring a change in a physical quantity of a tire of a running vehicle, using any one of a plurality of decision spaces created by the method for creating a decision space according to any one of claims 1 to 4, comprising the steps of: Obtaining judgment target data, which is the output waveform data measured to judge a road surface condition, and judgment driving conditions, which are driving conditions at the time of measuring the judgment target data; Selecting all of the judgment spaces that include the acquired judgment driving conditions as application judgment spaces from among the plurality of judgment spaces; applying the applied decision space to the decision target data; In the application judgment space, the road surface condition of the road surface during traveling is judged based on the road surface condition to which the judgment target data is linked. Method for determining road surface conditions.

6. A method for determining road surface conditions according to the present invention, comprising the steps of: creating new teacher data by linking the data to be determined obtained using the road surface condition determination method; and updating the determination space determined to include the data to be determined using the new teacher data. How to update the decision space.

7. A judgment space creating device that uses information on road surface conditions on a road surface on which a vehicle is traveling and output waveform data obtained by measuring changes in physical quantities of tires of a traveling vehicle, and creates a judgment space for determining road surface conditions from the output waveform data, comprising: a classification processing unit that acquires the output waveform data in a plurality of different road surface conditions and the driving conditions at the time of measuring the output waveform data, and classifies the output waveform data into a plurality of groups based on the driving conditions; an important region determination unit that performs machine learning using first teacher data in which the output waveform data is associated with the road surface condition, and determines, for each group, an important region in the output waveform data that is likely to differ depending on the road surface condition; an important portion extraction unit that extracts important portion waveform data, which is the output waveform data of the important region, from the output waveform data; a second judgment space creation unit that creates the judgment space for each of the groups by performing machine learning for each of the groups using second teacher data in which the important portion waveform data is linked to information on the road surface condition; The present invention is characterized in that it comprises Judgment space creation device.

8. A road surface condition determination device that determines a road surface condition from measured values ​​of output waveform data obtained by measuring changes in physical quantities of tires of a traveling vehicle, comprising: A measuring device for measuring a change in a physical quantity of the tire; a recording unit for storing a plurality of decision spaces created by the decision space creating method according to any one of claims 1 to 4; a judgment space selection unit that acquires judgment target data, which is the output waveform data measured to judge a road surface condition, and judgment driving conditions, which are driving conditions at the time of measuring the judgment target data, and selects, from among the multiple judgment spaces, all of the judgment spaces that include the acquired judgment driving conditions as application judgment spaces; and and a determination unit that determines whether the judgment target data is included for each of the application judgment spaces, and determines the road surface condition of the road surface during traveling from information on the road surface condition associated with the application judgment space that is determined to include the judgment target data. Road surface condition determination device.

9. The determination unit acquires the driving conditions from the measurement device or calculates and acquires the driving conditions from the output waveform data. The road surface condition determination device according to claim 8.

10. the physical quantities of the tires include a vehicle running speed and a tire wear degree at the time of measurement, the measuring device is a piezoelectric sensor attached to an inner surface of the tire, and outputs time-series data of the deformation speed of the tire as the output waveform data; the determination unit calculates a running speed of the vehicle from the periodicity of the output waveform data, and calculates a degree of wear of the tire from the magnitude of the output value. The road surface condition determination device according to claim 8.

11. an update unit capable of updating the judgment space stored in the recording unit, The update unit is generating additional teacher data that is new teacher data linking the output waveform data used in the determination of the road surface condition, the determination result of the road surface condition by the determination, and the driving conditions when the road surface condition was determined; An update group is created by adding the additional teacher data to the group of teacher data used to create the judgment space used as the applicable judgment space at the time of the judgment, to which information similar to the information linked to the additional teacher data is linked; performing machine learning on the update group to update the decision space; The road surface condition determination device according to claim 8.

12. A road surface condition determination device for determining a road surface condition from measured values ​​of output waveform data obtained by measuring changes in physical quantities of tires of a traveling vehicle, comprising: A measuring device for measuring a change in a physical quantity of the tire; a recording unit for storing a plurality of decision spaces created by the first creation method; a judgment space selection unit that acquires judgment target data, which is the output waveform data measured to judge the road surface condition, and judgment driving conditions, which are driving conditions at the time of measuring the judgment target data, and selects, from among the multiple judgment spaces, all of the judgment spaces that include the acquired judgment driving conditions as application judgment spaces; and a determination unit that determines whether the judgment target data is included for each of the applicable judgment spaces, and determines the road surface condition of the road surface during traveling from information on the road surface condition associated with the applicable judgment space that is determined to include the judgment target data; an update unit capable of updating the judgment space stored in the recording unit; Equipped with The plurality of judgment spaces created by the first creation method are for determining the road surface condition from the output waveform data by using information on the road surface condition of the road surface on which the vehicle is traveling and the output waveform data obtained by measuring a change in a physical quantity of the tire of the vehicle while the vehicle is traveling, In the first production method, Acquiring the output waveform data for a plurality of different road surface conditions and the driving conditions at the time of measuring the output waveform data; The output waveform data is classified into a plurality of groups based on the driving conditions. By performing machine learning for each group using teacher data in which the output waveform data is linked to information on the road surface condition, creating a corresponding judgment space for each of the groups; The update unit is generating additional teacher data that is new teacher data linking the output waveform data used in the determination of the road surface condition, the determination result of the road surface condition by the determination, and the driving conditions when the road surface condition was determined; An update group is created by adding the additional teacher data to the group of teacher data used to create the judgment space used as the applicable judgment space at the time of the judgment, to which information similar to the information linked to the additional teacher data is linked; The method is characterized in that the decision space is updated by performing machine learning on the update group. Road surface condition determination device.

13. 13. The road surface condition judgment device according to claim 11 or 12, further comprising a validity evaluation unit that acquires information on whether the road surface condition judged by the judgment unit is valid, and, if valid, sets the information on the road surface condition judged by the judgment unit to information on the road surface condition corresponding to the measurement value.

Images