A method for estimating water levels on roads while tires are moving.
A microphone-based method on the wheel arch analyzes high-frequency tire-water interactions to accurately estimate thin road water layers, improving vehicle safety by enhancing grip estimation and speed management.
Patent Information
- Application Number
- JP2022506625
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2019-08-01
- Filing Date
- 2020-07-20
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2040-07-20
AI Technical Summary
Existing vehicle safety systems fail to accurately detect thin layers of water on the road surface, which significantly reduce tire grip without visible water spray, leading to inaccurate speed management and safety device responses.
A method using a microphone sensor placed on the vehicle's wheel arch to analyze high-frequency acoustic signals from tire-water interactions, isolating frequency bands above 4kHz to accurately estimate water levels on the road, considering vehicle speed and tire characteristics.
Provides reliable and accurate water level estimation with a sensitivity of 1/10 mm, unaffected by vehicle response and road roughness, enhancing safety by improving speed management and grip estimation.
Smart Images

Figure 0007584493000001 
Figure 0007584493000002 
Figure 0007584493000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a method for estimating, in real time, the water level on a road on which the tires of a vehicle are traveling. [Background technology]
[0002] In the field of automotive safety, it is known that a road covered with a layer of water 0.2 millimeters high can worsen the grip conditions of tires on the road. In certain conditions, a reduction in grip of about 30% has been shown compared to a dry road, i.e. a vehicle traveling on a wet road has a significantly longer braking distance. Nevertheless, the driver of the vehicle is not necessarily aware of the fact that such a thin layer of water reduces the grip of the vehicle to such an extent. In fact, with a thickness of less than 0.5 mm, there is no water spray splashing behind other vehicles. In the absence of this spray, the driver has the feeling that the road is dry enough to allow the vehicle to travel at the same speed as on a completely dry road.
[0003] Furthermore, vehicles are nowadays equipped with many safety devices aimed at warning the driver of dangerous situations or even replacing him. For example, there are automatic speed control devices called ACC (Adaptive Cruise Control) or even devices that manage the vehicle's safety distance to the vehicle in front. Currently, to take into account wet roads, ACC type safety devices are based on the presence of water splashes on the windshield to adapt their response to the road conditions and increase the separation distance to the vehicle in front. However, when the vehicle is driving on wet ground with a water level of less than 0.5 millimeters, the water splashes do not reach the windshield. The safety device then manages the driving of the vehicle as if it were driving on a dry road, even though the grip is already greatly reduced.
[0004] Also known are other sensors for measuring water levels by infrared radiation that is reflected by the road coating after being refracted by passing through a layer of water. However, these measuring devices are very sensitive to the color and temperature of the road coating, making them somewhat inaccurate for everyday use. Furthermore, depending on the roughness of the ground on a millimeter scale, which corresponds to the macrotexture, the water level measurements obtained are affected by this characteristic.
[0005] Also known are vibration sensors placed on flexible plates made to vibrate under the impact of splashes generated by the movement of the tire on a wet road. However, these devices are vulnerable due to their location facing the trajectory of the splashes generated by the tire when moving, especially on gravel roads where the tire may capture and release gravel through the dimple pattern of its tread. These sensors are based on the vibration level of the plate excited by the splashes. The vibration analysis is in the low and mid-frequency range between 100 and 3000 Hz, which is sensitive to the deformation of the vehicle's panels, for example the wheel arches. Then, in the vibration response, it is difficult to separate those due to the vibrations of the panel generated by the vehicle excitation from those originating from the splashes. To solve this problem, the sensor is placed close to the splashes on a dedicated plate. [Prior art documents] [Patent documents]
[0006] [Patent Document 1] WO2015092253A1 Summary of the Invention [Problem to be solved by the invention]
[0007] The present invention comprises a method for estimating the water level on a road over which tires are traveling that is reliable, accurate, has a discrimination capability of the order of one tenth of a millimeter of water level, and is insensitive to vehicle response and non-intrusive to the vehicle. [Means for solving the problem]
[0008] The invention relates firstly to a method for estimating the water level on a road on which a tire of a mounted set of a vehicle is traveling, the mounted set being placed on a wheel arch of the vehicle, preferably on a wheel arch linked to a first axle of the vehicle with respect to the normal forward direction of the vehicle. The method comprises the following steps: Fixing the sensor on the vehicle, preferably in link with the wheel arch. Obtaining a frequency signal from a sensor corresponding to a vehicle traveling at a speed V on a road covered by water. Isolating a portion of the water level sensitive frequency signal bounded by two frequencies F1 and F2, F1 being lower than F2. Determining the energy vector linked to that portion of the frequency signal. Obtaining the water level on the road using a function that takes into account the energy vector and the vehicle speed V. The method is characterized in that the sensor is a microphone and that said portion of the frequency signal extends at least partially above 4 kHz, preferably above 7 kHz, particularly preferably above 10 kHz.
[0009] In this document, the term "water level" is understood to mean the average thickness of the water layer bounded on the one hand by the free surface of the water layer with air and on the other hand by the outer surface of the road. Thus, on a perfectly smooth road on a macroscopic scale of the order of millimeters, the water level corresponds to the thickness of the water layer uniformly distributed on the outer surface of the road. For rough roads on a millimeter scale, called macrotexture, the water level corresponds to the average thickness of the water layer located on top of the outer surface of the road over a certain area of the road.
[0010] The method uses a microphone as a sensor to record the vibration signal of the panel under splash excitation limited to 3000 Hz or 4000 Hz, as well as the acoustic signal of the interaction between the water layer and the tire up to several thousand Hertz. That is to say, this phenomenon includes different events such as the filling of the water into the dimple pattern of the tire tread, its release from the dimple pattern, the spraying of water droplets in the air, the impact of these droplets on the wheel arch wall, and the acoustic radiation of this panel that is made to vibrate. Furthermore, the acoustic signature of the water displacement, both the wave front in front of the tire displaced with the vehicle speed or the water discharge wave-like flow on the tire side, and the backflow that occurs behind the tire over large water levels, are also acoustic sources of this physical phenomenon. All these phenomena depend on the amount of water set in motion by the interaction with the tire, and they are directly linked to the water level on the road, ensuring the accuracy of the method.
[0011] The method focuses on a frequency band located at least partially or completely in the high frequencies above 4 kHz, where the airborne propagation of noise dominates compared to the solid or vibration propagation of the vehicle, thereby minimizing the vibration-acoustic response of the panel and highlighting other phenomena. The sensitivity of the acoustic response at high frequencies is higher than the sensitivity of the vibration response at low and intermediate frequencies. Furthermore, this frequency band highlights the acoustic events due to the interaction of the tire with the water layer. The reliability and accuracy of the method are reinforced by selecting a frequency band sensitive to the water level, where the physical events due to the interaction of the water layer with the tire will have an emergent acoustic response. However, at higher frequencies, this physical phenomenon becomes more prominent, since the sound level of the vehicle decreases with frequency. That is, the sensitivity of the method is increased by isolating the highest frequency blocks. Preferably, the frequency band is located at least partially or completely in the high frequencies above 7 kHz, particularly preferably in the high frequencies above 10 kHz.
[0012] The fact that a frequency band is selected that lies between the two boundaries F1 and F2, where F1 is lower than F2, makes it possible to compare the acoustic responses with each other and thus to estimate the water level on the road.
[0013] However, to enhance the accuracy of the method by suppressing the effects of conventional sound reduction in the vehicle, it is useful to construct a rectified frequency signal by comparing the frequency signal from the sensor with a characteristic curve of a frequency signal that incorporates the reduction in the frequency signature of the vehicle.
[0014] The location of the sensor on the vehicle and especially in the wheel arch also increases the reliability of the method. In fact, the wheel arch serves to amplify the acoustic source of the phenomenon by confining the acoustic waves in a small volume due to its shape, i.e., its concave surface. In addition to this, the reflection of these waves in this limited space concentrates the acoustic energy in this space. Finally, the wheel arch is a protection zone for the sensor against other acoustic sources, such as, for example, the aerodynamic flow of air around the vehicle when moving.
[0015] The presence of a layer of water on the road ensures a similar processing of the acoustic waves by this surface, regardless of the rough nature of the ground. Thus, the accuracy of the method is reinforced. Furthermore, the thin wall of the wheel arch can be utilized to place the inactive elements of the sensor, such as the power connection, on the other side of the wheel arch, facilitating the penetration of the microphone into the wheel arch and avoiding contact of most of the sensor with water. Finally, the location of the sensor is as close as possible to the acoustic source of the physical phenomenon, while being physically protected by the structure of the vehicle. Since the microphone could be entirely located behind the wheel arch, in which case the received signal would be attenuated by the sound insulation of the wheel arch, it is preferable to position the active parts of the microphone close to the wall. However, the active parts of the microphone can also be located in the volume bounded by the wheel arch in which the mounting assembly is located, and then the active parts of the sensor are preferably hidden behind a wall with many perforations. These perforations are dimensioned in such a way that the acoustic waves can pass unhindered while protecting the active parts of the sensor against water droplets. The high frequency range used favors this double effect of the perforated wall.
[0016] The determination of the blocks is performed using the rectified frequency signal. The analysis of this rectified frequency signal makes it possible to determine the blocks by comparing the frequency signal with a characteristic curve of the vehicle's frequency-acoustic response. Then, for each determined block, a characteristic value of the block must be evaluated. Finally, this characteristic value is simply evaluated with respect to a predefined threshold value in order to know whether the frequency band associated with the block is held in the portion of the frequency signal. In the affirmative, a frequency band of the block is selected and the frequency signal on this frequency band is separated so as to constitute a sequence of the portion of the frequency signal. Necessarily, the portion of the frequency signal is delimited by the extreme values of the frequency band selected on the rectified frequency signal.
[0017] From the separated frequency signal it is possible to determine a set of energy criteria that will give an energy vector, which is defined in at least one dimension. The choice of the energy criterion is due to the physical phenomenon that is sought to be studied. In the high frequency range in which the physical phenomenon applies, energy is the most discriminatory indicator.
[0018] Finally, the travel speed increases the amount of water in motion due to interaction with the tire. The speed of the vehicle also modifies the conditions of the physical phenomena of the interaction, for example by increasing the amount of water in motion or the propagation speed of the water droplets emitted from the tire, which directly increases the impact energy of the water droplets on the wall of the wheel arch. As a result, to accurately estimate the water level on the road, it is essential to take into account the travel speed of the vehicle. In fact, this speed directly affects some aspects of the physical phenomena.
[0019] The determination of the water level is obtained through a function, for example a correlation function, that relates the water level, the energy vector and the speed of the vehicle, which is appropriate for low water levels, regardless of the condition of the tire for carved tires, and regardless of the road macro-texture. This function is related to the vehicle or to the group of vehicles to which the vehicle is related. In fact, the acoustic measurements are performed outside the vehicle, close to the mounting assembly. For any automobile, the wheel arch is present and its dimensions are very similar. However, some vehicles have specific features in their wheel arch, such as the lack of soundproofing devices for low-cost vehicles or the ground clearance that affects the loudspeaker effect of the wheel arch.
[0020] Preferably, the frequency signal obtained is a power spectral density, called PSD, of a time-discretized signal obtained from a sensor while the vehicle is running.
[0021] Advantageously, the energy vector comprises an integral function, possibly rectified, of the DSP between frequencies F1 and F2.
[0022] This is a fast and cheap way to obtain the acoustic response of a phenomenon for which the sampling frequency of the time signal must be higher than 8 kHz. The running of a tire on a wet road is not a transient phenomenon, its running time is long with respect to the acquisition of the time signal. Obviously, the longer this signal is, the more likely it is to statically cut off the transients that disturb the acoustic response. Considering that the physical phenomenon is constantly periodic and the acquisition of the time signal is only an extraction of a part of this physical phenomenon, the power spectral density is a good tool to extract the appropriate energy information from the signal.
[0023] The energy vector comprises, for example, the sum of the products obtained by the amplitudes of the PSDs, for all discrete values of the acoustic response held between F1 and F2, discretized at the frequency interval between adjacent discrete values. Of course, other more elementary energy criteria can be used, for example constructed for the real or imaginary part of a frequency signal, or constructed from a frequency signal corresponding to a time response over a finite number of cycles of a physical phenomenon.
[0024] Of course, if the portion of the frequency signal is piecewise defined, then the analysis must be performed for each successive piece and the energy of all the pieces must be summed.
[0025] According to a particular embodiment, the portion of the frequency signal is limited to at least one block whose characteristic value exceeds a threshold value for the frequency signal.
[0026] Focusing the analysis of the acoustic response specifically on one or more blocks improves the sensitivity of the method and, therefore, its discrimination capabilities by keeping only the blocks with the greatest sensitivity to the water level. This selection of said blocks is done using a metric, i.e. a characteristic value of the block, and a threshold value that is necessarily linked to the frequency signal, since the frequency signal is not constant, making it possible to keep only those blocks that provide a significant acoustic weight with respect to the overall response of the vehicle.
[0027] The selection of blocks makes it possible to make the method robust, since the water level estimation is less sensitive to spot signals that are linked to vehicle signatures, for example. Furthermore, the method is more sensitive to physical phenomena that are broadband than those that we wish to separate. Thus, the reliability of the method is clearly improved.
[0028] According to a particular embodiment, the step of isolating the portion of the frequency signal comprises forming a rectified frequency signal as being a comparison of the frequency signal with a characteristic curve of the frequency signal.
[0029] To facilitate the selection of the blocks that make up that portion of the frequency signal, it is preferable to construct a rectified frequency signal, giving equal weight to the various blocks of the acoustic response by taking into account the reduction in the sound level of the vehicle proportional to the frequency. The measured frequency signal must be compared with a characteristic reduction curve in order to obtain a rectified frequency spectrum of the sound level of the vehicle. This comparison can be made either as a difference or as a ratio of the two components to be compared.
[0030] According to a very specific embodiment, the characteristic curve of the frequency signal is a curve that decreases continuously as a function of frequency above 4 kHz.
[0031] It is an elementary model that is sufficient to identify the block in question and can be obtained by a simple piecewise linear regression of the frequency signal, provided that one of the boundaries of the section is around 4 kHz. The natural falloff of the vehicle's acoustic signal makes it easy to identify the slope of the characteristic curve. This evaluation is sufficient to identify the acoustic response on a dry road. However, to completely determine the characteristic curve, a characteristic point on the measured acoustic signal must be identified, which will be called the 4 kHz reference point. In practice, the acoustic amplitude of this characteristic point depends not only on the vehicle's travel speed, but also on the water level and also on the nature of the ground. This characteristic point can be the value of the acoustic response at 4 kHz or the average of the discretized values of the frequency signal over a frequency band around 4 kHz.
[0032] Highly advantageously, at least one block of the rectified frequency signal is delimited by a crossing of a particular value by the rectified frequency signal.
[0033] In fact, the acoustic response of a vehicle decreases with frequency, regardless of whether it is traveling on a dry road or on a road covered with water. If the acoustic response of a vehicle is expressed on a logarithmic scale, its characteristic curve resembles a linear or piecewise linear curve above a frequency of 4 kHz. This decrease is unique to the acoustic response of a vehicle. To identify the slope of this characteristic curve, it is possible, for example, to use a linear regression method over the acoustic spectrum of the vehicle under driving conditions, regardless of the wet or dry nature of the road and coating. To construct the characteristic curve associated with a frequency signal, it is then sufficient to align the start of the curve on the frequency signal.
[0034] The rectified frequency signal is obtained in the form of the difference between the frequency signal and the characteristic curve of the frequency signal on a logarithmic scale. Blocks of the rectified frequency signal are delimited by two consecutive passes through the zero specific value of the rectified frequency signal expressed on a logarithmic scale. If the frequency signal is expressed on a linear scale, the construction of the rectified frequency signal is performed by the ratio of the frequency signal and the characteristic curve, both on a linear scale. The specific value which delimits the blocks of the rectified frequency signal is the value 1.
[0035] Depending on the particular representation mode, the characteristic value of the at least one block may be located in a group that includes the maximum, the mean, and the median.
[0036] In order to preserve the frequency band of a block of frequency signals, the block of frequency signals must be classified. This classification first involves the evaluation of a characteristic metric of the block of frequency signals. Simple metrics such as the maximum value of the frequency signal, its mean value or its median value over the frequency band of the block are suitable to characterize the block. This step is very effective for rectified frequency signals.
[0037] These characteristic values can equally be expressed on a linear scale, a logarithmic scale, or any other type of representation related to the representation of a frequency signal.
[0038] According to a very specific embodiment, the threshold value is a value of 3 dB, preferably a value of 4 dB, particularly preferably a value of 5 dB, when the frequency signal is represented on a logarithmic scale.
[0039] These thresholds indicate the significant difference of the blocks with respect to the acoustic response of the vehicle. As a result, this block is weighted more in the acoustic energy of the vehicle response. Therefore, if the characteristic value of the block exceeds the threshold, it is necessary to keep the frequency band of this block to delimit the sequence of that part of the frequency spectrum. Then, the energy vector becomes sensitive to the water level. This step is very effective for rectified frequency signals.
[0040] Advantageously, the stage of sensitivity of the frequency band of the frequency signal to the water level is carried out in a learning stage during which a vehicle equipped with a microphone drives on a road having at least two water levels which can be differentiated.
[0041] To evaluate the ability of the method to detect such differences, it is useful to perform a driving operation on a water level that can be distinguished in the order of tenths of a millimeter and that is controlled. The vehicle preferably drives on the same ground or on ground of comparable nature and under similar driving conditions such as linear speed, dynamic load of the vehicle, wind, ambient temperature. Furthermore, this sensitivity step must be performed over a range of driving speeds. This sensitivity study makes it possible to identify a correlation function that allows to distinguish the water level from the driving speed and from the energy criteria associated with that part of the frequency signal that is sensitive to the water level. This corresponds to calibrating the frequency signal and the energy criteria associated with each frequency band of the frequency signal to the water level on the road.
[0042] According to a very specific embodiment, the tire comprises an identification vector and the method comprises a step of obtaining an ID of the tire in order to obtain at least one geometric characteristic of the tire, the function of the step of obtaining the water level taking into account the at least one geometric characteristic of said tire.
[0043] A tire, in new condition, comprises at least one identification vector, for example through a label. A tire can also contain identification vectors that follow it throughout its life, in the form of physical, visible vectors attached to it, such as a barcode, a QR code, or in the form of invisible vectors, such as RFID (acronym for Radio Frequency Identification) or TMS (acronym for Tire Mounted Sensor) type electronic devices. The content of all these identification vectors is the ID of the tire. This ID of the tire may also be available directly on the vehicle, by manual, optical or radio frequency input of these various vectors. From this ID, a certain number of tire characteristics are available, especially geometric characteristics. It is thus possible to utilize the tire dimensions, including the tire width, which directly influence the width of the water droplets' release zone and impact zone. It is also possible to utilize the shape of the dimple pattern with respect to the tire's tread, which also results in the modification of the spray and impact zone of the water droplets trapped in the dimple pattern of the tire when driving on a wet road. For example, a summer type tire has longitudinal grooves as its main dimple pattern, which usually generate a longitudinal spray of water droplets, whereas a winter tire has a succession of V-shaped dimples as its dimple pattern, which also generates a 45° spray, which changes the impact zone of the water droplets on the wheel arch wall.
[0044] According to a preferred embodiment, the method includes a step of obtaining a wear state of the tyre, the function for obtaining the water level taking the wear state of the tyre into account.
[0045] When considering the geometrical characteristics associated with the tread dimple pattern, it is also necessary to address the wear state of the tire, which is reflected in the average variation of the tread height of the tire. Wear necessarily begins by scraping the outer surface of the tread. Wear can affect geometrical characteristics such as the dimple pattern, such as the water droplet release zones due to the appearance of new elements of the dimple pattern that were initially hidden in new condition, or the amount of water released due to saturation of the available dimple pattern due to tread wear.
[0046] This tire wear state can be obtained, for example, by devices located in the tire, such as resistive or capacitive sensors in the tread, by devices outside the tire, such as drive-over scanners, or by manual measurements during vehicle inspections, and this information is communicated to the vehicle or on the server by association with the tire ID. Taking into account the tire wear state, for example, for a specific tire, for the function that consequently links the water level on the road to other parameters, may be necessary in the case of a particular tire, where hidden dents in the tread appear as the tire wears. The change in the dent pattern due to tire wear then affects the acoustic signature of the physical phenomenon. Furthermore, tire wear has a slow dynamics in time with respect to the water level variations on the road. This stage of obtaining the tire wear state can therefore be performed well before the measurement that evaluates the water level on the road.
[0047] According to another preferred embodiment, the method includes a step of obtaining the road texture, preferably the MTD (acronym for Mean Texture Depth) of the ground in millimeter scale, and the function for obtaining the water level takes the road texture into account.
[0048] It is one of the parameters that can affect the water level assessment on a road. For this purpose, the macrotexture of the ground, i.e. the texture on a millimeter scale, is sufficient to certify the road with regard to its interaction with water. The MTD can be evaluated according to the standard NF EN 13036-1.
[0049] The road texture can be obtained using a road texture mapping associated with the vehicle's position, for example through a GPS (Global Positioning System) indication. It is also possible to use a device as shown in patent application WO2015092253A1, which also includes a microphone on the vehicle, and which allows the road texture state to be associated with a certain class. Since the influence of the road texture on the assessment of the water level is secondary, the qualitative and non-quantitative information on the road texture is sufficient to incorporate a certain correction for each texture class for the assessment of the water level. Finally, the step of obtaining the road texture can correspond to the analysis of a noise spectrum recorded in a frequency band lower than 4 kHz, for example 200-2000 Hz or 500-1000 Hz, by a microphone mounted on the vehicle as close as possible to the road, for example in the wheel arch or bumper. This noise spectrum is expressed, for example, in terms of an average level expressed using an RMS (root mean square) value. Depending on the qualitative level of this average level and on the knowledge of the vehicle speed, it is possible to infer from them the class of road texture, even on wet ground.
[0050] According to a specific embodiment, the fixing of the sensor is performed in the wheel arch, more specifically at the rear of the wheel arch with respect to the plane OYZ of the mounted set, preferably in an angle sector lying between 270° and 315° with respect to the axis Y of the geometric coordinate system of the tire.
[0051] The term geometric coordinate system of the tire, named OXYZ, is understood here to mean a coordinate system associated with the tire, composed of a centre O located at the intersection of the tire's natural axis of rotation and its median plane, with an axis Z, i.e. the vertical axis, corresponding to a direction perpendicular to the ground on which the tire runs, an axis X, called the longitudinal axis and oriented in the direction of movement of the tire as it runs on a straight line parallel to the ground, at right angles to axis Z, and a Y, i.e. the transverse axis, oriented according to the natural axes of the tire, at right angles to axes X and Z, and comprising a fixed Cartesian coordinate system associated with the tire.
[0052] This preferred positioning of the microphone has the advantage of positioning it far from the zone where water droplets impinge on the wall of the wheel arch, while being close to this and other acoustic sources. This distance makes it possible to guarantee the physical integrity of the sensor, especially in the case of gravel or solid objects expelled by the dimple pattern of the tread. The proximity to other acoustic sources makes it possible not to mask their response to the impact of water droplets.
[0053] The invention will be better understood on reading the following description, which application is given by way of example only and with reference to the accompanying drawings, in which: [Brief description of the drawings]
[0054] [Figure 1] FIG. 2 is a block diagram of the steps of the method for obtaining the water level. [Figure 2a] FIG. 13 shows the acoustic response of the same vehicle equipped with tires with different characteristics and traveling on the same water level. [Figure 2b] FIG. 13 shows the acoustic response of the same vehicle equipped with tires of different wear conditions and traveling on the same water level. [Figure 2c] FIG. 13 shows the acoustic response of the same vehicle traveling on the same water level equipped with identical tires but traveling on two different terrains. [Figure 3a] FIG. 1 shows the acoustic response of a vehicle equipped with the same tires traveling on the same water level at two different driving speeds. [Figure 3b] FIG. 13 illustrates the acoustic response of a vehicle equipped with the same tires and traveling at the same travel speed through different water levels. [Figure 4] FIG. 13 illustrates a rectified frequency signal with block identification. [Diagram 5] FIG. 1 illustrates a method for identifying blocks across a rectified frequency spectrum. [Figure 6] FIG. 2 illustrates the steps of obtaining a road texture class from sound volume. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0055] Figure 1 shows a block diagram of a method 1000 for assessing the water level on a road. It is divided into three main phases. A first learning phase consists in identifying certain parameters. A second phase of measurements on the vehicle consists in extracting information useful for the assessment of the water level on the road on which the vehicle is traveling. Finally, a third phase of assessment consists in determining the water level from the information collected in the previous two phases.
[0056] The learning phase consists firstly in identifying 1001 frequency bands sensitive to the water level in the acoustic response of the vehicle. This stage consists in obtaining noise spectra measured with a microphone fixed on the vehicle when the vehicle with a reference mounting set drives on roads with different known water levels. To know the water level, it is sufficient to use laboratory type measuring means known from the state of the art, which will serve as a benchmark. The rough nature of the road on which the vehicle drives is not important, especially at the millimeter scale. Neither is the nature and condition of the tires the vehicle is equipped with. In fact, these parameters are of secondary importance with respect to the acoustic response of the vehicle to the water level on the road.
[0057] These different spectra firstly and optionally make it possible to determine the slope of the characteristic curve above 4 kHz according to step 1002. For this determination, a simple linear regression of the spectrum is carried out over the complete frequency band of the spectrum or for each section of the frequency band. This determination can be carried out according to a single driving condition of the vehicle or by averaging according to the different driving conditions tested.
[0058] Measurements are taken under various water level conditions to identify frequency bands that show strong response sensitivity, particularly to the amplitude level of the acoustic response, which is dependent on the water level on the road. The results of this step 1001 will be retained for the evaluation phase.
[0059] Finally, since the variations as a function of the water level are observed, but also mainly as a function of the driving speed, it is possible to qualify the sensitivity of each sensitive frequency band to these two parameters according to step 1003 by using the energy criterion of step 3003. Secondly, in a more sophisticated water level evaluation mode, it is useful to qualify the sensitivity of the frequency bands as a function of the tire characteristics and the tire wear state. Thirdly, in an ideal mode, it is useful to evaluate the sensitivity of the frequency bands as a function of the texture of the road on which the tire is driving.
[0060] A multilinear function of various parameters is then generated, including at least the water level and the driving speed, which successively describes the energy vector of step 3003 and passes through the points measured in this step 1003. This function will be used in step 3006 to determine the water level on the road as a function of the vehicle's driving speed and the energy vector associated with the vehicle's acoustic response.
[0061] Finally, step 1003 requires the determination of the conditions of the energy criterion of step 3003 in order to select the blocks useful for the analysis. As a result, step 1004 is executed which consists in selecting characteristic values and threshold values of the various blocks for the vehicle frequency signal, for example for the subsequent rectified frequency signal, which will be used in the evaluation phase in search mode.
[0062] The second phase consists in acquiring information about the vehicle while it is traveling on any road in wet conditions where the water level is not known.
[0063] In a first optional step 2002, it is advisable to identify the tires the vehicle is equipped with and to obtain information about their wear state. However, these data are auxiliary and are only necessary to refine the final assessment of the water level. On the other hand, it is necessary to estimate the travel speed of the vehicle at the moment of measurement on the wet road according to step 2001. The average value of the speed over the acquisition time of the acoustic signal is sufficient.
[0064] The third phase is an evaluation phase that is carried out embedded in the vehicle through a processor with memory space and microphones or communication means with other parts of the vehicle, in particular the tires. These communication means can also be, for example, radio frequency transceivers for communicating with an RFID tag associated with the tire, a TMS sensor provided in the tire, or a TPMS sensor provided in the mounting set. Also, these communication means can be a microphone or a wired link with the vehicle's CAN bus. Thus, on the one hand, the processor can extract all the useful information in the learning phase, such as, for example, sensitive frequency bands, slopes of the characteristic curves, characteristic values, and thresholds for each sensitive frequency band. On the other hand, the processor can extract not only the running speed of the vehicle, but also potentially the wear state of the tires, for example according to the mileage they have traveled, together with the ID or geometrical characteristics of the tires equipped on the vehicle.
[0065] A first optional step of the evaluation phase consists in generating a rectified frequency signal. For this purpose, it is preferable to combine the frequency signal of the measurement phase with a characteristic curve of the vehicle. However, the characteristic curve is determined by two parameters: on the one hand, the characteristic curve passes through an amplitude level at a frequency of 4 kHz which is a function of the water level and the driving speed, and on the other hand, the slope of the characteristic curve is independent of the water level. This second parameter was evaluated in the learning phase of step 1002. It is therefore appropriate to determine the amplitude level at a given frequency, for example 4 kHz, from the frequency signal obtained in step 2003. For this purpose, one way consists in linearizing the frequency signal around the sampling frequency of the frequency signal. Another way consists in determining this amplitude as the average value of the frequency signal over a frequency band around a target frequency of 4 kHz.
[0066] Depending on whether the work is performed on a linear or logarithmic scale, the comparison consists either in establishing the ratio between the frequency signal and the characteristic curve or in establishing the difference between the two curves. That is to say, according to step 3001 a rectified frequency signal is obtained.
[0067] A second step 3002 consists in identifying blocks of frequency signals above 4 kHz, optionally rectified frequency signals. To do so, the frequency signal is sampled and each of the samples must be compared with a value. For the rectified frequency signal, this specific value is the value 1, if the work is done on a linear scale. If the work is done on a logarithmic scale, this specific value is the value zero. A simple comparator between successive samples makes it possible to identify a passage of the rectified frequency signal to this specific value. If this passage is reflected by an overshoot above the specific value, it corresponds to the first point of the block. The end point of the block is determined by a new passage of the rectified frequency signal below the specific value. That is to say, it is possible to identify all blocks of the rectified frequency signal which are the response of the rectified frequency signal exceeding the specific value.
[0068] Then, step 3003 consists in qualifying the identified blocks. The aim is to keep only those blocks that have a strong sensitivity to the water level and whose acoustic response is significant with respect to the acoustic response of the vehicle. Obviously, for this second embodiment, it is easier to focus on the rectified frequency signal so as not to exclude high frequency blocks.
[0069] This stage 3003 consists first in assigning to the block a metric that is representative of the block's acoustic energy while avoiding overemphasizing very local phenomena in the frequency range. For this purpose, the maximum value of the block over its frequency range is the first indicator. However, it does not necessarily reflect the shape of the block well, which is why in certain embodiments the mean or median value of the block is preferred over it.
[0070] This metric then needs to be positioned relative to a threshold in order to select only those blocks that have a strong contribution to the acoustic response of the vehicle, which makes it possible to increase the sensitivity of the method and to increase its separation capabilities. For this purpose, it is useful and important to use as threshold a value of at least 3 dB, which represents twice the acoustic power of the block, when the representation of the acoustic response of the vehicle is made in a logarithmic scale. Obviously, this threshold can be further increased to reduce the analysis to a small number of blocks of the acoustic response and increase the separation or differentiation capabilities of the method.
[0071] The next step 3004 is the selection of the most significant blocks from step 3003 in terms of acoustic power associated with the frequency bands sensitive to the water level identified in step 1001. Thus, only those frequency bands of blocks that are both audible with respect to the vehicle response and sensitive to the water level are kept. Ideally, only a single block is kept, but it is entirely possible to select multiple blocks. At the end of this step 3004, that part of the signal that will be used to determine the water level has been selected.
[0072] Step 3005 consists in generating an energy vector from the portion of the signal selected in the previous step. In the framework of the invention, the power spectral density is fully fully represented in terms of the frequency signal. The energy vector then contains the integral function of this power spectral density over the portion of the signal, whether this is a frequency signal obtained from a microphone or a rectified frequency signal. In fact, this integral function is an indicator of the acoustic energy associated with the frequency band of the acoustic response. Obviously, if the portion of the frequency signal is subdivided into several sections, it is possible to calculate for each successive section of frequency the integral function of the acoustic response over the band of frequencies that delimit each section. Each integral function can then be an element of the energy vector. It is also possible to combine the various elements of the energy vector, assigning them each a specific weight, in order to generate only a single element. The integral function of the portion of the frequency signal will be evaluated on the representation of the acoustic signal on a linear scale.
[0073] Finally, step 3006 consists in estimating the water level on the road on which the vehicle is traveling from the energy vector of step 3005 and the travel speed of the vehicle obtained in step 2001. Indeed, in step 1003, the quantification of the different frequency bands sensitive to the water level is carried out for any road by varying at least the water level and the travel speed of the vehicle, making it possible to identify a function that continuously describes the energy vector identified in step 3005. This learning phase, step 1003, makes it possible to estimate the water level on the road from the travel speed and the energy vector identified in step 3005 using at least a bilinear function of the energy vector and the travel speed parameters.
[0074] FIG. 2a shows the acoustic response of a vehicle traveling on a road with a defined water level under equally stabilized driving conditions such as travel speed, tire pressure, and load. The dotted curve 2101 represents the response of a first tire characterized by a special pattern of tread indentations of the "all season" type, called V-design. The continuous curve 2102 represents the acoustic response of a second tire, whose indentation pattern differs from the first tire, due to its "summer" type design with four longitudinal grooves across the width of the tread. The acoustic responses of the two tires are very similar. However, based on the frequency band, different behaviors are observed between the two tires. Indeed, in a wide frequency band from 4 to 15 kHz, the summer type tire emits more noise than the all season type tire. Then, above 15 kHz, the all season type tire becomes noisier than the summer type tire. These variations in the acoustic behavior of the two tires are clear, but remain negligible when the travel speed or the water level on the road changes.
[0075] Figure 2b shows the effect of tread wear level for the same tire running on the same wet ground under equal driving conditions. The dotted curve 2201 shows the response for a first wear state. The continuous curve 2202 shows the acoustic response for a second wear state of the tire. Again, there is little dynamics linked to the tire wear level. Furthermore, the sensitivity as a function of frequency band is inverted, especially around 15kHz.
[0076] Figure 2c represents the effect of the ground on the acoustic response for the same tire running on the same water level under equal driving conditions. The dotted curve 2301 represents the response of the tire on a first ground surface. The continuous curve 2302 represents the acoustic response of the tire running on a second ground surface. This time the dynamics linked to the nature of the ground are small.
[0077] Fig. 3a shows the acoustic response of vehicles equipped with the same tires, in particular tires with the same wear level, when they are traveling on the same wet road, i.e. on the same water level. However, the continuous curves 3101 and 3102 show the acoustic response of the vehicle when it is traveling on two different water levels at 30 km / h, whereas the dotted curves 3103 and 3104 show the acoustic response when it is traveling on the same two water levels at 100 km / h. There is a very strong difference between these acoustic responses as a function of the variable traveling speed, which is even more significant than the case of the water level change, making it essential to take into account at least this parameter to estimate the water level on the road.
[0078] Figure 3b shows the acoustic responses under equal driving conditions for vehicles equipped with the same tires. However, the road on which the vehicles travel has different water levels ranging from 300 to 1500 micrometers between the various acoustic responses. These water levels were measured using a laboratory type measuring device. This is for example the MARWIS 8900.U03 device from the company LUFT.
[0079] The increasing order of numbering of the response curves between 3201 and 3204 corresponds to the increasing order of the water level on the road in steps of about 375 micrometers between 300 and 1500 micrometers. It should be noted that these curves are classified in terms of the amplitude level according to the water level on the road. The deviation between the curves is significant from 4 kHz. The higher the observed frequency of the acoustic response, the greater the deviation between the responses, with the crossing of the levels from 7 to 10 kHz being particularly notable. Indeed, from 4 to 7 kHz, the dynamics according to the water level are less than 5 dB, from 7 to 10 kHz, more than 5 dB and less than 7 dB. Above 10 kHz, we obtain the dynamics between the extreme curves that can reach and exceed 10 dB.
[0080] Fig. 4 shows the construction of a rectified frequency signal from the frequency signals according to a first preferred embodiment. After identifying the slope of the characteristic curve of the acoustic response of the vehicle according to step 1002, in step 2003 a characteristic curve of the frequency signal associated with the driving speed is constructed from the recorded frequency signals. For this purpose a reference point of this characteristic curve or a number of points if it is defined piecewise must be identified. This reference point corresponds to the value for a given frequency of the characteristic curve associated with the driving speed. This point is preferably taken in a frequency band below 7 kHz, since the response dynamics are low for low frequencies. Furthermore it is also possible to take the average value of the frequency signal over the frequency bands insensitive to the water level identified in step 1001 of the learning phase. The graph first shows curve 4101 which is the frequency signal measured in step 2003 and a dotted characteristic curve 4102 associated with the frequency signal, the latter here being a linear curve whose slope over the frequency band 4 kHz to 22 kHz was identified in step 1002, whereas the 4 kHz reference point was evaluated according to step 1001 as the median value over a frequency band that is insensitive to the water level around 4 kHz. The rectified frequency signal represented by the thick grey curve 4103 is obtained by the difference on a logarithmic scale between frequency signal 4101 and characteristic curve 4102.
[0081] FIG. 5 represents a rectified frequency signal 5101. The latter oscillates around a zero value in a logarithmic representation. The first step consists in identifying positive blocks 5102 in this rectified frequency signal. Indeed, these so-called positive blocks are important energy contributors of the acoustic response. For this, each block located above a zero value separated by two frequencies corresponding to a zero value crossing 5103 is isolated. The isolated blocks must then be qualified through a simple metric 5104 representative of the acoustic energy of the block. For this purpose, the mean or median value of the isolated blocks is a good estimator of the acoustic energy that is not at all concerned by the shape of the block. The maximum value of the block is an indicator of the energy that is easy to implement, but it does not correlate very well with the total energy of the block. Finally, this estimator of the blocks must be compared with a threshold in order to keep only the blocks with the highest energy. For this purpose, a threshold value 3 dB higher than the zero value is a good indicator that allows to keep only a small number of blocks. By keeping only these blocks 5105, there is a guarantee that they have a strong dynamics for classification, which increases the discrimination capabilities of the method.
[0082] After identifying the most energetic block, this selection must be combined with the block that is most sensitive to the water level. The combination of these two blocks allows to identify the block that is most likely to identify the water level on the ground, since it is the most sensitive to the water level and at the same time certainly improves the discrimination function of the method. However, this block is also one of the most energetic blocks that allows to identify the complete acoustic spectrum. That is to say, an effective metric is constructed to identify the water level on the road, especially for water levels below 1.5 mm. This effective metric is the integral of the corresponding part of the frequency signal on a linear scale separated by the frequency band of the most energetic and sensitive to the water level block.
[0083] Finally, having evaluated this metric and having identified the travel speed of the vehicle at which the frequency signal was obtained, the water level on the road must be identified using a bilinear function linking the water level on the road identified in this stage 1003, the effective metric of the frequency signal and the travel speed of the vehicle. This bilinear function is for example the product of two functions with independent parameters: on the one hand the travel speed and, on the other hand, the effective metric of the integral of that part of the frequency signal on a linear scale. The result of this product is directly the water level on the road.
[0084] Figure 6 shows a representation of the sound level in decibel scale using a frequency pitch of several tens of Hertz. This type of representation makes it possible to highlight the overall sound level while limiting the modes isolated from mechanical parts, whether the vehicle or the mounting set. In this case, the recordings correspond to the same driving speed of a vehicle moving in a straight line. In order to limit the influence of water on the road, the microphone measurement is made at the rear of the vehicle, for example at the wheel arch of the second axle of the vehicle. In this case, the water level remaining on the road after the passage of the first axle of the vehicle is limited, and this parameter has a negligible influence on the sound level recorded by the microphone. Of course, measurements at the front axle are also intended, as long as the water level on the road is reasonable. The qualification of the road texture is appropriate for low water levels of less than 2 millimeters.
[0085] Curves 6101 to 6104 correspond to the acoustic signatures of different ground surfaces, where the MDT (acronym for Mean Texture Depth) varies from less than 0.1 mm, corresponding to a smooth ground, to 2.6 mm for a very rough ground, with two intermediate steps. At low frequencies, the curves are strongly differentiated. From 2 kHz onwards, the acoustic signature of the road is more influenced overall by the road texture.
[0086] For each curve, the RMS (Root Mean Square) value of the acoustic signature is calculated over frequency bands sensitive to the road texture; i.e. for a given speed, the RMS value is related to a class of road texture defined by the MTD. Here, the RMS value is evaluated over two frequency bands: a wide window from 200 Hz to 2 kHz and a narrow window from 500 Hz to 1 kHz. For any analyzed frequency band, the classification of the acoustic signature by the RMS value remains unchanged. Also, the deviations between the RMS values are significant and sufficiently discriminatory. [Explanation of symbols]
[0087] 1001 Identification of sensitive frequency bands 1002 Determination of the slope of the characteristic curve 1003 Certification of sensitivity for each frequency band 1004 Selection of characteristic values and thresholds 2001 Acquisition of driving speed
Claims
1. 1. A method for estimating a water level on a roadway on which a tire of a vehicle is traveling, comprising the steps of: The mounted set of the vehicle is disposed in a wheel arch of the vehicle; Fixing a sensor on a vehicle; obtaining from said sensor a frequency signal corresponding to the movement of a vehicle at a speed V on a road covered with water; isolating a portion of said frequency signal that is sensitive to water level, the portion being bounded by two frequencies F1 and F2, F1 being lower than F2; determining an energy vector linked to said portion of said frequency signal; obtaining the water level on the road using a function that takes into account the energy vector and the velocities V of both front wheels; Including, the sensor is a microphone; the portion of the frequency signal extends generally above 4 kHz; A method comprising:
2. 2. A method for estimating the water level on a road on which tires of a mounted set are traveling, as claimed in claim 1, characterized in that the frequency signal is the power spectral density, called PSD, of a discretized time signal from the microphone obtained while the vehicle is traveling.
3. 3. The method of claim 2, wherein the energy vector comprises an integral function of the PSD between the frequencies F1 and F2.
4. 4. A method for estimating the water level on a road on which a tire of a mounted set is traveling, according to claims 1 to 3, characterized in that the portion of the frequency signal is limited to at least one block for which a characteristic value for the frequency signal is above a threshold value, the characteristic value being a maximum value, an average value or a median value.
5. 5. A method for estimating the water level on a road on which tires of a mounted set are traveling, according to any one of claims 1 to 4, characterized in that the step of isolating a portion of the frequency signal comprises constructing a rectified frequency signal as a comparison between the frequency signal and a characteristic curve of the frequency signal.
6. 6. The method of estimating the water level on a road on which tires of a mounted set are traveling, as claimed in claim 5, characterized in that the characteristic curve of the frequency signal decreases continuously above 4 kHz.
7. 7. A method for estimating a water level on a road on which tires of a mounted set are traveling, according to any one of claims 5 and 6 in combination with claim 4, characterized in that the at least one block of frequency signals is delimited by crossings of specific values by the rectified frequency signal.
8. 5. A method for estimating the water level on a road on which a tire of a mounted set is traveling, according to claim 4, characterized in that the frequency signal is expressed on a logarithmic scale and the threshold value is a value of 3 dB, and according to any one of claims 5 to 7 in combination with claim 4.
9. 9. Method for estimating the water level on a road on which a tire of a mounted set according to any one of claims 1 to 8 is running, characterized in that the sensitivity stage of the frequency band of the frequency signal to the water level is achieved in a learning stage during which the vehicle equipped with the microphone runs on a road having at least two water levels which can be differentiated.
10. the tire includes an identification vector; The method includes obtaining an identification number of the tire to obtain at least one geometric characteristic of the tire; the function for obtaining the water level takes into account the at least one geometric characteristic of the tire. Method for estimating the water level on a road on which a tire of a mounted set according to any one of claims 1 to 9 is traveling.
11. The method includes obtaining a state of wear of the tire; the function for obtaining the water level takes into account the state of wear of the tire. Method for estimating the water level on a road on which a tire of a mounted set according to any one of claims 1 to 10 is traveling.
12. The method includes obtaining a texture of the road; Method for estimating the water level on a road on which a tire of a mounted set according to any one of claims 1 to 11 is traveling.
13. 13. A method for estimating the water level on a road on which a tire of a mounted set according to any one of claims 1 to 12 is running, characterized in that the step of fixing the sensor is performed in the wheel arch.
14. 14. The method for estimating the water level on a road on which tires of a mounted set are traveling, as claimed in claim 13, characterized in that the fixing step of the sensor is performed rearward of the wheel arch with respect to a plane OYZ of the mounted set.
Citation Information
Patent Citations
Road surface condition measuring device
JP1994138018A
Detecting apparatus for road surface condition
JP1994174543A
Method for estimating speed generating hydroplaning phenomenon and method for detecting generation area thereof
JP1996184533A
Road surface state detecting device
JP1996298613A
Sound diagnostic device and sound diagnostic method
JP2002323371A
