JP5097470B2 - Road surface condition determination method and apparatus - Google Patents

Description

  The present invention relates to a road surface condition determination method and apparatus for determining the road surface condition of a road on which a vehicle such as an automobile or a motorcycle travels.

  Conventionally, information on a road surface sensor that detects the road surface state of a road on which a vehicle such as an automobile or a motorcycle is running provides information promptly to operators, drivers, and road managers of a snow removal station instead of visual information such as road patrol. This is one of the most promising road weather sensors in the future. However, since the road surface condition sensor has many devices with low accuracy stability, the road surface condition sensor is not sufficiently utilized as information for determining the snow removal start, and the start determination depends on visual information such as an ITV image or road patrol. This is the current situation.

  Since road surface information on winter roads is the most familiar information for local residents in snowy countries, the percentage of requests for “provision of detailed information (including road surface conditions)” for winter road information has become very high. . In particular, the provision of high-accuracy road weather information is useful not only for road management such as snow removal, but also for local residents in the future, so active development is expected.

  Conventionally, as a road surface condition determination method and apparatus therefor, as disclosed in Patent Document 1, for example, a sound sensor such as a microphone is used as a road surface condition sensor for detecting a road surface condition of a road on which a vehicle such as an automobile or a motorcycle travels. This sound sensor detects the contact sound between the tire of the vehicle traveling on the road and the road surface, and the detected contact sound is converted into a digital signal and input to the analysis unit. In this analysis unit, the actual frequency value and sound pressure level value are obtained from the digital data by spectral analysis, and the actual frequency value and sound with respect to the preset road surface condition value of the frequency value and sound pressure level value registered in the analysis unit are obtained. The pressure level value is compared and determined, and the comparison determination result is output via the output unit.

  In Patent Document 2, the sound recorded by the sound sensor is converted into digital recording data, and then subjected to Fourier transform in almost real time by a fast Fourier transformer (FFT) to obtain a frequency spectrum. It is judged from the distribution of the frequency spectrum whether the road surface condition is wet, dry, or compressed snow.

  By the way, the vehicle running sound is mainly the frictional sound between the road surface and the tire. When it is “dry”, it becomes the frictional sound of the road surface, air, and tire. High frequency sound is mainly used, and “pressure snow” is snow and tire friction sound. Therefore, since the road surface is concealed by snow, the characteristic frequency sound is changed and the low frequency sound is mainly used. From such a situation, the vehicle running sound is picked up by a sound sensor, and the road surface condition is determined by feature extraction by frequency analysis (frequency filter) processing.

  For example, FIG. 30 shows the result of digital recording of the vehicle running state, where the horizontal axis indicates time (minutes) and the vertical axis indicates intensity (sound amplitude). The portion where the amplitude rises greatly indicates the passage of the vehicle, and the portion indicated by an ellipse indicates the passage of a large vehicle and a small vehicle (ordinary vehicle) as a representative vehicle type vehicle. The sampling frequency in the example of FIG. 30 was 22.05 kHz, and the road surface condition was dry.

  Since the digital data as described above is prepared for each road surface condition, the frequency analysis for each vehicle type is performed.

  For example, FIG. 31 shows the result of applying Fourier transform to the traveling sound of the time zone (each elliptical part in FIG. 30) when the small car (ordinary car) and the large car traveled on the dry road surface and the wet road surface, respectively. is there. The number of digital data is 65536 (about 3 seconds). In addition, the horizontal axis of this graph indicates the frequency component of the sound, and the vertical axis indicates the intensity of each frequency component.

  From the above results, it can be seen that the intensity of high frequency sound of about 1 kHz or more is stronger on the wet road surface than on the dry road surface in both small cars (ordinary cars) and large cars. This means that, on a daily basis, the wet road surface has a higher frequency on the wet road surface than the dry road surface state, as the vehicle travel is represented by the “sher” sound on the wet road surface. Yes.

Therefore, it can be seen that the intensity of the high frequency component sound is strong on the wet road surface regardless of the vehicle type. Therefore, if a high frequency component sound is extracted by the frequency filter, the dry state and the wet state of each vehicle type can be distinguished.
JP 2005-156236 A JP 2002-256521 A

  By the way, in the conventional road surface condition determination method and its apparatus, it is inherent as a problem that a difference due to a vehicle type and a difference due to a road surface state cannot be determined by simple spectrum analysis.

  For example, in the case where the results of the dry state and the wet state in the small car and the large car are expressed on the same graph as shown in FIG. 31, the sound intensity of the large car is stronger than that of the small car, even if the frequency filter is higher. Even if the sound of the frequency is extracted, it becomes impossible to distinguish between dry and wet due to the difference in the vehicle type. That is, when the vehicle running sound itself is large as in a large vehicle, the strength of the running noise in the wet state of the small vehicle is exceeded even in the dry state, and the discrimination becomes difficult when comparing the intensity of simple frequency components.

  In order to achieve the problem to be solved by the above invention, a road surface condition determination method according to the present invention detects a contact sound between a tire of a vehicle traveling on a road and a road surface with a sound sensor, and the detected contact sound is detected. The digital data converted to digital signal is Fourier transformed to the frequency intensity of the sound for each frequency, and the intensity of each frequency component of each vehicle sound that has passed within a preset time is converted to the total sum of all the bands Calculate the frequency intensity, extract each vehicle part from this full-band frequency intensity based on the threshold for vehicle extraction, and normalize the frequency characteristics of each extracted vehicle running sound with the high-pass filter frequency Processing to calculate the standardized frequency intensity of all vehicles, and based on the standardized frequency intensity of all vehicles by this standardization process, a first threshold is set between the wet and dry road surface conditions, and the road surface conditions Dry A second threshold value is set between the pressure snow and the road surface condition is determined to be wet when the normalized frequency intensity is greater than or equal to the first threshold value, and a difference between the first threshold value and the second threshold value is determined. When it is in between, the road surface state is determined to be dry, and when it is smaller than the second threshold value, the road surface state is determined to be compressed snow.

  The road surface condition determining method according to the present invention is the road surface condition determining method, wherein the first threshold value and the second threshold value are obtained by averaging standardized frequency intensities of all vehicles by the standardization process, and by road surface condition. It is preferable to calculate the average normalized frequency intensity and set based on the average normalized frequency intensity for each road surface condition.

  Further, in the road surface condition determining method according to the present invention, in the road surface condition determining method, when the normalized frequency intensity is equal to or lower than a first threshold value, the road surface condition is again a normalized frequency according to road surface conditions of dry and compressed snow. Based on the strength, the standardized frequency strength of all vehicles is calculated by standardizing with a new high-pass filter frequency, and the road surface condition is determined to be dry and compressed snow based on the standardized frequency strength of all vehicles by this standardization processing. A third threshold value is set in between, and when the normalized frequency intensity is equal to or greater than the third threshold value, the road surface condition is determined to be dry, and when it is smaller than the third threshold value, the road surface condition is determined to be compressed snow. It is preferable.

  Further, the road surface condition determining method of the present invention is the road surface condition determining method, wherein the first threshold value and the second threshold value or the third threshold value are standardized for all vehicles by corresponding normalization processing. It is preferable to average the frequency intensity to calculate the average normalized frequency intensity for each road surface condition, and to set based on the average normalized frequency intensity for each road surface condition.

  Further, the road surface condition determining method according to the present invention is the road surface condition determining method, wherein in addition to the sound sensor, a snow detector detects whether or not it is snowing, and the normalized frequency intensity is a third threshold value. Even if it is the above, when the snowfall detector is detecting snowfall, it is preferable to determine a road surface state as snow pressure.

A road surface condition determination device according to the present invention includes a sound sensor that detects a contact sound between a tire of a vehicle traveling on a road and a road surface,
A Fourier transform processing unit for Fourier transforming digital data obtained by converting the contact sound detected by the sound sensor into a digital signal to the frequency intensity of the sound for each frequency;
A full-band frequency intensity calculating unit that calculates the full-band frequency intensity by converting the intensity of each frequency component of the sound of each vehicle that has passed in a preset time into an overall total intensity;
A vehicle extractor for extracting each vehicle part from the full-band frequency intensity calculated by the full-band frequency intensity calculator based on the vehicle extraction threshold;
A normalization processing unit that calculates the normalized frequency intensity of all vehicles by normalizing with a high-pass filter frequency with respect to the frequency characteristics of each vehicle running sound extracted by the vehicle extraction unit,
Based on the normalized frequency intensity of all vehicles by the standardization processing unit, a first threshold value is set between the wet and dry road surface conditions, and a second threshold value is set between the dry and compressed snow. When the normalized frequency intensity is greater than or equal to the first threshold value, the road surface condition is determined to be wet, and when it is between the first threshold value and the second threshold value, the road surface condition is determined to be dry. A road surface discriminating unit that determines that the road surface state is compressed snow when it is smaller than the second threshold value;
It is characterized by having.

  In the road surface condition determining apparatus according to the present invention, in the road surface condition determining apparatus, when the normalized frequency intensity is equal to or lower than a first threshold value, the road surface state is again determined to be dry or compressed snow. Based on the standardized frequency strength for each road surface, normalization processing is performed with a new high-pass filter frequency to calculate the standardized frequency strength for all vehicles, and the road surface condition is based on the standardized frequency strength for all vehicles by this standardization processing. Sets a third threshold value between dry and compressed snow, and determines that the road surface condition is dry when the normalized frequency intensity is greater than or equal to the third threshold value, and road surface when smaller than the third threshold value. It is preferable that the state is determined to be snow pressure.

  The road surface condition determining apparatus according to the present invention further includes a snowfall detector that detects snowfall in addition to the sound sensor in the road surface condition determining apparatus, and the road surface determining unit has the normalized frequency intensity of the first. Even when the threshold value is 3 or more, it is preferable that the road surface condition is determined to be compressed snow when the snowfall detector detects snowfall.

  In the road surface condition judging device according to the present invention, in the road surface condition judging device, a sound sensor is attached to a vehicle traveling on the road at a certain height position of a support member standing near the road. It is preferable.

  In the road surface condition determining apparatus according to the present invention, in the road surface condition determining apparatus, it is preferable that the sound sensor is attached to a part of the vehicle traveling on the road.

  In the road surface condition determining apparatus according to the present invention, in the road surface condition determining apparatus, the sound sensor is preferably attached to the tire house itself or in the vicinity thereof in the vehicle.

  In the road surface condition determining apparatus according to the present invention, the sound sensor is preferably an electric condenser microphone element whose periphery is sealed with resin.

  As can be understood from the means for solving the problems as described above, according to the road surface condition determination method of the present invention, the sound of each vehicle that has passed within a preset time is Fourier-transformed to obtain the sound of each vehicle. By converting the intensity of each frequency component to the total intensity, calculating the full-band frequency intensity, and standardizing the frequency characteristics of each vehicle running sound with the high-pass filter frequency, The same judgment level can be used, and the road surface condition can set the first and second threshold values between wet, dry and compressed snow, and the road condition of wet, dry and compressed snow can be judged accurately. can do.

  Further, according to the road surface condition determination device of the present invention, the full-band frequency intensity calculation unit sums the intensity of each frequency component of the sound of each vehicle obtained by Fourier transforming the sound of each vehicle that has passed within a preset time. The frequency intensity of the entire band is calculated by converting the intensity to the specified intensity, and the normalization processing is performed with the high-pass filter frequency on the frequency characteristics of each vehicle running sound by the normalization processing unit, regardless of the difference depending on the vehicle type. In addition to being able to set the same judgment level, the road surface determination unit can set the first and second threshold values between wet, dry and compressed snow as the road surface condition. It can be judged accurately.

  Embodiments of the present invention will be described below with reference to the drawings.

  Referring to FIG. 1, for example, an automobile C such as a general passenger car or a large vehicle as a vehicle is traveling to the front side in a direction orthogonal to the paper surface in FIG. 1 on a road R made of, for example, asphalt or concrete. . For example, a pole 1 as a support member is erected in the vicinity of the road R, for example, on the left side in FIG. A bracket 3 is attached to the pole 1 at a certain height, for example, at a height of 4 m, and a sound sensor as a sound sensor as an input unit in the road surface condition determination device 5 is attached to the tip of the bracket 3. A converter, for example, a microphone 7 is provided so as to incline in the diagonally lower right direction toward the automobile C where the vehicle 7 travels. In addition, a road surface condition determining device main body 9 in the road surface condition determining device 5 is provided in the vicinity of the microphone 7 provided on the pole 1.

  A bracket 11 is attached to the pole 1, and a snowfall detector 13 is provided at the tip of the bracket 11.

  Referring to FIG. 2, a configuration block diagram of the road surface condition judging device 5 is shown. The road surface condition determination device 5 includes a CPU 15 as a central processing unit. The CPU 15 is a sound sensor as an input unit that detects a contact sound between a tire of a vehicle traveling on a road and a road surface. A microphone 7 and a memory 17 for storing digital data (raw data) obtained by converting a contact sound input from the microphone 7 into a digital signal are provided.

  Further, the CPU 15 includes a Fourier transform processing unit 19 (FFT processing unit) that Fourier-transforms digital data into the frequency intensity of sound for each frequency, and the intensity of each frequency component of each vehicle sound that has passed in a preset time. Is converted into a total sum intensity, and an all-band frequency intensity calculating unit 21 that calculates an all-band frequency intensity, and a vehicle extraction threshold 23 (see FIG. 5) from the all-band frequency intensity calculated by the all-band frequency intensity calculating unit 21. 7), a vehicle extraction unit 25 for extracting each vehicle portion is provided. The vehicle extraction threshold value 23 will be described later in detail.

  Further, the CPU 15 normalizes the magnitude of each frequency component higher than the high-pass filter frequency with respect to the frequency characteristics of each vehicle running sound extracted by the vehicle extraction unit 25 by the overall sound volume. The standardization processing unit 27 that calculates the standardized frequency intensity of all the vehicles and performs the normalization process, and the road surface state is between wet and dry based on the standardized frequency intensity of all the vehicles by the standardization processing unit 27. Is set to the first threshold value 29 (see FIG. 12), the road surface condition is set to the second threshold value 31 (see FIG. 12) between dry and compressed snow, and the normalized frequency intensity is set to the first value. When the threshold value is 29 or more, the road surface condition is determined to be wet, and when it is between the first threshold value 29 and the second threshold value 31, the road surface condition is determined to be dry. When it is small, a road surface discrimination unit 33 that determines that the road surface state is compressed snow It is gills.

  The road surface discriminating unit 33 averages the standardized frequency intensities of all the vehicles by the standardization processing unit 27 to calculate the average standardized frequency strength by road surface condition, and based on the average standardized frequency intensity by road surface condition. The first threshold value and the second threshold value can be set.

  Further, the CPU 15 is connected to a snowfall detector 13 and a display device 35 such as a CRT or a liquid crystal panel as an output unit for displaying the result determined by the road surface determination unit 33.

  Next, a road surface condition determination method according to the first embodiment of the present invention will be described using the road surface condition determination device 5 described above.

  Referring to FIG. 3, as a schematic flow of the road surface condition determination method, in step S <b> 1, vehicle running sound digitally recorded by the microphone 7 for 10 minutes as a preset time is subjected to FFT processing (fast Fourier transform: A Fourier transform is performed by Fast Fourier Transform, and the intensity for each frequency is calculated.

  In step S2, the time-series full-band frequency intensity of the sound of each vehicle in step S1 is calculated.

  In step S3, each vehicle portion is extracted from the full-band frequency intensity in step S2 based on the vehicle extraction threshold value 23.

  In step S4, if there is a vehicle part extracted in step S3, the process proceeds to step 5, and if there is no extracted vehicle part, it is output to the display device 35.

  In step S5, the high-pass filter process (HPF) is performed on the frequency characteristics of each vehicle running sound extracted in step S3, and the normalization process is performed.

  In step S6, based on the normalized frequency intensity standardized in step S5, it is determined whether the road surface state is wet, dry, or compressed snow and output to the display device 35.

  The above steps S1 to S6 will be described in detail.

  Referring to FIG. 4, in step S1, the microphone 7 detects a contact sound between a tire of a vehicle traveling on the road and the road surface, and the detected contact sound is converted into a digital signal, for example, 10 minutes. The vehicle running sound digitally recorded is Fourier-transformed by FFT processing (Fast Fourier Transform) in the Fourier transform processing unit 19 to calculate the intensity for each frequency.

When the sampling frequency at the time of digital recording is, for example, 22.05 kHz, FFT processing is performed using 65536 pieces of digital data and data every [(1/22050) ^* 65536] for about 3 seconds. Therefore, in 10 minutes, about 200 FFT processing results are obtained.

  Here, a series of processing methods from the FFT processing to the normalization processing will be described in detail.

  FIG. 4A shows the result of digital recording of the vehicle running state, where the horizontal axis indicates time (minutes) and the vertical axis indicates intensity (sound amplitude). The portion where the amplitude rises greatly indicates the passage of the vehicle, and the portion indicated by an ellipse indicates the passage of a large vehicle and a small vehicle (ordinary vehicle) as a representative vehicle type vehicle. In addition, the sampling frequency of the example of this FIG. 4 (A) was 22.05 kHz, and the road surface condition was a dry state.

  Since the digital data as described above is prepared for each road surface condition, the frequency analysis for each vehicle type is performed. For example, as described with reference to FIG. 31 in the related art, Fourier is applied to the traveling sound in the time zone [each elliptical portion in FIG. 4A] when a small car (ordinary car) and a large car travel on a dry road surface and a wet road surface, respectively. From the result of the conversion, if a small car (ordinary car) and a large car are distinguished from each other, a high frequency component sound can be extracted by a frequency filter, and a dry state and a wet state in each vehicle type can be distinguished. Actually, since there are various types of vehicles that pass within a preset time, it becomes impossible to distinguish between dry and wet.

Therefore, the intensity fω (x) of each frequency component is expressed as a ratio to the overall intensity. That is, it is converted into a numerical value indicating how much ratio Fω (x) each individual frequency component fω (x) indicates with respect to the total intensity F _total of the total frequency components. This conversion is as shown in Equation (1).

  Where Fω (x) is the intensity of the x frequency, ωmin is the minimum frequency, and ωmax is the maximum frequency.

  In Formula (1), Fω (x) indicates the ratio of the intensity of the x frequency to the intensity of the entire frequency.

  Therefore, when the frequency component of each vehicle type is converted to Formula (1) and expressed in a graph, it is as shown in FIG. The vertical axis of the graph is expressed on a logarithmic scale. By converting in this way, the intensity is expressed within a numerical range of 1.0 at the maximum regardless of the vehicle type, and converted to the same scale.

  From the result of FIG. 5, it can be seen that the conversion intensity is increased at a frequency higher than about 2 kHz regardless of the vehicle type. That is, it is expressed that the vehicle running sound on the wet road surface has a high ratio of the high frequency to the whole regardless of the vehicle type. In other words, by normalizing the magnitude of each frequency component with the overall loudness, it was possible to remove the influence of the magnitude of the running sound of the vehicle.

From the result expressed in this way, for example, when the intensity at 6 kHz is larger than 0.002, it can be determined that the liquid is wet. However, although it is possible to discriminate even by paying attention to specific frequency components in this way, in the result of FIG. 5, the same magnitude relationship is continued for all frequencies above 2 kHz. Therefore, Equation (1) is rewritten as Equation (4) based on Equations (2) and (3).

Where fω (x) is the intensity of the x frequency, ωmin is the minimum frequency, ωmax is the maximum frequency, ωHPF is the high-pass filter frequency, and F _STD is the normalized frequency intensity.

In the F _{HPF of the} equation (4), for example, in FIG. 6, the frequency components from ωHPF to ωmax are extracted and integrated with respect to the intensity of the x frequency of the dry road surface and the wet road surface of the large vehicle. That is, the frequency component is extracted and integrated by the high pass filter. By dividing this result by the integral value F _total of the entire frequency components, the ratio of the frequency intensity after the high-pass filter to the entire intensity can be expressed. As a result, since all frequency components having characteristics can be added, more stable measurement can be performed as compared with comparison at a specific frequency.

In the first embodiment, the processing of Equation (4) is referred to as “standardization processing”, and F _STD is referred to as “normalized frequency strength”. Of course, as in Equation (1), F _STD is 1.0 at the maximum, and the influence on the difference in sound intensity depending on the vehicle type is removed, and it can be handled in the same dimension.

In the example of FIG. 5, the result of calculating the normalized frequency intensity F _STD when ωHPF is 2 kHz is 0.219 for small car drying, 0.399 for small car drying, and 0.188 for large car drying. It becomes 0.432 when the large vehicle is wet.

From this result, when discriminating between wet and dry, it is possible to discriminate the road surface state regardless of the vehicle type of the passing vehicle if the normalized frequency intensity F _STD is 0.25 or higher. However, in road surface condition measurement for snow and ice management of roads, it is basic to measure at least every 10 minutes rather than judging by the passage of one vehicle. It is necessary to make a comprehensive judgment for 10 minutes.

  In step S2, the all-band frequency intensity calculation unit 21 converts the intensity of each frequency component of the sound of each vehicle into an intensity that is totaled according to the above-described equation (2). That is, as a result of performing the processing about 200 times on the digital recording data of FIG. 4A, as a result of integration of all the frequency components, all time-series bands as shown in FIG. A frequency intensity is calculated.

  In step S3, the arrow part of FIG. 4 (A), (B) has shown passing of a vehicle, and when driving noise of a vehicle is made into object, it is necessary to extract this vehicle passing part. Therefore, regarding the vehicle separation, the vehicle extraction unit 25 can extract the vehicle portion based on the vehicle extraction threshold value 23 having the magnitude of the entire frequency band intensity as shown in FIG. The vehicle extraction result is as shown in FIG.

  In step S4, the vehicle extraction result of FIG. 8 is extracted as one point data or a plurality of points because the existence time (vehicle passing time) differs depending on the traveling speed of the vehicle and the loudness of the traveling sound. Some are extracted as data. However, the extraction result of a plurality of points of continuous data can be regarded as passing one vehicle. That is, these extraction points indicate the running sound when passing through the vehicle. Therefore, it can be determined whether there is a passing vehicle.

  In step S5, based on the vehicle extraction result extracted in step S3, the frequency characteristic of the vehicle running sound in the passing vehicle time portion is as shown in FIG. 9A. Using this result, the normalization processing unit 27, the normalization process of Formula (4) is performed. The standardization processing result for each vehicle can be expressed as a standardized numerical value from 0 to 1.0 as shown in FIG. Note that the horizontal axis of FIG. 9A is a logarithmic scale in terms of frequency expression.

  In the above normalization process, ωHPF is set to 2 kHz as shown in FIG. Furthermore, for a case where a plurality of points are extracted with respect to one vehicle passing, a point indicating the maximum value among the plurality of points is used as a representative point of a plurality of points, but an average value of a plurality of points may be used. .

  However, as a method for determining the above-described ωHPF, for example, FIG. 9A shows frequency characteristics regarding the passing sound of one vehicle, but this frequency characteristic varies depending on individual vehicles and road surface conditions. . Therefore, FIG. 10 shows the result of averaging the frequency characteristics of the traveling sounds of all the vehicles extracted in one hour in each road surface state. In addition, about this normalization intensity | strength, after calculating Formula (1), it will average.

  First, when attention is paid to drying and wetting, it can be seen that the frequency characteristics of drying and wetting intersect at approximately 2 kHz. In other words, in the wet state, the intensity of the high frequency is stronger than in the dry state, and when performing the calculation of Equation (4), if ωHPF is selected to be 2 kHz, only the frequency range that is different will be added. , You can see that the difference is most obvious. That is, it is possible to determine the optimal ωHPF by collecting traveling sounds of typical road surface conditions and obtaining intersection points of the normalized frequency characteristics (average).

  In step S6, the road surface discriminating unit 33 determines whether the road surface is wet, dry, or compressed snow. FIG. 11 shows the characteristic extraction result of the traveling sound for each road surface state obtained as shown in FIG. 9B. In FIG. 11, ωHPF is set to 2 kHz, and the number of running vehicles for one hour in each road surface state and the normalized frequency intensity value of each vehicle are represented on the graph. The horizontal axis of FIG. 11 is expressed as a number, but the result is that the vehicle parts extracted from the running sound during one hour (similar to FIG. 9B) are assigned numbers in order. .

  In FIG. 11, although the values for each point (one vehicle) vary, it can be seen that each road surface state unit is distributed in a band shape. Here, from the positioning of the road surface condition determination device 5 of this embodiment, it is not necessary to determine the road surface state from a single vehicle running sound, and it is sufficient to determine once every 10 minutes in relation to the existing telemeter. . Therefore, FIG. 12 shows the result of expressing the values in FIG. 11 as average values in units of 10 minutes.

From FIG. 12, the band-like distribution of FIG. 11 is averaged every 10 minutes, so that the difference in the distribution becomes clear, and the first threshold of the normalized frequency intensity F _STD is determined for the discrimination between dry and wet. It can be seen that the value 29 should be 0.4. It can also be seen that the second threshold value 31 of the normalized frequency intensity F _STD can be set to 0.3 for drying and snow pressure.

Therefore, in the road surface discrimination method of the first embodiment, as shown in the flowchart of FIG. 13, when the normalized frequency strength F _STD is, for example, 0.4 or more as the first threshold value 29, the road surface state Is determined to be wet, and when the first threshold value 29 is between 0.4 and 0.3 as the second threshold value 31, for example, the road surface condition is determined to be dry and the second threshold value 31 is determined. For example, when it is smaller than 0.3, it is possible to determine that the road surface condition is snow.

  As described above, the all-band frequency intensity calculation unit 21 performs Fourier transform on the sound of each vehicle that has passed within a preset time, and converts the intensity of each frequency component of the sound of each vehicle into an overall intensity. The frequency intensity of all bands is calculated, and the normalization processing unit 27 performs normalization processing on the frequency characteristics of each vehicle running sound with a high-pass filter frequency, so that the same judgment level can be obtained regardless of the type of vehicle. In addition, the road surface discriminating unit 33 can set the first and second threshold values 31 between wet, dry, and compressed snow, and accurately determine the road surface state of wet, dry, and compressed snow. can do.

  Next, the road surface discriminating method of the second embodiment will be described. As shown in FIG. 12, the difference between dry and compressed snow is smaller than the discrimination between dry and wet. Returning to the graph of frequency characteristics comparison for each road surface condition in FIG. 10 again, paying attention to the intersection on the frequency characteristics of the dry and compressed snow state, in the dry and compressed snow state, the low frequency sound in the case of the compressed snow. The intersection on the frequency characteristic is in the vicinity of 1 kHz.

  In the first embodiment, the difference between dryness and wetness has been mainly handled as ωHPF being 2 kHz. However, considering the above points, it is optimal to set ωHPF to 1 kHz in order to discriminate between dry and compressed snow. It can be said that.

  Therefore, as in FIG. 11, the normalized frequency intensity of each vehicle in the dry and compressed snow state when ωHPF is 1 kHz is expressed in the graph of FIG. Further, FIG. 15 shows the result of expressing the values in FIG. 14 as average values in units of 10 minutes.

  Comparing FIG. 15 with FIG. 12 described above, it can be seen that the difference between dry and compressed snow is greater in FIG. That is, road surface discrimination can be performed more stably. Eventually, there is an optimum frequency value for ωHPF for standardization processing, and the frequency value can be known from the intersection in the frequency characteristics of typical road surface conditions.

From the above, the road surface discrimination method according to the second embodiment is performed when the normalized frequency strength F _STD is, for example, 0.4 or more as the first threshold value 29 as shown in the flowchart of FIG. Determines that the road surface is wet. On the other hand, when the normalized frequency intensity F _STD is smaller than 0.4 as the first threshold value 29, for example, the new high-pass filter frequency is again determined based on the normalized frequency intensity for each road surface condition of dry and compressed snow. When normalization processing is performed with ωHPF and the road surface condition is between dry and compressed snow, for example, 0.55 is set as the third threshold 37, and when the normalized frequency strength F _STD is greater than or _equal to the third threshold 37 When the road surface condition is determined to be dry and the third threshold value 37 is smaller than 0.55, for example, the road surface condition can be determined to be compressed snow.

  As described above, by newly setting the third threshold value 37, it is possible to reduce the determination error between drying and snow pressure. In the road surface discriminating unit 33, the third threshold value 37 is also normalized by the standardization processing unit 27 for all vehicles in dryness and snow pressure, similarly to the first threshold value 29 and the second threshold value 31 described above. The frequency intensity is averaged to calculate the average normalized frequency intensity for each road surface condition, and can be set based on the average normalized frequency intensity for each road surface condition.

  Next, a road surface discriminating method according to the third embodiment will be described. As can be seen from FIG. 12, the value of the normalized frequency intensity is large in the wet state and small in the compressed snow state, mainly in the dry state. That is, the dry state has a boundary with other states. In particular, regarding compressed snow and dryness, mistakes between the two can be said to be errors that must be avoided in road management, but errors may occur as long as there is a boundary between the two values.

  Therefore, it is possible to predict that the snowfall detector 13 is also provided and knowing the snowfall situation at the time of the road surface determination and the previous snowfall condition, so that it is clearly not dry, so even if the normalized frequency intensity is in the range of compressed snow However, this error can be reduced if the judgment is changed depending on the situation of the snowfall detector 13.

The road surface discriminating method of the third embodiment finally includes the snowfall detector 13 as shown in the flowchart of FIG. For example, in the flowchart of FIG. 16 of the road surface discrimination method of the second embodiment described above, the flow is the same until the third threshold value 37 is set between drying and snow pressure, but the normalized frequency intensity F _STD Is, for example, 0.55 or more as the third threshold value 37, when the snowfall detector 13 does not detect snowfall, the road surface condition is determined to be dry, and the snowfall detector 13 detects snowfall. In this case, even if the normalized frequency strength F _STD is equal to or greater than the third threshold value 37, the road surface condition can be determined as compressed snow.

On the other hand, when the normalized frequency intensity F _STD is smaller than 0.55 as the third threshold value 37, for example, the road surface state is determined to be compressed snow.

  In the above case, only the presence / absence of snowfall by the snowfall detector 13 is reflected in the judgment of the road surface condition. For this judgment, for example, “There has been snowing since one hour ago” It also includes referring to a flow of various situation changes such as “via a wet state”.

  As described above, by using the snowfall detector 13 together, it is possible to further reduce the judgment error between snow pressure and dryness.

  The above description is an example in which a microphone 7 (fixed point type sound sensor) as a sound sensor is attached to a predetermined height position of, for example, the pole 1 as a support member standing near the road R. However, even if the sound sensor is attached to, for example, a part of an automobile as a vehicle, it is possible to determine the road surface state. This will be described in detail.

  As shown in FIG. 18A, a microphone 39 as a sound sensor is attached in the vicinity of a tire house 43 in, for example, the rear wheel 41 of, for example, an automobile C as a vehicle. More specifically, in the enlarged view shown in FIG. 18 (B), the traveling sound of the vehicle C traveling on the road R is captured by the microphone 39 that captures the sound that the water or snow ice splashed by the tire hits the tire house 43. Sound data was collected as described above.

  As shown in FIG. 19, the automobile equipped with the microphone 39 traveled on the wet road surface R <b> 1 sprayed on the road R and the traveling sound was recorded by the microphone 39. The result is shown in FIG. Further, FIG. 21 shows the result of calculating the normalized strength of the sound data by the above equation (1). Note that the microphone 7 (fixed-point sound sensor) of the above-described sound sensor extracts the time when the vehicle is running and performs frequency analysis for about 3 seconds, but the microphone 39 (vehicle-mounted sound sensor) is mounted on the vehicle. Therefore, it is possible to measure everything when the car is running. Here, frequency analysis is performed by extracting approximately 1.5 seconds from the state in which the vehicle travels stably.

As a result, as in the case of the microphone 7 (fixed point type sound sensor) described above, the difference between the dryness and the wetness clearly appears as shown in FIG. 21, so that the sound hitting the tire house 43 is measured. By doing so, it is possible to determine the road surface condition. However, some measures are required because the microphone 39 can be easily estimated to be affected by a person's conversation or radio sound by placing the microphone 39 in the car.

  Originally, when capturing the running sound of a car in a car, there is a type that captures the sound that enters through the gaps of windows and the like, and a type that captures the vibration of the vehicle body such as the tire house 43 as described above. It is done. However, the former is disadvantageous when the vehicle has high airtightness, and the latter has a problem of attenuation in space, and mixing with other in-vehicle sounds is a problem. Therefore, the following approach was taken from the viewpoint of “vibration hitting the tire house 43”, which can reduce attenuation in the space as much as possible and expect stable measurement without being affected by the sound of the vehicle interior.

  When collecting sound in the car, it is unavoidable to be affected by human conversation and radio sound. Therefore, it is considered that the above problem can be solved if the “sound” of water or snow ice splashed by the tire hits the tire house 43 is measured as “vibration”.

  Therefore, as shown in FIG. 22, for example, a vibration meter 45 as a sound sensor is directly attached to the tire house 43 and a running experiment similar to that described above was performed. The normalized intensity obtained by frequency analysis of the collected sound data at that time is shown in FIG. From this result, as with the microphone 39, it was confirmed that the vibrometer 45 can distinguish between dry and wet.

  Next, an electric condenser microphone (hereinafter referred to as ECM) element 47 is used as a microphone to be used, and the ECM element 47 is directly attached to the tire house 43 as shown in FIG. We decided to measure the vibration as the sound (pressure) and the vibration with maximum efficiency.

  The normalized intensity results obtained by frequency analysis of the collected sound data are shown in FIG. Note that the periphery of the ECM element 47 was sealed with resin 49 as shown in FIG.

  Thus, even when the ECM element 47 was attached to the tire house 43 and measured, the same result as that of the vibrometer 45 could be obtained.

  Using the same method as the fixed-point sound sensor, we tried to determine the road surface condition.

  For example, Fig.25 shows the frequency characteristics when ωHPF is determined. The frequency characteristics are clearly different depending on the road surface condition. Further, FIG. 26 shows the result of averaging road surface conditions with different dates and places assuming normal driving.

  First, in FIG. 26, when attention is paid to drying and wetting, it can be seen that the frequency characteristics of drying and wetting intersect at approximately 700 Hz. That is, in the wet state, the intensity of the high frequency is strong against drying, and if ωHPF is selected to be 700 Hz, only the frequency range with a difference is added, so that the difference becomes clearest. In other words, the optimum ωHPF can be determined by collecting the running sound of typical road surface conditions and finding the intersection of the normalized frequency characteristics (average).

  Accordingly, FIG. 27 shows the result of arranging the normalized frequency intensities in time series (at intervals of about 1.5 seconds) with respect to the running data with ωHPF in the above-described equation (3) set to 700 Hz. As described above, since the normalized frequency intensity is separated according to the road surface condition, it is possible to easily determine the road surface condition by setting a threshold value at this boundary as in the above-described fixed-point sound sensor. Become. In FIG. 27, the first threshold value for determining wetness and dryness can be set to “0.3”, and the second threshold value for determining dryness and snow pressure can be set to “0.1”.

  Basically, it was possible to determine the road surface condition by measuring the sound hitting the tire house 43 with the microphone 39 placed in the car, but other sounds (conversation, radio, etc.) generated in the car were obtained. Therefore, if the sound hitting the tire house 43 is measured as vibration, the previous problem can be solved, and an experiment using the vibrometer 45 can be performed to obtain the expected result. It was. In addition, instead of the vibration meter 45, the ECM element 47 is directly attached to the tire house 43, and by taking measures to prevent sound from leaking, an approach to measure vibration as sound can be obtained and good results can be obtained. It was.

  By using a method similar to that of the fixed point type sound sensor described above, it is possible to easily determine the road surface condition of dry, wet, and compressed snow.

  A road surface condition determination method according to a fourth embodiment of the present invention will be described using the road surface condition determination device 5 described above.

  Referring to FIG. 28, as an outline flow of the road surface condition determination method, in step S7, for example, vehicle running sound digitally recorded by the ECM element 47 for 1.5 seconds as a preset time is subjected to FFT processing (high speed Fourier transform is performed by Fast Fourier Transform), and the intensity for each frequency is calculated.

  In step S8, the time-series full-band frequency intensity of the sound of each vehicle in step S7 is calculated.

  In step S9, the high-pass filter process (HPF) is performed on the frequency characteristic of each vehicle running sound extracted in step S8, and the normalization process is performed.

  In step S10, based on the normalized frequency intensity standardized in step S9, it is determined whether the road surface state is wet, dry, or compressed snow and output to the display device 35.

In the road surface discrimination method of the fourth embodiment, as shown in the flowchart of FIG. 29, when the normalized frequency intensity F _STD is, for example, 0.3 or more as the first threshold value 29, the road surface state is wet. Judge. On the other hand, when the normalized frequency strength F _STD is smaller than 0.3, for example, as the first threshold value 29, the new high-pass filter frequency is again determined based on the standardized frequency strength according to the road surface condition of dry and compressed snow. When normalization processing is performed with ωHPF and the road surface condition is between dry and snow-capped, for example, 0.1 is set as the second threshold 31 and when the normalized frequency strength F _STD is greater than or _equal to the second threshold 31 When the road surface condition is determined to be dry and the second threshold value 31 is smaller than 0.1, for example, the road surface condition is determined to be compressed snow.

  From the above, the fixed point sound sensor has a large number of installations to manage the entire road surface, and it cannot be determined unless the vehicle passes, but since it is a constant monitoring type, real-time information output Can be made possible. In addition, the vehicle-mounted sound sensor is convenient for grasping the road surface condition of the entire route although it does not have real-time characteristics because it involves travel time. In addition, since the traveling sound of the host vehicle is always generated, it is possible to always make a determination. Therefore, for the manager who manages the road, both can be used properly depending on the usage, and both are effective.

It is the front view which provided the road surface condition determination apparatus of embodiment of this invention in the vicinity of the road. 1 is a block diagram illustrating a configuration of a road surface condition determination apparatus according to an embodiment of the present invention. It is a flowchart which shows the road surface condition determination method of embodiment of this invention. (A) is a graph which shows digital sound recording data, (B) is a graph which shows a full band frequency intensity from the graph of (A). It is the graph which expressed the intensity | strength of each frequency component in the graph of FIG. 19 with the ratio with respect to the whole intensity | strength. It is the graph which performed the normalization process by performing the high-pass filter process with respect to the frequency component of the large vehicle in the graph of FIG. It is a graph when the threshold value for vehicle extraction is set with respect to the graph of the frequency intensity of all the bands of FIG. It is a graph which shows the result of having extracted the vehicle from FIG. (A) is a graph which shows the frequency characteristic of the vehicle running sound in a passing vehicle time part, (B) is a graph which normalized and processed using the result of (A) in each vehicle. It is a graph which shows the average frequency characteristic comparison according to each road surface state. It is a graph which shows the normalized frequency intensity according to the road surface state of each vehicle when ωHPF is 2 kHz. It is a graph which shows the average normalized frequency strength averaged for every 10 minutes with respect to the graph of FIG. It is a flowchart which shows the road surface discrimination method of 1st Embodiment of this invention. It is a graph which shows the normalized frequency intensity | strength of each vehicle in the dry and pressure-snow condition when (omega) HPF is 1 kHz. It is a graph which shows the average normalized frequency intensity averaged for every 10 minutes with respect to the graph of FIG. It is a flowchart which shows the road surface discrimination method of 2nd Embodiment of this invention. It is a flowchart which shows the road surface discrimination method of 3rd Embodiment of this invention. (A) is the perspective view which has arrange | positioned the microphone near the tire house in the rear wheel of a motor vehicle, (B) is the enlarged view which has arrange | positioned the microphone near the tire house. It is the figure which showed the state of the wet road surface sprinkled on the road. It is a graph which shows digital recording data in the state of a wet road surface and a dry road surface. It is a figure which shows the graph which converted the state shown in FIG. 20 into the frequency and the normalization intensity | strength. It is the figure which has arrange | positioned the vibrometer in the tire house of a vehicle. It is a figure which shows the graph which converted the state which recorded the sound of the road with the vibrometer shown in FIG. 22 into the frequency and the normalization intensity | strength. It is the figure which has arrange | positioned the ECM element in the tire house of a vehicle. It is a figure which shows the graph which converted the state which recorded the sound of the road with the ECM element shown in FIG. 24 into the frequency and the normalization intensity | strength. It is a figure which shows the graph converted into the frequency and the normalization intensity | strength which averaged the road surface state from which a date and a place differ. It is a graph which shows the average normalized frequency intensity averaged at intervals of 1.5 seconds with respect to the graph of FIG. It is a flowchart which shows the road surface condition determination method of embodiment of this invention. It is a flowchart which shows the road surface discrimination method of 4th Embodiment of this invention. It is a graph which shows digital recording data. It is the graph which performed Fourier transformation with respect to the running sound when a small vehicle and a large vehicle each drive | worked the dry road surface and the wet road surface.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Pole 3 Bracket 5 Road surface condition determination apparatus 7 Microphone (input part; sound sensor)
9 Road surface condition judging device body 11 Bracket 13 Snowfall detector 15 CPU
17 Memory 19 Fourier transform processing unit (FFT processing unit)
21 Full Band Frequency Intensity Calculation Unit 23 Vehicle Extraction Threshold 25 Vehicle Extraction Unit 27 Standardization Processing Unit 29 First Threshold 31 Second Threshold 33 Road Surface Discrimination Unit 35 Display Device (Output Unit)
37 Third threshold 39 Rear wheel 41 Microphone (sound sensor)
43 Tire house 45 Vibrometer (Sound sensor)
47 ECM element 49 Resin

Claims (12)

The contact sound between the tire and the road surface of the vehicle traveling on the road is detected by the sound sensor, and the digital data obtained by converting the detected contact sound into a digital signal is Fourier-transformed into the frequency intensity of the sound for each frequency and set in advance. The total frequency intensity is calculated by integrating the intensity of each frequency component of the sound of each vehicle that has passed within the specified time with respect to all frequencies, and the portion where the intensity is greater than the vehicle extraction threshold is calculated from this all-band frequency intensity. extracted as a vehicle running sound, standardization of all vehicle performs normalization processing for dividing the entire band frequency intensity integrated value on the high-pass filter frequency or higher frequency for the frequency characteristic of each vehicle traveling sound the extracted A frequency intensity is calculated, a first threshold value is set between the wet and dry road surface conditions based on the normalized frequency intensity of all vehicles by this normalization process, and the road surface conditions are dry and compressed snow. Set the smaller second threshold than the first threshold value during said normalized frequency intensity is determined to a wet road surface state when the above first threshold value, the first threshold and the 2. A road surface condition determining method, characterized in that a road surface condition is determined to be dry when it is between two threshold values, and that the road surface condition is determined to be compressed snow when smaller than a second threshold value.   The first threshold value and the second threshold value are obtained by averaging the standardized frequency intensities of all vehicles by the standardization process to calculate the average standardized frequency strength by road surface condition, and the average standardized frequency intensity by road surface condition. The road surface condition determination method according to claim 1, wherein the road surface condition determination method is set based on When the normalized frequency intensity is equal to or lower than the first threshold value, again, all values obtained by integrating the frequency characteristics of each vehicle running sound with respect to a frequency higher than a new high-pass filter frequency lower than the high-pass filter frequency are all obtained. A standardization process for dividing by the band frequency intensity is performed to calculate a standardized frequency intensity for all vehicles, and the road surface condition is between the dry and the compressed snow based on the standardized frequency intensity of all vehicles by this standardization process. A threshold value is set, and when the normalized frequency intensity is greater than or equal to a third threshold value, the road surface condition is determined to be dry, and when it is less than the third threshold value, the road surface condition is determined to be compressed snow. The road surface condition determination method according to claim 1.   The first threshold value and the second threshold value or the third threshold value are calculated by averaging the standardized frequency intensities of all vehicles by the corresponding standardization processing to calculate the average standardized frequency strength by road surface condition, 4. The road surface condition determination method according to claim 1, wherein the road surface condition determination method is set based on the average normalized frequency intensity for each road surface condition.   In addition to the sound sensor, the snow detector detects whether it is snowing or not, and even if the normalized frequency intensity is greater than or equal to a third threshold, the snow detector detects snowfall 4. The road surface condition determining method according to claim 3, wherein the road surface state is determined to be snow pressure. A sound sensor for detecting a contact sound between a tire of a vehicle traveling on a road and a road surface;
A Fourier transform processing unit for Fourier transforming digital data obtained by converting the contact sound detected by the sound sensor into a digital signal to the frequency intensity of the sound for each frequency;
A full-band frequency intensity calculation unit that calculates the full-band frequency intensity by integrating the intensity of each frequency component of the sound of each vehicle that has passed within a preset time with respect to all frequencies ;
A vehicle extraction unit that extracts a portion having a strength greater than the vehicle extraction threshold from the full-band frequency intensity calculated by the full-band frequency intensity calculation unit;
The standardized frequency intensity of all vehicles is obtained by dividing the frequency characteristics of each vehicle running sound extracted by the vehicle extraction unit by the integrated value of the frequency above the high-pass filter frequency by the full-band frequency intensity. A normalization processing unit to calculate,
It sets the first threshold value during a road surface condition based on the normalized frequency intensity of all vehicles by the normalization processing unit and the drying and wetting, the first threshold road condition between the drying and the pressed snow A second threshold value smaller than the first threshold value, and when the normalized frequency intensity is equal to or greater than the first threshold value, the road surface condition is determined to be wet, and between the first threshold value and the second threshold value. A road surface determination unit that determines that the road surface state is dry when there is, and determines that the road surface state is compressed snow when less than the second threshold;
A road surface condition judging device characterized by comprising: When the normalized frequency intensity is less than or equal to the first threshold value, the road surface discriminating unit again integrates a value obtained by integrating a frequency characteristic of each vehicle running sound with respect to a frequency higher than a new high pass filter frequency. The standardized frequency intensity is divided by the frequency intensity to calculate the standardized frequency intensity of all the vehicles. Based on the standardized frequency intensity of all the vehicles by this standardizing process, the road surface condition is third between dry and compressed snow. A threshold value is set, and when the normalized frequency intensity is greater than or equal to a third threshold value, the road surface condition is determined to be dry, and when it is less than the third threshold value, the road surface condition is determined to be compressed snow. The road surface condition determination apparatus according to claim 6, wherein   In addition to the sound sensor, a snowfall detector for detecting snowfall is provided, and the road surface detection unit detects the snowfall even if the normalized frequency intensity is a third threshold value or more. 8. The road surface condition determining apparatus according to claim 7, wherein the road surface condition is determined to be compressed snow when the vehicle is on.   The road surface condition according to claim 6, 7 or 8, wherein the sound sensor is attached to a vehicle traveling on the road at a fixed height position of a support member erected in the vicinity of the road. Judgment device.   7. The road surface condition judging device according to claim 6, wherein the sound sensor is attached to a part of the vehicle traveling on the road.   The road surface condition determination apparatus according to claim 10, wherein the sound sensor is attached to the tire house itself or in the vicinity thereof in the vehicle.   12. The road surface condition judging device according to claim 11, wherein the sound sensor is an electric condenser microphone element whose periphery is sealed with resin.

Images