KR102415195B1 - Apparatus for estimating of road surface type using atmospheric attenuation corrected sound wave and method thereof - Google Patents

Abstract

The present invention relates to an apparatus and method for estimating the type of road surface using sound waves in which the amount of atmospheric attenuation is corrected, and more particularly, to transmit a sound wave signal to receive a reflected signal, and to correct the attenuation amount of the sound wave in the air with respect to the received signal Then, there is provided an apparatus and method for estimating the type of road using sound waves with the amount of atmospheric attenuation corrected for estimating the type of the corresponding road surface by learning the attenuation amount-corrected signal using an artificial neural network.

Description

Apparatus and method for estimating road surface type using sound waves corrected for atmospheric attenuation

The present invention relates to an apparatus and method for estimating the type of road surface using sound waves in which the amount of atmospheric attenuation is corrected, and more particularly, to transmit a sound wave signal to receive a reflected signal, and to correct the attenuation amount of the sound wave in the air with respect to the received signal Then, the present invention relates to an apparatus and method for estimating a road surface type using a sound wave having an atmospheric attenuation corrected amount, for estimating the type of the corresponding road surface by learning the attenuation amount corrected signal using an artificial neural network.

In the United States, approximately 6 million car accidents occur each year. In particular, an average of more than 1.2 million accidents occur annually in the United States alone, and more than 5,000 people die every year from these accidents. Accordingly, the slippery road surface is emerging as a social problem that places a great burden on the state and insurance companies.

Korea is no exception. Black ice accidents occur every winter, 6,500 cases have occurred in the last 5 years, and the death toll is about 200.

In Korea, the snow melting process consists of four stages, and the transmission of all these processes is passively transmitted by humans, and it takes several hours to reach the final stage. In the case of the last black ice accident on the Sangju-Yeongcheon Expressway, the official in charge checked the forecast but did not request snowmelting in advance.

This accident is due to the absence of a sensor capable of detecting the state of the road surface in real time and a notification system using the same.

On the other hand, there are two major methods that have been proposed so far as a method for estimating the type of road surface or the friction coefficient of the road surface.

One is a method to find out by causing physical friction in a direct method, and the other is a method using radio waves or sound waves in an indirect method.

First, the direct method includes a wheel-based method and a vehicle dynamics-based method.

What these two methods have in common is to apply the brake regardless of the driver's intention to estimate the friction coefficient of the road to the degree of deceleration of the vehicle.

In this method, the friction coefficient can be known only by stepping on the relevant road surface, and a vehicle dynamics model is essential. In addition, there is a limit that affects the riding comfort and fuel economy because the brake must be operated.

On the other hand, indirect methods include methods using electromagnetic waves and sound waves. Both methods do not require physical friction, so they can be used independently for the vehicle dynamics model, and the advantage of knowing the road surface condition in advance without contact with the corresponding road surface There is this.

The electromagnetic wave-based method uses a camera, lidar, radar, etc., and although it can see quite far, the accuracy is relatively low, and the price of the system is formed in hundreds to thousands of dollars, so it is very expensive. In addition, there is a difficulty in creating a data set because the amount of data is large.

Also, there is a method using an image sensor, which is one of the electromagnetic wave methods, and the type of road surface is classified using a camera, which is an image sensor. However, in the case of an image sensor, it is greatly affected by the illuminance, and as black ice is invisible to the human eye, the vision sensor cannot detect it. Due to this limitation, the average performance of the method using the image sensor is less than 70%, and the cost is relatively high because not only the image sensor but also a high-performance image processing device is required.

Black ice is invisible to the human eye and cannot be detected by an image sensor. That is, the image sensor method has a limitation in not being able to distinguish various types of road surfaces.

On the other hand, as a method using the sound wave, there are a passive sensing method and an active sensing method. In the passive sensing method, noise generated from a tire for a certain period of time is recorded with a microphone and then classified using an artificial neural network. There is this. In this way, the type of road can be known only for the road on which the tire has passed, and there is a limitation that the classification takes a long time. On the other hand, among active sensing methods, there is a study for classifying a reflected road surface using only distance and reflection intensity information, which has a disadvantage in that it is weak against ambient disturbance.

Meanwhile, on the road these days, there are a lot of electric cars and hydrogen cars with sky-blue license plates. According to Bloomberg, more than half of vehicles sold by 2040 will be electric, and a third of vehicles on the road will be electric.

To increase the mileage, which is the driving range of electric vehicles, vehicle manufacturers are actively using regenerative braking. However, it is not difficult to hear the news that the car slipped due to regenerative braking and caused an accident. The regenerative braking causes the rear wheel to lose traction or the wheel locks.

If you look at Tesla's current regenerative braking settings, you can change it to standard and low, but there is no option to turn it off. The reason for such an accident is that the control period of the regenerative braking module is 100ms longer than other systems, and the reason the manufacturer does not allow the regenerative braking function to be turned off is that the vehicle's moving distance is rapidly decreasing.

It would be best to apply regenerative braking depending on the condition of the driving road surface, but it can be said that this problem occurs because the type of road surface cannot be known in advance before the vehicle passes.

On the other hand, when the distance (interval) between transmitting and receiving sound waves is long, the transmitted sound wave signal and the received sound wave signal are transmitted through the atmosphere (air) and attenuation occurs naturally. Accordingly, there is a problem in that the accuracy of the system for determining the road surface condition is deteriorated when the attenuated received sound wave signal is used.

Korean Patent Registration [10-1716270] discloses an apparatus and method for estimating a front ground state.

Korean Patent Registration [10-2136576] discloses a road surface condition measurement device and a road surface condition diagnosis system.

Korean Patent Registration [10-2156295] discloses a method and apparatus for estimating a road surface type using an ultrasonic signal.

Korean Patent Registration [10-2207816] discloses a method and apparatus for measuring a road surface condition using sensor data.

Korean Patent Registration [10-1716270] (Registration Date: 2017.03.08) Korean Patent [10-2136576] (Registration Date: 2020. 07. 16) Korean Patent [10-2156295] (Registration Date: 2020. 09. 09) Korean Patent [10-2207816] (Registration Date: 2021. 01. 20)

Accordingly, the present invention has been devised to solve the above problems, and an object of the present invention is to transmit a sound wave signal, receive a reflected signal, and correct the attenuation amount of the sound wave in the air for the received signal, To provide an apparatus and method for estimating a road surface type using sound waves with a corrected amount of atmospheric attenuation for estimating the type of the corresponding road surface by learning the attenuation amount-corrected signal using an artificial neural network.

The purpose of the embodiments of the present invention is not limited to the above-mentioned purpose, and other objects not mentioned will be clearly understood by those of ordinary skill in the art to which the present invention belongs from the following description. .

In order to achieve the above object, an apparatus for estimating a road surface type using a sound wave having an atmospheric attenuation amount corrected according to an embodiment of the present invention transmits a sound wave signal to a corresponding road surface for which the type is to be known, and then collects the reflected signal. Sound wave transceiver 100 for receiving; an analog-to-digital converter 200 for converting an analog signal of the received signal into a digital signal; an atmospheric attenuation correction unit 300 for calculating and correcting the atmospheric attenuation amount of the digital signal; an artificial neural network 400 that uses the corrected digital signal as an input signal and classifies the input signal based on the learned road surface classification model to estimate the type of road surface; and a control unit (MCU) 500 for controlling the operation of the sound wave transceiver, the analog-to-digital converter, the atmospheric attenuation correction unit, and the artificial neural network.

In addition, in the method for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to an embodiment of the present invention for achieving the above object, the sound wave signal is transmitted to the corresponding road surface for which the type is to be known under the control of the controller. After transmitting, a reflected signal receiving step of receiving the reflected signal (S20); a signal conversion step (S30) of converting an analog signal into a digital signal for a preset region of the received signal under the control of the controller; an atmospheric attenuation correction step (S40) of calculating and correcting the atmospheric attenuation amount of the digital signal under the control of the control unit; a frequency domain signal obtaining step of obtaining a frequency domain signal by performing frequency conversion on a preset region of the corrected digital signal under the control of the controller (S50); and a first road surface that uses the frequency domain signal as an input signal of a neural network under the control of the controller, extracts and classifies characteristics of the input signal based on a learned road surface classification model, and determines the type of the road surface It is characterized in that it includes a type determination step (S60).

In addition, in the method for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to another embodiment of the present invention for achieving the above object, the sound wave signal is transmitted to the corresponding road surface for which the type is to be known under the control of the controller. After transmitting, a reflected signal receiving step of receiving the reflected signal (S20); a signal conversion step (S30) of converting an analog signal into a digital signal for a preset region of the received signal under the control of the controller; an atmospheric attenuation correction step of calculating and correcting the atmospheric attenuation amount of the digital signal (S40); a convolution operation step (S70) of receiving a digital signal with a corrected amount of attenuation in the air from the artificial neural network under the control of the control unit and performing a convolution operation multiple times; and a second road surface type determination step of extracting and classifying characteristics of the convolutional signal based on the road surface classification model learned from the artificial neural network under the control of the controller and determining the type of the road surface (S80) ) is characterized in that it contains.

In addition, according to an embodiment of the present invention, there is provided a computer-readable recording medium storing a program for implementing the method for estimating the type of road using the sound wave in which the atmospheric attenuation amount is corrected.

In addition, according to an embodiment of the present invention, a program stored in a computer-readable recording medium is provided to implement the method for estimating the type of road using the sound wave in which the atmospheric attenuation amount is corrected.

According to the apparatus and method for estimating the type of road surface using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention, a sound wave signal is transmitted and a reflected signal is received, and the attenuation amount of the atmospheric sound wave is corrected for the received signal. Then, it is possible to estimate the type of the road surface by learning the attenuation amount-corrected signal using an artificial neural network, so that the type of road surface can be determined with high accuracy through an uncomplicated configuration.

In addition, according to the apparatus and method for estimating the type of road using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention, atmospheric information (temperature, humidity, atmospheric pressure, etc.) is collected in real time and the received signal is in standby mode. Since the attenuation amount of the sound wave can be corrected immediately, there is an effect of accurately analyzing the reflected signal according to the type of road surface according to the actual signal.

In addition, according to the apparatus and method for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to an embodiment of the present invention, the type or condition of the road surface can be determined with high accuracy, so that the black ice on which thin ice is formed Since it is possible to quickly check whether or not it is generated, there is an effect of being able to make a prediction in advance for the driver who operates the relevant section.

In addition, according to the apparatus and method for estimating the type of road using sound waves with the amount of atmospheric attenuation corrected according to an embodiment of the present invention, the apparatus for estimating the type of road surface can be implemented using an inexpensive 40 kHz ultrasonic sensor and a light algorithm, The processing speed is fast and it has an economical effect.

1 is a block diagram of an apparatus for estimating a road surface type using sound waves in which an atmospheric attenuation amount is corrected according to an embodiment of the present invention;
FIG. 2 is a view for explaining a transmission signal and a reception signal in an apparatus for estimating a road surface type using a sound wave in which an atmospheric attenuation amount is corrected according to an embodiment of the present invention; FIG.
3 is a configuration diagram of an artificial neural network in an apparatus for estimating a road surface type using sound waves with a corrected atmospheric attenuation amount according to an embodiment of the present invention;
4 is a view for explaining a signal converter in the apparatus for estimating the type of road surface using sound waves according to an embodiment of the present invention;
5 is a block diagram of another embodiment of an artificial neural network in the apparatus for estimating the type of road using sound waves with the amount of atmospheric attenuation corrected according to the present invention;
FIG. 6 is a view for explaining an operation of the convolution performing unit of FIG. 5;
7 is a view for explaining a code of a convolution performing unit in the apparatus for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to the present invention;
8A is a table showing a test result in the case where the atmospheric attenuation amount is not corrected in the apparatus for estimating the type of road using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention;
FIG. 8B is a table showing test results when the atmospheric attenuation amount is corrected in the apparatus for estimating the type of road using sound waves with the atmospheric attenuation amount corrected according to an embodiment of the present invention; FIG.
9 is a flowchart illustrating a method for estimating the type of road surface using sound waves in which atmospheric attenuation is corrected according to an embodiment of the present invention;
10 is a flowchart of another embodiment of a method for estimating a road surface type using a sound wave in which an atmospheric attenuation amount is corrected according to the present invention;

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, and it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

When a component is referred to as being “connected” or “connected” to another component, it may be directly connected or connected to the other component, but it is understood that other components may exist in between. it should be

On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle.

The terms used herein are used only to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as "comprise" or "have" are intended to designate that a feature, number, process, operation, component, part, or a combination thereof described in the specification exists, but one or more other features It is to be understood that this does not preclude the existence or addition of numbers, processes, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having meanings consistent with the meanings in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings. Prior to this, the terms or words used in the present specification and claims should not be construed as being limited to conventional or dictionary meanings, and the inventor should properly understand the concept of the term in order to best describe his invention. Based on the principle that can be defined, it should be interpreted as meaning and concept consistent with the technical idea of the present invention. In addition, if there is no other definition in the technical terms and scientific terms used, it has the meaning commonly understood by those of ordinary skill in the art to which this invention belongs, and the summary of the present invention in the following description and accompanying drawings Descriptions of known functions and configurations that may be unnecessarily obscure will be omitted. The drawings introduced below are provided as examples so that the spirit of the present invention can be sufficiently conveyed to those skilled in the art. Accordingly, the present invention is not limited to the drawings presented below and may be embodied in other forms. Also, like reference numerals refer to like elements throughout. It should be noted that the same components in the drawings are denoted by the same reference numerals wherever possible.

1 is a block diagram of an apparatus for estimating a road surface type using sound waves in which an atmospheric attenuation amount is corrected according to an embodiment of the present invention.

As shown in FIG. 1 , the apparatus for estimating the type of road using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention includes a sound wave transceiver 100 , an analog-digital converter (ADC), and air. It includes an attenuation correction unit 300 , an artificial neural network 400 , and a control unit (MCU) 500 .

The sound wave transceiver 100 transmits a sound wave signal to the corresponding road surface for which the type is to be known, and then receives the reflected signal.

The sound wave transceiver 100 includes a sound wave transmitter 101 that outputs a transmission signal under the control of the control unit 500, and a sound wave receiver 102 that receives a reflected signal from which the transmitted signal is reflected back on an arbitrary surface. ) is included.

The analog-to-digital converter 200 converts the analog signal of the received signal into a digital signal.

The atmospheric attenuation correction unit 300 calculates and corrects the atmospheric attenuation amount of the digital signal.

The artificial neural network 400 uses the corrected digital signal as an input signal, classifies the input signal based on the learned road surface classification model, and estimates the type of the road surface.

On the other hand, in the artificial neural network 400, decision trees, linear discriminant analysis, logistic regression classifiers, naive Bayes classifier, support vector machine (support vector) machine), nearest neighbor classifiers, and ensemble classifiers are used to classify and learn.

Decision trees include Fine tree, Medium tree, Coarse tree, All tree, and Optimizable tree, and linear discriminant analysis includes Linear discriminant, Quadratic discriminant, All discriminants, Optimizable discriminant, and naive The Naive Bayes classifier includes Gaussian Naive Bayes, Kernel Naive Bayes, All Naive Bayes, and Optimizable Naive Bayes, and the support vector machine (SVM) includes Linear SVM, Quadratic SVM, Cubic SVM, and Fine Gaussian. It includes SVM, Medium Gaussian SVM, Coarse Gaussian SVM, All SVM, and Optimizable SVM, and the nearest neighbor classifiers are Fine KNN, Medium KNN, Coarse KNN, Cosine KNN, Cubic KNN, Weighted KNN, All KNN, Includes Optimizable KNN, and ensemble classifiers include Boosted trees, Bagged trees, Subspace Discriminant, Subspace KNN, RUSBoosted trees, All Ensembles, and Optimizable Ensemble.

The control unit (MCU) 500 controls the operation of the sound wave transceiver 100 , the analog-to-digital converter 200 , the atmospheric attenuation correction unit 300 , and the artificial neural network 400 .

The analog-to-digital converter 200, the atmospheric attenuation correction unit 300, and the artificial neural network 400 express software implemented as a program as a component.

Meanwhile, although not shown in the drawings, the apparatus for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to the present invention includes a storage (memory) in which the learned road surface classification model and software implemented with the program are stored. The storage (memory) may be configured to be included in the control unit (MCU) 500 .

On the other hand, the apparatus for estimating the type of road using the sound wave in which the atmospheric attenuation amount is corrected according to the present invention further includes an atmospheric information measuring unit 600 capable of measuring the temperature, humidity, and atmospheric pressure in the atmosphere.

Under the control of the control unit 500, the atmospheric information including the temperature, humidity, and atmospheric pressure measured by the atmospheric information measurement unit 600 may be used in the atmospheric attenuation correction unit 300, and the artificial neural network ( 400) as an input.

FIG. 2 is a view for explaining a transmission signal and a reception signal in the apparatus for estimating the type of road surface using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention.

As shown in FIG. 2 , the control unit (MCU) 500 transmits a trigger signal having a preset size (v: trigger voltage) and a preset transmission period (p: transmission period) to the sound wave transceiver 100 . do.

Then, the sound wave transmitter 101 of the sound wave transceiver 100 outputs the sound wave signal 201 having 40 kHz to the corresponding road surface for which the type is to be known.

Thereafter, the sound wave receiver 102 of the sound wave transceiver 100 receives the reflected signal reflected from the road surface.

Here, the signal 202 received on the same time line as the sound wave signal 201 is a crosstalk signal of the sound wave signal transmitted by the sound wave transceiver 100 . In addition, the controller 500 determines the signal 203 for a preset time based on the point at which the amplitude of the received signal is the largest after the transmission delay as the received signal.

For example, assuming that t_0 is the point at which the amplitude of the signal received after the crosstalk signal is greatest, a total of (a+b) ms can be observed from t_0 - a [ms] to t_0 + b [ms]. , in the received signal 203 of FIG. 2, a is 0.2 and b is 5. Of course, depending on the environment or conditions, a and b can be adjusted.

Here, 10 ms, which is the time until the transmitted sound wave signal sufficiently disappears, was set as one transmission period, and the sampling frequency of the sound wave transceiver 100 was set to 1M Hz for sampling 25 times the 40 kHz sound wave frequency.

Hereinafter, with reference to [Equation 1] to [Equation 8], the process of calculating the amount of attenuation of the sound wave signal in the atmospheric attenuation correction unit 300 will be described.

First, the saturation pressure Psat is calculated using the following <Equation 1>.

[Equation 1]

Here, To1 is the triple point of the atmosphere [K], and T is the current temperature [K].

Absolute humidity (h) is calculated using the following <Equation 2>.

[Equation 2]

where hrin is relative humidity [%], Psat is saturation pressure [ unit ], and Ps is static pressure [atm].

Meanwhile, the scaled relaxation frequency for nitrogen (FrN) of nitrogen occupying 78% of the atmosphere is calculated using the following <Equation 3>.

[Equation 3]

Here, To is the reference temperature [K], and T is the current temperature [K].

Meanwhile, the scaled relaxation frequency for oxygen (FrO) of oxygen occupying 21% of the atmosphere is calculated using the following <Equation 4>.

[Equation 4]

where h is the absolute humidity.

Meanwhile, the attenuation coefficient (α: attenuation coefficient [nepers/m]] is calculated using the following <Equation 5>.

[Equation 5]

where Ps is the static pressure, F is the frequency of the sound wave signal (transmitted sound wave signal) , T is the current temperature [K], To is the reference temperature [K], FrO is the extended relaxation frequency of oxygen, and FrN is It is the extended relaxation frequency of nitrogen.

On the other hand, the attenuation ratio (A, unit: dB) of the sound wave signal is calculated using the following <Equation 6>.

[Equation 6]

Here, α is an attenuation coefficient, and d is a distance between the sound wave transmitter 100 and the corresponding road surface for which the type is to be known.

The d is calculated using the following <Equation 7> using the required time t (time of flight) from being transmitted from the transmitter until it is reflected by the road surface and sensed by the receiver and the speed of sound (Vair) in the air.

[Equation 7]

Here, t is the required time, and Vair is the speed of sound in the air [m/s].

On the other hand, the speed of sound in the air is calculated using the following <Equation 8>.

[Equation 8]

Here, Ks is the isentropic coefficient of stiffness of the object, and ρ is the density of the object (atmosphere).

At this time, assuming air (atmosphere) as an ideal gas, Ks = γP, γ is the heat capacity ratio (1.4 in the case of air), P is the pressure, R is the ideal gas constant, and T is the absolute temperature [K]. . Since it is a constant except for temperature, it can also be expressed approximately.

The speed of sound in the atmosphere may be compensated for attenuation by correcting it according to the temperature, atmospheric pressure, and humidity of the atmosphere.

3 is a configuration diagram of an artificial neural network in an apparatus for estimating a type of road using sound waves with a corrected atmospheric attenuation amount according to an embodiment of the present invention.

As shown in FIG. 3 , the artificial neural network 400 in the apparatus for estimating the type of road using sound waves in which the atmospheric attenuation amount is corrected according to the present invention includes a signal converter 410 and a neural network 420 .

The signal converter 410 obtains a frequency domain signal by performing frequency conversion on a preset region on the time domain of the received signal.

The signal converter 410 may be a Short-Time Fourier Transform (STFT) converter, a Fast Fourier Transform (FFT), a cepstruam, or a wavelet transform. The frequency domain signal may be 2D or 3D.

The neural network 420 uses the frequency domain signal as an input signal, extracts and classifies characteristics of the input signal based on the learned road surface classification model, and estimates the type of the road surface.

The neural network 420 is a deep neural network (DNN) of a multi-layer perceptron algorithm including an input layer 401 , a plurality of hidden layers 402 and an output layer 403 .

Meanwhile, the structure of the neural network 420 is not limited to the aforementioned DNN.

The input layer 401 flattens data of the frequency domain signal to receive 1D input.

Data input to the input layer 401 is feature extracted and classified through a plurality of hidden layers 402 .

The output layer 403 outputs a probability value for each type of the learned road surface.

The neural network 420 determines and outputs the type of road surface having the highest probability among the probability values output from the output layer 403 using the softmax 404 .

Meanwhile, the neural network 420 may receive atmospheric information (temperature, humidity, and atmospheric pressure information) and use it as an input of the input layer 401 .

Also, the neural network 420 may frequency-convert the sound wave signal 201 transmitted to the road surface and use it as an input of the input layer 401 .

4 is a view for explaining a signal converter in the apparatus for estimating the type of road using sound waves according to an embodiment of the present invention.

In FIG. 4 , the STFT converter is used as the signal converter 410 as an example.

As shown in FIG. 4, the STFT converter excludes the crosstalk signal 202 of the sound wave signal transmitted by the sound wave transceiver 100 among the reflected signals received in FIG. 2, and receives the preset received after the transmission delay. Short-Time Fourier Transform is performed on the signals 203 and 411 for time to obtain a spectrogram 412 .

Fourier transform may be performed on the signal 203 for one period, or Fourier transform may be performed on the received signal for multiple periods.

In the present invention, materials are classified using acoustic impedance and surface roughness information. However, the acoustic impedance is not a constant, but varies according to the frequency at which the sound wave vibrates. Therefore, analysis in the frequency domain is necessary. Time Fourier Transform, which is one of several methods for transforming a received signal in the time domain into a frequency domain signal, can be used. In order to check the FFT for every time (sampling time), the Short Time Fourier Transform (Short-Time Fourier Transform) is used.

In addition, frequency analysis is possible using not only the Short-Time Fourier Transform but also a wavelet, and the present invention has been described as using the STFT to reduce the amount of computation and secure sufficient data.

The Short-Time Fourier Transform (STFT) is a method devised to consider changes with respect to time, which cannot be solved by the existing Fourier transform. STFT is to apply a Fourier transform after dividing a long signal that changes with time into short time units.

However, STFT also has disadvantages. Since the signal is separated according to the window length, the length of the signal used for the Fourier transform is reduced, and thus the resolution of the frequency is deteriorated. However, even if the frequency resolution is improved by increasing the window length, the temporal resolution is adversely deteriorated. In order to overcome the limitation of resolution due to the trade-off relationship between frequency and time, a wavelet transform has emerged.

If the window length is fixed in STFT, the WT performs STFT several times while changing the window length. Also, if a sinusoid that extends to infinity in time is used as a basic function in STFT, wavelet has several types of functions that exist for a finite period. Wavelet functions include Morlet, Daubechies, Coiflets, Biorthogonal, Mexican Hat, and Symlets.

5 is a block diagram of another embodiment of an artificial neural network in the apparatus for estimating the type of road using sound waves with the amount of atmospheric attenuation corrected according to the present invention.

5 is a diagram for explaining an artificial neural network in an apparatus for estimating a type of road using sound waves in which atmospheric attenuation is corrected according to an embodiment of the present invention.

As shown in FIG. 5 , the artificial neural network 400 is a multi-layer perceptron including a convolution performing unit 501 , a transmission layer 502 , a plurality of hidden layers 503 and an output layer 504 . Perceptron) is a deep convolutional neural network (DCNN) algorithm.

The convolution performing unit 501 performs 1D convolution operation multiple times on the received digital input signal, and performs batch normalization, Relu function, and MaxpPooling function for each convolution operation. , the output of the last convolution operation outputs the flattened data to the transfer layer 502 .

The transfer layer 502 receives the flattened output data of the convolution performing unit 501 as 1D input.

Data input to the transfer layer 502 is feature extracted and classified through a plurality of hidden layers 503 .

The output layer 504 outputs a probability value for each type of the learned road surface.

The artificial neural network 400 determines and outputs the type of road surface having the highest probability among the probability values output from the output layer 504 using the softmax 505 .

Meanwhile, the artificial neural network 400 may receive atmospheric information (temperature, humidity, and atmospheric pressure information) and use it as an input of the convolution performing unit 501 .

Also, the artificial neural network 400 may use the sound wave signal 201 transmitted to the road surface as an input of the convolution performing unit 501 .

FIG. 6 is a diagram for explaining an operation of the convolution performing unit of FIG. 5 .

As shown in FIG. 6 , the convolution performing unit 501 of FIG. 5 performs a 1D convolution operation on the input signal 5 times, and for each convolution operation, a batch normalization (Batch Normalization), a Relu function, and Executes the MaxpPooling function, and the output of the last convolution operation is flattened data.

The input signal 601 becomes about 7000 received signals for 7 ms, and the result of the first convolution 602 is 1D conv (64,16), BN, ReLU, and MP(8) to the input signal 601. The result of performing the second convolution 603 is the result of performing 1D conv (32,32), BN, ReLU, and MP(8) on the result of the first convolution 602, and the third convolution The execution result 604 is the result of performing 1D conv (16,64), BN, ReLU, and MP(8) on the second convolution result 603, and the fourth convolution result 605 is the first 1D conv (8,128), BN, and ReLU are performed on the 3 convolution result 604, and the 5th convolution result 606 is 1D conv (4,2568) on the 4 convolution result 605. , BN, and ReLU are the results.

7 is a view for explaining a code of a convolution performing unit in the apparatus for estimating the type of road using sound waves in which atmospheric attenuation is corrected according to the present invention.

FIG. 7 is a code in which a part of the convolution performing unit shown in FIG. 6 is implemented in software.

It includes a number of batch normalization (BatchNorm) functions, and a max pooling function (MaxPool).

As one usable example according to an embodiment of the present invention, one-dimensional (1D) convolution followed by one-dimensional batch normalization followed by maxpooling is repeated as one set for 4 sets, and finally through a fully connected layer By outputting as many values as the number of road surfaces to be classified, the probability is output for each road surface.

Meanwhile, in the present invention, a method of performing a one-dimensional (1D) convolution operation has been described as an example, but the convolution operation can be performed not only in 1D, but also in 2D and 3D.

8A is a table showing test results when the atmospheric attenuation amount is not corrected in the apparatus for estimating the type of road using sound waves in which the atmospheric attenuation amount is corrected according to an embodiment of the present invention, and FIG. This table shows the test results when the atmospheric attenuation amount is corrected in the road surface type estimation device using the sound wave with the attenuation amount corrected.

The classes classified in FIGS. 8a and 8b are asphalt (1), cement (2), dirt (3), ice (4), marble (5), paint (6), slush (7), snow (8), and water (9).

Actual laboratory test data include asphalt (1) (348 pcs), cement (2) (166 pcs), ice (4) (85 pcs.), paint (6) (390 pcs.), and snow (8) (255 pcs.) A total of 1244 datasets including

As shown in FIG. 8A , when the atmospheric attenuation amount was not corrected, the classification performance was 0.6%, so there was no classification at all.

As shown in FIG. 8B, when the atmospheric attenuation amount is corrected according to the present invention, asphalt is 99.7% accurate, cement is 91.6% accurate, ice is 95.3% accurate, and paint ) was estimated with 99.0% accuracy and snow with 97.3% accuracy, and overall classification performance of 97.6% was obtained.

9 is a flowchart illustrating a method for estimating the type of road surface using sound waves in which atmospheric attenuation is corrected according to an embodiment of the present invention.

First, in order to perform the method for estimating a road surface type using sound waves according to the present invention, the learning step S10 must be preceded and a road surface classification model must be generated.

In the learning step (S10), after transmitting a sound wave signal to a plurality of types of road surfaces, receiving a reflected signal, converting the signal into a digital signal, and correcting the amount of attenuation in the air for the converted digital signal Thereafter, it is converted into a frequency domain signal, and the frequency domain signal is input to the neural network 420 to learn the road surface classification model.

Here, in the learning step (S10), a Short-Time Fourier Transform (STFT) transformer, a Fast Fourier Transform (FFT), a cepstruam, or a wavelet transform in order to convert the frequency domain signal. can be used The frequency domain signal may be 2D or 3D.

Thereafter, under the control of the controller 500 , the sound wave signal is transmitted to the corresponding road surface for which the type is to be determined, and then the reflected signal is received ( S20 ).

Thereafter, the analog signal is converted into a digital signal with respect to a preset region of the received signal under the control of the controller 500 (S30).

In the signal conversion step (S30), in the received signal, excluding the crosstalk signal of the transmitted sound wave signal, the point at which the amplitude of the received signal after the transmission delay is the largest in each period of the sound wave signal to convert the signal for a preset time into a digital signal.

For example, assuming that t_0 is the point at which the amplitude of the signal received after the crosstalk signal is greatest, a total of (a+b) ms can be observed from t_0 - a [ms] to t_0 + b [ms]. , in the received signal 203 of FIG. 2, a is 0.2 and b is 5. Of course, depending on the environment or conditions, a and b can be adjusted.

Thereafter, under the control of the controller, the amount of attenuation in the air of the digital signal is calculated and corrected (S40).

In the atmospheric attenuation correction step (S40), the atmospheric attenuation amount is calculated and corrected using the following <Equation 1> to <Equation 8>.

First, the saturation pressure Psat is calculated using the following <Equation 1>.

[Equation 1]

Here, To1 is the triple point of the atmosphere [K], and T is the current temperature [K].

Absolute humidity (h) is calculated using the following <Equation 2>.

[Equation 2]

Here, hrin is relative humidity [%], Psat is saturation pressure [unit], and Ps is static pressure [atm].

Meanwhile, the scaled relaxation frequency for nitrogen (FrN) of nitrogen occupying 78% of the atmosphere is calculated using the following <Equation 3>.

[Equation 3]

Here, To is the reference temperature [K], and T is the current temperature [K].

Meanwhile, the scaled relaxation frequency for oxygen (FrO) of oxygen occupying 21% of the atmosphere is calculated using the following <Equation 4>.

[Equation 4]

where h is the absolute humidity.

Meanwhile, the attenuation coefficient (α: attenuation coefficient [nepers/m]] is calculated using the following <Equation 5>.

[Equation 5]

where Ps is the static pressure, F is the frequency of the sound wave signal (transmitted sound wave signal), T is the current temperature [K], To is the reference temperature [K], FrO is the extended relaxation frequency of oxygen, and FrN is It is the extended relaxation frequency of nitrogen.

On the other hand, the attenuation ratio (A, unit: dB) of the sound wave signal is calculated using the following <Equation 6>.

[Equation 6]

Here, α is an attenuation coefficient, and d is a distance between the sound wave transmitter 100 and the corresponding road surface for which the type is to be known.

The d is calculated using the following <Equation 7> using the time t (time of flight) from being transmitted from the transmitter until it is reflected by the road surface and sensed by the receiver and the speed of sound in the air (Vair).

[Equation 7]

Here, t is the required time, and Vair is the speed of sound in the air [m/s].

On the other hand, the speed of sound in the air is calculated using the following <Equation 8>.

[Equation 8]

Here, Ks is the isentropic coefficient of stiffness of the object, and ρ is the density of the object (atmosphere).

At this time, assuming air (atmosphere) as an ideal gas, Ks = γP, γ is the heat capacity ratio (1.4 in the case of air), P is the pressure, R is the ideal gas constant, and T is the absolute temperature [K]. . Since it is a constant except for temperature, it can also be expressed approximately.

The speed of sound in the atmosphere may be compensated for attenuation by correcting it according to the temperature, atmospheric pressure, and humidity of the atmosphere.

Thereafter, under the control of the controller 500, a frequency domain signal is obtained by performing signal conversion on a preset region of the corrected digital signal (S50).

In the frequency domain signal acquisition step (S50), the crosstalk signal of the transmitted sound wave signal is excluded from the corrected digital signal, and for each period of the sound wave signal, the signal for a preset time received after the transmission delay is applied. The frequency domain signal is obtained by performing frequency conversion in the signal converter 410 .

Thereafter, under the control of the control unit 500, the frequency domain signal is used as an input signal of the neural network 420, and based on the learned road surface classification model, characteristics of the input signal are extracted and classified, A type is determined (S60).

The neural network 420 is a deep neural network (DNN) of a multi-layer perceptron algorithm including an input layer 401, a plurality of hidden layers 402 and an output layer 403, The output layer 403 outputs a probability value for each type of the learned road surface, and the neural network determines and outputs the type of road surface having the highest probability using a softmax 404 . .

Meanwhile, the structure of the neural network 420 is not limited to the aforementioned DNN.

The neural network 420 may receive atmospheric information (temperature, humidity, and atmospheric pressure information) and use it as an input of the input layer 401 .

Also, the neural network 420 may frequency-convert the sound wave signal transmitted to the road surface and use it as an input of the input layer 401 .

10 is a flowchart of another embodiment of a method for estimating a road surface type using a sound wave having an atmospheric attenuation amount corrected according to the present invention.

First, in order to perform the method for estimating the road surface type using the sound wave in which the atmospheric attenuation amount has been corrected according to the present invention, the second learning step S90 must be preceded and a road surface classification model must be generated.

In the second learning step (S90), after transmitting a sound wave signal to a plurality of types of road surfaces, receiving a reflected signal, converting the signal into a digital signal, and determining the amount of attenuation in the air for the converted digital signal After correction, it is input to the artificial neural network 400 to perform convolutional calculations multiple times to learn the road surface classification model.

Thereafter, under the control of the controller 500 , the sound wave signal is transmitted to the corresponding road surface for which the type is to be determined, and then the reflected signal is received ( S20 ).

Thereafter, the analog signal is converted into a digital signal with respect to a preset region of the received signal under the control of the controller 500 (S30).

In the signal conversion step (S30), in the received signal, excluding the crosstalk signal of the transmitted sound wave signal, the point at which the amplitude of the received signal after the transmission delay is the largest in each period of the sound wave signal to convert the signal for a preset time into a digital signal.

Thereafter, under the control of the controller 500, the amount of attenuation in the air of the digital signal is calculated and corrected (S40).

In the atmospheric attenuation correction step (S40), the atmospheric attenuation amount is calculated and corrected using <Equation 1> to <Equation 8> as described in FIG. 9 .

Thereafter, under the control of the controller 500 , the artificial neural network 400 receives the digital signal in which the amount of attenuation in the air is corrected, and performs a convolution operation multiple times ( S70 ).

In the convolution operation step (S70), 1D convolution operation is performed multiple times on the digital signal in which the amount of attenuation in the air is corrected, and for each convolution operation, batch normalization (Batch Normalization), Relu function, and MaxpPooling (MaxpPooling) ) function, and the output of the last convolution operation is flattened data.

Thereafter, according to the control of the controller 500 , based on the road surface classification model learned in the artificial neural network 400 , the characteristics of the convolutionally calculated signal (flattened data) are extracted and classified, and the road surface determines the type of (S80).

The artificial neural network 400 includes a convolution performing unit 501 that receives the digital signal and performs multiple convolution operations, a transmission layer 502, a plurality of hidden layers 503 and an output layer 504. It is a deep convolutional neural network (DCNN) of a multi-layer perceptron algorithm, and the output layer 504 outputs a probability value for each type of the learned road surface, and the artificial neural network 400 , determines and outputs the type of road surface having the highest probability using a softmax 505 .

Meanwhile, the structure of the artificial neural network 400 is not limited to the aforementioned DCNN.

The artificial neural network 400 may receive atmospheric information (temperature, humidity, and atmospheric pressure information) and use it as an input of the convolution performing unit 501 .

Also, the artificial neural network 400 may use the sound wave signal transmitted to the road surface as an input of the convolution performing unit 501 .

In the above, the method for estimating the type of road using the sound wave with the amount of atmospheric attenuation corrected according to an embodiment of the present invention has been described. It goes without saying that a program stored in a computer-readable recording medium for implementing a method for estimating a road surface using a sound wave in which an amount of medium and atmospheric attenuation has been corrected can also be implemented.

That is, those skilled in the art can easily understand that the above-described method for estimating the type of road using sound waves with the amount of atmospheric attenuation corrected may be provided by being included in a computer-readable recording medium by tangibly implementing a program of instructions for implementing the method. you will understand In other words, it may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and used by those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and floppy disks. magneto-optical media, and hardware devices specially configured to store and carry out program instructions, such as ROM, RAM, flash memory, USB memory, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention as claimed in the claims.

100: sound wave transceiver
101: sound wave transmitter 102: sound wave receiver
200: analog-to-digital converter 300: atmospheric attenuation correction unit
400: artificial neural network
410: signal converter 420: neural network
501: convolution performer 502: transfer layer
503: hidden layer 504: output layer
500: control unit (MCU) 600: atmospheric information measurement unit
S10: first learning stage
S20: reflected wave receiving step
S30: signal conversion step
S40: Atmospheric attenuation compensation step
S50: Frequency domain signal acquisition step
S60: first road surface type determination step
S70: Convolution operation step
S80: Second road surface type determination step
S90: second learning stage

Claims (11)

In an apparatus for estimating a road surface type using sound waves in which atmospheric attenuation is corrected,
a sound wave transceiver 100 for transmitting a sound wave signal to a corresponding road surface for which a type is to be known, and then receiving a reflected signal;
an analog-to-digital converter 200 for converting an analog signal of the received signal into a digital signal;
an atmospheric attenuation correction unit 300 for calculating and correcting the atmospheric attenuation amount of the digital signal;
an artificial neural network 400 that uses the corrected digital signal as an input signal and classifies the input signal based on the learned road surface classification model to estimate the type of road surface; and
A control unit (MCU) 500 for controlling the operation of the sound wave transceiver, the analog-to-digital converter, the atmospheric attenuation correction unit, and the artificial neural network
including,
The control unit 500,
In the received signal, except for the crosstalk signal of the transmitted sound wave signal, each period of the sound wave signal receives a signal for a preset time based on the point at which the amplitude of the signal received after the transmission delay is greatest It is characterized in that it is determined by a signal,
It is characterized in that a total of (a+b) ms from t_0 - a [ms] to t_0 + b [ms] is determined as the received signal when the time point at which the amplitude of the received signal is greatest after the crosstalk signal is t_0 with
The apparatus for estimating the type of road using sound waves in which atmospheric attenuation is corrected, characterized in that the time point of “t_0 - a” is after the end time of the crosstalk signal.
According to claim 1,
Air information measurement unit 600 for measuring atmospheric information (temperature, humidity, atmospheric pressure information)
further comprising,
The atmospheric attenuation correction unit 300 receives the standby information and calculates the atmospheric attenuation amount of the digital signal,
The control unit controls the operation of the atmospheric information measurement unit, the road surface type estimation apparatus using a sound wave in which the atmospheric attenuation amount is corrected.
According to claim 1,
In the atmospheric attenuation correction unit 300,
The saturation pressure (Psat) is calculated using the following <Equation 1>,
[Equation 1]

(Where To1 is the triple point of the atmosphere [K], and T is the current temperature [K])

Absolute humidity (h) is calculated using the following <Equation 2>,
[Equation 2]

(Where hrin is relative humidity [%], Psat is saturation pressure [ unit ], and Ps is static pressure [atm].)

The extended relaxation frequency (FrN: scaled relaxation frequency for Nitrogen) of nitrogen occupying 78% of the atmosphere is calculated using the following <Equation 3>,
[Equation 3]

(Here, To is the reference temperature [K], and T is the current temperature [K].)

The extended relaxation frequency (FrO: scaled relaxation frequency for Oxygen) of oxygen occupying 21% of the atmosphere is calculated using the following <Equation 4>,
[Equation 4]

(Where h is the absolute humidity.)

Attenuation coefficient (α: attenuation coefficient [nepers / m]] is calculated using the following <Equation 5>,
[Equation 5]

(Where Ps is the static pressure, F is the frequency of the sound wave signal (transmitted sound wave signal), T is the current temperature [K], To is the reference temperature [K], FrO is the extended relaxation frequency of oxygen, FrN is the extended relaxation frequency of nitrogen.)

The attenuation ratio (A, unit: dB) of the sound wave signal is calculated using the following <Equation 6>,
[Equation 6]

(Here, α is the attenuation coefficient, and d is the distance between the sound wave transmitter 100 and the corresponding road surface for which the type is to be known.)
The d is calculated using the following <Equation 7> using the time t (time of flight) and the speed of sound in the air (Vair) from being transmitted from the transmitter to being reflected by the road surface and detected by the receiver,
[Equation 7]

(Where t is the required time, and Vair is the speed of sound in the air [m/s])

The speed of sound in the air is calculated using the following <Equation 8>,
[Equation 8]

(Here, Ks is the isentropic coefficient of stiffness of the object, and ρ is the density of the object (atmosphere). At this time, assuming air (atmosphere) as an ideal gas, Ks = γP, and γ is the heat capacity ratio (1.4 for air), P is the pressure, R is the ideal gas constant, and T is the absolute temperature [K].)
Estimation of road surface type using sound waves with the amount of atmospheric attenuation corrected, characterized in that the received signal is corrected by calculating the amount of attenuation (attenuation ratio) (A) in the air using the [Equation 1] to [Equation 8] Device.
According to claim 1,
The artificial neural network 400 is
a signal converter 410 for obtaining a frequency domain signal by performing frequency conversion on a preset region on the time domain of the corrected digital signal received from the atmospheric attenuation correction unit; and
and a neural network 420 that uses the frequency domain signal as an input signal and extracts and classifies the characteristics of the input signal based on the learned road surface classification model to estimate the type of road surface,
The neural network 420 is
A road surface type estimation apparatus using sound waves with corrected atmospheric attenuation, characterized in that it is a deep neural network of a multi-layer perceptron algorithm including an input layer, a plurality of hidden layers, and an output layer.
According to claim 1,
The artificial neural network 400 is
A multi-layer perceptron including a convolution performing unit 501 , a transfer layer 502 , a plurality of hidden layers 503 , and an output layer 504 that receives the corrected digital signal and performs convolution operation multiple times Perceptron) algorithm's deep convolutional neural network (DCNN).
6. The method of claim 5,
The convolution performing unit 501,
1D convolution operation is performed multiple times on the input signal, batch normalization, Relu function, and MaxpPooling function are performed for each convolution operation, and the output of the final convolution operation is flattened data It is characterized in that
The output layer is
Characterized in outputting a probability value for each type of the learned road surface,
The artificial neural network is
An apparatus for estimating a road surface type using a sound wave corrected for atmospheric attenuation, characterized in that it determines and outputs the type of road surface having the highest probability using Softmax.
In the method for estimating the type of road surface using sound waves in which atmospheric attenuation is corrected,
A reflected signal receiving step (S20) of transmitting a sound wave signal to the corresponding road surface for which the type is to be known and then receiving the reflected signal under the control of the controller;
a signal conversion step (S30) of converting an analog signal into a digital signal for a preset region of the received signal under the control of the controller;
an atmospheric attenuation correction step (S40) of calculating and correcting the atmospheric attenuation amount of the digital signal under the control of the control unit;
a frequency domain signal acquisition step of obtaining a frequency domain signal by performing frequency conversion on a preset region of the corrected digital signal under the control of the controller (S50); and
A first road surface type that uses the frequency domain signal as an input signal of a neural network under the control of the controller, extracts and classifies characteristics of the input signal based on a learned road surface classification model, and determines the type of the road surface Decision step (S60)
including,
The signal conversion step (S30),
In the received signal, except for the crosstalk signal of the transmitted sound wave signal, for each period of the sound wave signal, the signal for a preset time based on the point at which the amplitude of the signal received after the transmission delay is greatest It is characterized in that it is converted to a digital signal,
It is characterized in that a total (a+b) ms is converted from t_0 - a [ms] to t_0 + b [ms] when the time point at which the amplitude of the signal received after the crosstalk signal is greatest is t_0,
The method for estimating the type of road using the sound wave in which the atmospheric attenuation amount is corrected, characterized in that the time point of “t_0 - a” is after the end time of the crosstalk signal.
8. The method of claim 7,
For a plurality of types of road surfaces, after transmitting a sound wave signal, receiving a reflected signal, converting the signal into a digital signal, correcting the amount of attenuation in the air for the converted digital signal, and converting it into a frequency domain signal, , a first learning step of learning the road surface classification model by inputting the frequency domain signal to the neural network (S10)
A method for estimating the type of road using sound waves in which atmospheric attenuation has been corrected, characterized in that it further comprises.
In the method for estimating the type of road surface using sound waves in which atmospheric attenuation is corrected,
a reflected signal receiving step (S20) of transmitting a sound wave signal to a corresponding road surface for which a type is to be known under the control of the controller and receiving the reflected signal;
a signal conversion step (S30) of converting an analog signal into a digital signal for a preset region of the received signal under the control of the controller;
an atmospheric attenuation correction step of calculating and correcting the atmospheric attenuation amount of the digital signal (S40);
a convolution operation step (S70) of receiving a digital signal with a corrected amount of attenuation in the air from the artificial neural network under the control of the control unit and performing a convolution operation multiple times; and
A second road surface type determination step (S80) of extracting and classifying characteristics of the convolutional signal based on the road surface classification model learned from the artificial neural network under the control of the controller and determining the type of the road surface (S80)
including,
The signal conversion step (S30),
In the received signal, except for the crosstalk signal of the transmitted sound wave signal, for each period of the sound wave signal, the signal for a preset time based on the point at which the amplitude of the signal received after the transmission delay is greatest It is characterized in that it is converted to a digital signal,
It is characterized in that a total (a+b) ms is converted from t_0 - a [ms] to t_0 + b [ms] when the time point at which the amplitude of the signal received after the crosstalk signal is greatest is t_0,
The method for estimating the type of road using the sound wave in which the atmospheric attenuation amount is corrected, characterized in that the time point of “t_0 - a” is after the end time of the crosstalk signal.
10. The method of claim 9,
After transmitting a sound wave signal to multiple types of road surfaces, receiving a reflected signal, converting the signal into a digital signal, correcting the amount of attenuation in the air for the converted digital signal, and inputting it to the artificial neural network A second learning step (S90) of learning the road surface classification model by performing multiple convolutional calculations
A method for estimating the type of road using sound waves in which atmospheric attenuation has been corrected, characterized in that it further comprises.
11. The method of claim 7 or 10,
In the atmospheric attenuation correction step (S40),
The saturation pressure (Psat) is calculated using the following <Equation 1>,
[Equation 1]

(Where To1 is the triple point of the atmosphere [K], and T is the current temperature [K])

Absolute humidity (h) is calculated using the following <Equation 2>,
[Equation 2]

(Where hrin is relative humidity [%], Psat is saturation pressure [ unit ], and Ps is static pressure [atm].)

The extended relaxation frequency (FrN: scaled relaxation frequency for Nitrogen) of nitrogen occupying 78% of the atmosphere is calculated using the following <Equation 3>,
[Equation 3]

(Here, To is the reference temperature [K], and T is the current temperature [K].)

The extended relaxation frequency (FrO: scaled relaxation frequency for Oxygen) of oxygen occupying 21% of the atmosphere is calculated using the following <Equation 4>,
[Equation 4]

(Where h is the absolute humidity.)

Attenuation coefficient (α: attenuation coefficient [nepers / m]] is calculated using the following <Equation 5>,
[Equation 5]

(Where Ps is the static pressure, F is the frequency of the sound wave signal (transmitted sound wave signal) , T is the current temperature [K], To is the reference temperature [K], FrO is the extended relaxation frequency of oxygen, FrN is the extended relaxation frequency of nitrogen.)

The attenuation ratio (A, unit: dB) of the sound wave signal is calculated using the following <Equation 6>,
[Equation 6]

(Here, α is the attenuation coefficient, and d is the distance between the sound wave transmitter 100 and the corresponding road surface for which the type is to be known.)
The d is calculated using the following <Equation 7> using the time t (time of flight) and the speed of sound in the air (Vair) from being transmitted from the transmitter to being reflected by the road surface and detected by the receiver,
[Equation 7]

(Where t is the required time, and Vair is the speed of sound in the air [m/s])

The speed of sound in the air is calculated using the following <Equation 8>,
[Equation 8]

(Here, Ks is the isentropic coefficient of stiffness of the object, and ρ is the density of the object (atmosphere). At this time, assuming air (atmosphere) as an ideal gas, Ks = γP, and γ is the heat capacity ratio (1.4 for air), P is the pressure, R is the ideal gas constant, and T is the absolute temperature [K].)
Estimation of road surface type using sound waves with the amount of atmospheric attenuation corrected, characterized in that the received signal is corrected by calculating the amount of attenuation (attenuation ratio) (A) in the air using the [Equation 1] to [Equation 8] Way.

Images