EP1480182B1 - Contactless axle counter for road traffic - Google Patents

Abstract

Contactless axle counting system for road transport has at least a measurement arrangement (1) for measuring changes in environmental parameters such as sound, temperature or vibrations caused by the tires of a vehicle passing near the measurement area. The measurement arrangement is placed at or near the contact point of the vehicle tires on the road surface. The measurement arrangement can be positioned above the road surface, e.g. in an existing bridge or overhead gantry. Suitable measurement arrangements can be based on laser-vibrometers or directional microphone systems.

Description

The invention relates to a non-contact axle counting system for road traffic comprising at least one measuring device for detecting the changes in the environmental parameters, such as sound, temperature or vibration, produced by a moving vehicle to a measuring range.

Non-contact axle counting systems are known in which induction loops are integrated into the roadway. Such systems have the disadvantage that they must be installed in the roadway. Even in case of maintenance, the road would be dug up, which is not only associated with high costs, but also with long-lasting traffic obstructions.

There are also optical non-contact axle counting systems, such as laser scanners, known, however, which are only suitable for detecting vehicles on single-lane road sections, because the corresponding measuring devices can only be mounted laterally on the edges of the road.

No. 5,821,879 A shows such a non-contact axle counting system for multi-lane lanes. The axle counting system includes an LED array on one lane side and a photodiode array on the opposite lane side so that a plurality of light barriers are formed. The signal interruptions recorded by these light barriers are directed to a signal evaluation system and from this the traffic volume is derived on each lane.

The object of the invention is therefore to provide a non-contact axle counting system which avoids the above disadvantages and allows for low installation and maintenance effort a reliable axle counting for multi-lane roads.

According to the invention, this is achieved in that at least one measuring device for detecting the movement of a moving vehicle in a measuring range, such. At or near the point of contact from tire to roadway, changes in environmental parameters such as sound, temperature or vibration are provided. In this way, measuring devices can conveniently and inexpensively on any existing superstructures of the roadway, such as bridges or signage, mounted and maintained, with an assignment of the measuring devices to lanes can be particularly easily implemented, and such a measuring device is by their Focusing on a suitable measuring range suitable to detect the vehicles without contact also in multi-lane road sections.

In a particular embodiment of the invention can be provided that the at least one of the at least one measuring device is designed as a laser vibrometer. Laservibrometer allow a high-precision non-contact detection of road vibrations, as caused by the weight and / or the self-vibration of moving vehicles on the tires on the road.

In another embodiment of the invention can be provided that the at least one of the at least one measuring device is designed as a directional microphone, in particular as a series microphone system. Such an arrangement is suitable for detecting and locating tire rolling noise particularly well.

In a further embodiment of the invention can be provided that an analysis unit is provided for the analysis of measurement signals. Thus, the interpretation of the measurement signals can be facilitated and improved.

In particular, it can be provided that the analysis unit comprises a signal amplification device for amplifying the measurement signals, in particular by a factor of 100 to 10,000. This means that even very weak signals can be made accessible to an evaluation.

Furthermore, it can be provided that the analysis unit comprises an A / D converter with a sampling rate of in particular 1 kHz to 50 kHz, or more particularly between 6 kHz and 16 kHz. Such digitized measurement signals are particularly well suited for subsequent data analyzes.

In particular, it can be provided that the A / D converter has a resolution of 8 to 24 bits, preferably 12 to 20 bits. Such a resolution is sufficient for subsequent data analyzes and leads at the same time to still easily manageable amounts of data.

In a particular embodiment of the invention can be provided that the analysis unit comprises a computing unit, which is arranged downstream of the A / D converter. Thus, the digitized measurement signals can be prepared for a classification of the vehicles to be detected.

In a further embodiment of the invention it can be provided that a memory and / or a user interface is connected to the arithmetic unit. The memory allows the storage of parameters influencing the measurement, while corresponding measurement or environmental parameters can be changed via the user interface.

In particular, it can be provided that the arithmetic unit comprises a time segmentation device for subdividing discrete time signals into time signal blocks, in particular from a block length of 10 ms to 1500 ms. A temporal segmentation device allows a higher resolution in the analysis of the measurement signals.

According to a further embodiment of the invention, it can be provided that the arithmetic unit comprises a spectral analyzer for calculating the frequency spectrum of the measurement signals and / or an energy calculation unit for calculating the signal energy, in particular the short-term signal energy. From the data thus calculated, data patterns can be generated which are particularly well suited for a subsequent classification.

In a further development of the invention it can be provided that the arithmetic unit comprises a classifier, which is arranged downstream of the spectral analyzer and / or the energy calculation unit, and which classifies the data pattern resulting from the corresponding calculation results of the spectral analyzer and / or the energy calculation unit. Classified data samples allow a particularly reliable determination of the number of axles of different vehicle types.

In particular, it can be provided that the classifier can be operated in a learning mode for generating mathematical models for specific vehicle classes. In this way, the axle counting system can be calibrated to a large number of vehicle types to be determined, and can also be retrofitted cost-effectively in the event of a change in demand to new vehicle types.

It can be provided that the learning mode is operable using learning algorithms.

Furthermore, it can be provided that the learning algorithm can be operated using the Hebbian learning method, the backpropagation rule or the forward-backward algorithm.

In a further consequence it can be provided that the classifier can be operated in a recognition mode for the assignment of a current measurement signal to a vehicle class. Thus, a safe, reliable and tailor-made operation of the non-contact axle counting system can be achieved.

Finally, it can be provided in a further development of the invention that the analysis unit is arranged downstream of a communication unit to which the calculation result of the classifier is transferred. Thus, even at a further distance, for example in a central office, the number of axles of vehicles passing a measuring point can be detected and monitored.

• Fig. 1 ein Fahrzeug auf einer Fahrbahn in Seitenansicht,
• Fig. 2 eine mehrspurige Fahrbahn mit Messeinrichtungen, einer Analyse- und einer Kommunikationseinheit, im Grundriss,
• Fig. 3 ein Blockdiagramm einer Analyseeinheit zwischen Messeinrichtungen und
• Kommunikationseinheit und
• Fig. 4 ein Blockdiagramm einer Rechnereinheit zwischen A/D-Konverter und Kommunikationseinheit.

The invention will be further described with reference to the accompanying drawings, in which embodiments are shown. Showing:

• 1 is a vehicle on a road in side view,
• 2 a multi-lane road with measuring devices, an analysis and a communication unit, in plan view,
• 3 is a block diagram of an analysis unit between measuring devices and
• Communication unit and
• 4 shows a block diagram of a computer unit between A / D converter and communication unit.

A vehicle 10 is via its tires 12 at contact points 11 in mechanical interaction with the roadway 20. Due to the weight, the proper motion (vibration) and the relative movement of the vehicle 10 to the roadway 20 11 vibrations are now mainly in the region of the contact points excited. These are on the one hand sound vibrations caused mainly by the rolling of the tires 12 on the roadway 20, but also by engine and flow noise, on the other hand vibrations of the surface of the roadway 20 itself, the amplitude may be on the order of up to several millimeters , These physical quantities, which are subject to change if a motor vehicle passes the location of the measurement, can be detected by different measuring principles.

The amplitude of the radiated sound waves can be described in different ways - e.g. as the velocity of movement of the air molecules (sound velocity) or as pressure which is generally preferred for describing the amplitude. The sound pressure results from the fluctuation of the air pressure above and below the atmospheric air pressure. Measuring devices 1, which serve to convert the sound pressure into an electrical signal, can be any suitable sound sensors, such as e.g. electroacoustic transducers. In this case, the measuring device 1 can also be designed as a directional microphone or as a series microphone system, which contributes to an improved spatial resolution of a measurement signal.

The vibrations of the surface of the traffic route can in principle be measured with accelerometers which work according to the piezoelectric effect - but not without contact. By means of optical methods, these vibrations can also be measured without contact. For detecting the vibrations of the surface of the roadway 20 are measuring devices 1 in an embodiment of Laservibrometem especially good. Laser vibrometers work on the principle of Doppler frequency shift. In this case, the laser light scattered back by a vibrating object (for example, the surface of a roadway 20) provides all information for the determination of surface velocity and absolute oscillation amplitudes. In contrast to other optical methods (eg laser scanners), it is not the distance of an existing object that is of interest, but the vibration velocity of its surface. With the help of this high-precision measurement, the least vibration excitations of the surface of the roadway 20 can be detected.

The measuring devices 1 for measuring the changes of the environmental influences measure the strength of the sound waves, preferably the sound pressure of the sound waves or the vibrations of the surface of the roadway, when vehicles 10 approach or pass the mounting location of a measuring device 1. Preferably, the measuring devices 1 are mounted above the roadway 20 on superstructures such as bridges, signaling devices or signage (see Fig. 2), so that the monitoring of several parallel lanes is easily possible.

The measuring devices 1 convert the sound or vibration signals into electrical energy. The measuring devices 1 generate analog or digital measuring signals as a function of time. The measurement signals generated by the measuring devices 1 are forwarded to the analysis unit 2 via separate signal lines or after modulation or coding via a common signal line. However, the measurement signals can also be transmitted to the analysis unit 2 by a wireless connection (eg radio, infrared, etc.) or via local area networks (LAN) or wireless LAN (WLAN).

The analysis unit 2 in turn communicates via a communication unit 8, for example with a central unit (not shown) for collecting, evaluating or further processing the results supplied by the analysis unit 2. The connection between the communication unit 8 and the central unit can be wired or wireless.

FIG. 3 shows the block diagram of the hardware structure of the analysis unit 2. The measurement signals are forwarded to the signal amplification device 3, which amplifies the measurement signals by a fixed factor, preferably between 100 and 10,000, or with an automatic adjustment.

The following analog / digital converter 4 converts the analog signals into discrete values. The sampling rate of the A / D converter 4 may vary from system to system and is generally between 1 kHz and 50 kHz. Particularly suitable are sampling frequencies between 6 kHz and 16 kHz. The resolution of the A / D converter 4 is in the range of 8 to 24 bits, with the range of 12 to 20 bits is preferably used. The system has a computing unit 5, which is connected to the A / D converter 4 and the data memory 6. The arithmetic unit 5 is used to execute the calculation steps that are applied to the digitized measurement signals. The user interface 7 and the memory 6 are connected to the arithmetic unit 5. Through the user interface 7 inputs can be made by a user. The input by the user may be made by any suitable device, such as a keyboard, a mouse, a touch screen, or any combination of these devices.

The result of the analysis is transferred via an output to the interface of the communication unit 8 and further processed there.

FIG. 4 shows, in the form of a block diagram, the analysis of the measurement signals as they occur in the arithmetic unit 5. The measurement signals are supplied from the A / D converter 4 to a time segmentation 51 in order to subdivide the discrete time signals into time blocks, wherein the block length may be between 10 ms and 1500 ms. The individual blocks are further extracted from the signal with an overlap. This overlap serves to increase the resolution and can take values between 20% and 70%. After the temporal segmentation 51, the data are transferred on the one hand to the energy calculation unit 53 and also to the spectrum analyzer 52.

Subsequently, the functionality of the analysis unit 2 will be described. Here, the variable t stands for the time, T for the length of a signal block and i stands for the number of the block within the entire acoustic signal. s _i (t) denotes the measurement signal in the time domain of the i- th block, S _i (w) the frequency spectrum of the i- th block. The variable w corresponds to the instantaneous frequency.

The analysis unit 2 consists of a spectrum analyzer 52 and an energy calculation unit 53 for calculating the signal energy. The spectral analyzer 52 transforms the individual signal blocks from the time domain into the frequency domain. By default, these transformations are performed using Fourier transform. The so-called Fast Fourier Transformation (FFT) is particularly well suited for this purpose. The Fast Fourier Transformation corresponds to a digital approximation of the Fourier transformation. The Fourier transformation of a function s (t) is defined as follows: $$S\left(w\right)=\int s\left(t\right){\mathrm{<figure-callout\; id="2"\; label="e"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">e</figure-callout>}}^{2\mathit{\pi ift}}dt$$

The output of the spectrum analyzer 52 corresponds to the power density spectrum of the signal at the input and describes the amount of energy at a particular frequency interpolation point. The number of frequency nodes is dependent on the number of discrete samples taken for the Fourier transform from the temporal signal, and is directly related to the above-mentioned block length T. Besides the Fourier transform for calculating the spectral content a time signal can also other methods, such as Linear Prediction Coding is used, which is known from the literature.

Another component of the analysis unit 2 is the energy calculation unit 53. In general, the relationship for determining the signal energy is described by the following equation: $$e_n={\displaystyle \underset{-\infty }{\overset{+\infty }{\Sigma }}}{s}^2\left(t\right)$$

wobei $$w\left(t\right)=\left\{\begin{array}{ccc}1& \mathit{für}& 0\le t\le T\\ 0& \mathit{für}& \begin{array}{}\end{array}\mathit{sonst}\end{array}\right\}$$

Since the course of the energy as a function of time is of interest for the description of acoustic signals, the use of the short-term signal energy is particularly suitable. It is defined as: $$e_n={\displaystyle \underset{k=-\infty }{\overset{\infty }{\Sigma }}}[s\left(k\right)\cdot w\left(t-k\right){]}^2$$

in which $$w\left(t\right)=\left\{\begin{array}{ccc}1& \mathit{For}& 0\le t\le T\\ 0& \mathit{For}& \begin{array}{}\end{array}\mathit{otherwise}\end{array}\right\}$$

By choosing the window length of the resulting waveform of the signal energy can be significantly influenced. The use of large window lengths corresponds to a low-pass filtering with a low cut-off frequency and has the consequence that short-term fluctuations of the signal in the signal energy are not reflected. By contrast, short window lengths result in a course of the signal energy that largely follows the temporal structure of the acoustic signal. The block lengths T for calculating the energy correspond to the above values and can be between 10 ms and 1,500 ms.

The calculation results obtained from the energy calculation and the spectrum analyzer 52 are subsequently processed in the classifier 54. The classifier 54 is capable of determining certain data patterns generated by the Calculation results are formed to classify. That is to say, when vehicles 10 with different numbers of axles pass by, certain data patterns result for the calculated energy as well as for the spectrum, which the classifier 54 can classify by the learning patterns previously presented to it. The classifier 54 can be operated in two modes.

The first mode is also called a learning mode. The learning mode serves to generate a mathematical model for each class to be recognized. A learning algorithm is a sequence of mathematical calculation steps to iteratively approximate a mathematical model. Learning algorithms such as the Hebbian learning method, the backpropagation rule and the forward-backward algorithm are known from the literature. The second mode is called the detection mode and represents the mode used in the normal operation of the detection system. During the recognition phase, one of the K signal classes is allocated to a currently present signal, and thus the number of axles of the currently passing vehicle 10 is determined. The result of the classification is passed on to the communication unit 8.

Claims (17)

1. A contactless axle counter for road traffic, comprising at least one measuring device (1) for detecting changes in the ambient parameters such as noise, temperature or vibration caused by a traveling vehicle (10) in a measuring section, characterized in that the at least one measuring device (1) is fixed above the pavement (20) and above the height of the vehicle (10).
2. An axle counter system according to claim 1, characterized in that the at least one of the at least one measuring device (1) is arranged as a laser vibrometer (1').
3. An axle counter system according to claim 1, characterized in that the at least one of the at least one measuring device (1) is arranged as a directional microphone (1"), especially as a microphone array system.
4. An axle counter system according to one of the claims 1 to 3, characterized in that an analyzing unit (2) is provided for analyzing the measuring signals.
5. An axle counter system according to claim 4, characterized in that the analyzing unit (2) comprises a signal amplification device (3) for amplifying the measuring signals, especially by a factor of 100 to 10,000.
6. An axle counter system according to one of the claims 4 or 5, characterized in that the analyzing unit (2) comprises an analog-to-digital converter (4) with a sampling rate of especially 1 kHz to 50 kHz, or more specially between 6 kHz and 16 kHz.
7. An axle counter system according to one of the claims 4 to 6, characterized in that the analog-to-digital converter (4) has a resolution of 8 to 24 bits, preferably 12 to 20 bits.
8. An axle counter system according to one of the claims 6 or 7, characterized in that the analyzing unit (2) comprises a computing unit (5) which is connected with the analog-to-digital converter (4) in outgoing circuit.
9. An axle counter system according to one of the claims 6 to 8, characterized in that a memory (6) and/or a user interface (7) is connected to the computing unit.
10. An axle counter system according to one of the claims 6 to 9, characterized in that the computing unit (5) comprises a time segmenting device (51) for dividing discrete time signals into time signal blocks, especially having a block length of 10 ms to 1,500 ms.
11. An axle counter system according to one of the claims 6 to 10, characterized in that the computing unit (5) comprises a spectral analyzer (52) for calculating the frequency spectrum of the measuring signals and/or an energy calculating unit (53) for calculating the signal energy, especially the short-term signal energy.
12. An axle counter system according to claim 11, characterized in that the computing unit (5) comprises a classifier (54) which is connected in outgoing circuit to the spectral analyzer (52) and/or the energy calculating unit (53) and which classifies the data pattern generated from the respective calculation results of the spectral analyzer (52) and/or the energy calculating unit (53).
13. An axle counter system according to claim 12, characterized in that the classifier (54) can be operated in a learning mode for generating mathematical models for certain classes of vehicles.
14. An axle counter system according to claim 13, characterized in that the learning mode can be operated by using learning algorithms.
15. An axle counter system according to claim 14, characterized in that the learning algorithm can be operated by using the Hebbian learning method, back propagation rule or the forward-backward algorithm.
16. An axle counter system according to claim 12 to 15, characterized in that the classifier (54) can be operated in a recognition mode for allocating a current measuring signal to a vehicle class.
17. An axle counter system according to one of the claims 12 to 16, characterized in that a communication unit (8) is connected in outgoing circuit with the analyzing unit (2), which communication unit is supplied with the calculation result of the classifier (54).

Images