JP4224411B2 - Vehicle detection device - Google Patents

Description

  The present invention relates to a vehicle detection device that detects a vehicle traveling on a road and is used for measuring a traffic flow rate, and more particularly to a magnetic field sensitive vehicle detection device that determines passage of a vehicle from a change in magnetic field by the vehicle.

  If a magnetic field is detected at a certain point in the vicinity of the road, a certain value that is constant in time is obtained. This is due to geomagnetism. When a vehicle that is a lump of magnetic material passes on the road, magnetic flux concentrates on the magnetic material, so that the magnetic field at the point changes with time. When the vehicle is magnetized, the magnetic field is stronger if the direction of magnetization is the same as the direction of geomagnetism, and the magnetic field is weaker if the direction of magnetization is opposite to the direction of geomagnetism. It changes a lot. Patent Documents 1 to 7 describe such a magnetic field sensitive vehicle detection device.

  In these vehicle detection devices, when the vehicle passes on the road, the output (intensity / amplitude) of the magnetic sensor installed near the road changes. Based on the output waveform of the magnetic sensor at this time, it is determined that the vehicle has passed by capturing an event such as exceeding a threshold value. By counting the number of vehicle detection signals notifying this determination, the number of vehicles passing through can be obtained. Since the two magnetic sensors are installed at an interval along the longitudinal direction of the road, a vehicle detection signal having a time difference can be obtained from the two magnetic sensors with respect to the passage of one vehicle. By dividing the distance between the magnetic sensors by the time difference, the passing speed of the vehicle can be obtained.

  FIG. 7 shows a conventional vehicle detection algorithm, and FIG. 8 shows an output waveform of the magnetic sensor.

  In step 15, the output value (output waveform 19) of each magnetic sensor is read, and in step 16, a preset offset value (due to geomagnetism) 21 is subtracted from the output value to obtain a measured value. In step 17, vehicle measurement calculation (determination of vehicle passage) is performed, and in step 18, a determination value (determination result) is output to the outside. Repeat these steps periodically.

  The contents of the vehicle measurement calculation will be described. When the measured value is outside the range between the preset upper threshold 22 and lower threshold 23, it is determined that the vehicle is present (passed), and the determination value 20 is turned ON. Even if the measured value returns within the range, the determination by the measured value is suspended until the filling time 27 elapses after the vehicle is finally determined to be present, and the measured value is again within the range during the filling time 27. When the vehicle goes outside, it is determined that the same vehicle is present, and it is determined that there is no vehicle if the measured value stays within the above range after the hole filling time 27 has elapsed. Since one magnetic sensor detects a magnetic field in one direction in one axis, if a magnetic field due to passing through the vehicle remains around the magnetic sensor, a magnetic field in the opposite direction due to passing through the next vehicle cannot be detected well, and the judgment value 20 May be turned OFF, but the tendency of the judgment value 20 to be turned OFF can be reduced by the processing using the hole filling time 27.

  Patent Document 2 shows that hysteresis is given to a threshold value. Patent Document 3 discloses that the output value of a magnetic sensor is differentiated. Patent Document 4 discloses that the first peak and the last peak are detected from the output value of the magnetic sensor.

JP-A-6-325288 Japanese Patent Laid-Open No. 9-138284 JP-A-10-208187 JP 2002-251691 A JP 2003-132487 A JP 2003-187281 A JP 2003-308592 A

  Even if a road has a plurality of lanes and a magnetic sensor is installed in the vicinity of the lane in order to detect a vehicle passing through one of the lanes, a vehicle in which a lane other than the lane is greatly magnetized Is passed, it is determined in the conventional vehicle detection algorithm that the vehicle has passed, and the determination value 20 is turned ON. For this reason, the vehicle passage determination in the lane becomes inaccurate, and the accuracy of vehicle detection decreases.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a vehicle detection device that solves the above-described problems and can accurately determine the passage of a vehicle in a required lane.

In order to achieve the above object, the present invention provides a magnetic sensor for measuring a magnetic field generated by a vehicle passing on a road with a plurality of different axes, a memory device for storing a time series of these measured values as waveform data, As an evaluation factor from the waveform data of the measurement axis, a pulse starts when the absolute value of the time differential value of the measurement value exceeds a preset threshold value, and then the absolute value of the time differential value exceeds the preset threshold value again. When the pulse ends, the pulse length Ti that is the time from the start to the end of the pulse , the absolute value integration value Si that is the time integration value at the time Ti of the absolute value of the measurement value, and the time Ti of the measurement value at the time Ti time integration value in which the integrated value Ai, the measured value and shape factor Pi is a time integral value at time Ti of the product of time (i is measured axis number) and calculating means for calculating each The vehicle on the basis of these evaluation elements in which it and the vehicle has passed and a determination unit configured to determine the lane that has passed.
Each of the time integral values may be a time integral value in a time range where the time t when the measured value at the center of the pulse length Ti is 0 is 0 and the time t is from −Ti / 2 to + Ti / 2.

  The determination means calculates two first evaluation amounts S1 / T1 and A1 / S1 from evaluation elements related to an arbitrary one measurement axis (i = 1), and a waveform is obtained by combining these two first evaluation amounts. Data may be classified and lanes may be determined based on this classification.

  The determination unit may have a correlation map between the two first evaluation quantities in advance, and the classification may be performed using the correlation map.

  The determination means calculates two second evaluation amounts P2 / (S2T2) and (S1-S2) / (S1 + S2) from evaluation elements related to any two measurement axes (i = 1, 2), You may determine by the combination of 2nd evaluation amount.

  The determination unit may have a correlation map between the two second evaluation quantities for each classification in advance, and may perform determination using a correlation map according to the classification.

  A magnetic sensor having two or three detection axes may be installed at one location as the magnetic sensor, and the magnetic field of a desired measurement axis may be obtained by performing rotation calculation on the output value of each detection axis.

  The present invention exhibits the following excellent effects.

  (1) Since the passage of the vehicle and the lane through which the vehicle has passed can be determined, the passage of the vehicle in the required lane can be accurately determined.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

  1A and 1B show an example of the arrangement of magnetic sensors in the vehicle detection device according to the present invention. As shown in FIG. 1A, when viewed from the front of the vehicle, the two magnetic sensors 1, 2 are embedded in one lane, for example, below the right lane. The depth from the road surface to the magnetic sensors 1 and 2 is arbitrary, and in the case of a bridge or the like, the magnetic sensors 1 and 2 may be attached to the back side of the bridge. Each of the magnetic sensors 1 and 2 has a single detection axis (direction unique to the magnetic sensor), and the detection axes are set to coincide with desired measurement axes (directions to be measured). In other words, in the illustrated example, the measurement axis 3 of the magnetic sensor 1 is tilted somewhat to the right while the measurement axis 4 of the magnetic sensor 2 is tilted somewhat to the left. Further, as shown in FIG. 1B, the magnetic sensor 1 is installed downstream of the magnetic sensor 2 than the magnetic sensor 2. Thus, the two magnetic sensors 1 and 2 are installed at intervals along the longitudinal direction of the road. The measurement axis number i of the magnetic sensor 1 is set to 1, and the measurement axis number i of the magnetic sensor 2 is set to 2.

  In many vehicles, when the magnetic field is measured directly below the vehicle, the vertical magnetic field changes mainly due to the concentration of geomagnetism. On the other hand, when the magnetic field is measured at a location away from just below the vehicle, the magnetic field in a direction substantially parallel to the straight line connecting the vehicle and the magnetic sensor mainly changes. The magnetic sensor 1 on the downstream side often detects the magnetic field generated by the vehicle on the right lane, but does not detect the magnetic field generated by the vehicle on the left lane. On the other hand, the magnetic sensor 2 on the upstream side well detects the magnetic field generated by the vehicle on the right lane and also detects the magnetic field generated by the vehicle on the left lane. Therefore, the output waveforms of the magnetic sensors 1 and 2 differ between when the vehicle passes through the right lane and when the vehicle passes through the left lane.

  The vehicle detection algorithm will be described with reference to FIG.

  In step 5, the output waveforms of the magnetic sensors 1 and 2 are sampled at a predetermined sampling period, a time series of measurement values is read, and a time series for a predetermined period is stored in the memory device as waveform data. The measured value is obtained by subtracting a preset offset value (due to geomagnetism) from the output values of the magnetic sensors 1 and 2.

  In step 6, the calculation means calculates an evaluation factor for each stored waveform data. Evaluation elements are a pulse length Ti, an absolute value integration value Si, an integration value Ai, and a shape factor Pi (i is a measurement axis number). The calculation formula for each evaluation element is as follows. The time t, measured value Hi, and integration range of each calculation formula are shown in FIG.

  FIG. 3 shows a typical waveform common to all measurement axes. Actually, the waveform is different for each measurement axis. As shown in FIG. 3, the pulse length Ti is a time for the measured value Hi to return to 0 after passing through the positive maximum peak, 0, and negative maximum peak, and corresponds to the magnetic field change due to the passage of one vehicle. . The method of calculating the pulse length Ti is to calculate the time differential value (speed of change of the magnetic field) of the measured value Hi, the pulse starts when the absolute value of the differential value exceeds a preset threshold value, and then again The pulse ends when the absolute value of the differential value exceeds a preset threshold, and the time from the start to the end of the pulse is the pulse length Ti. However, during the hole filling time 27 described with reference to FIG. 8, the end of the pulse is not determined and the pulse is considered to be continuing.

  In the waveform of FIG. 3, the time t at Hi = 0 at the center of the pulse length Ti is set to 0, and is distributed to ± Ti / 2. The pulse length Ti includes vehicle speed information. The absolute value integration value Si includes information on the magnitude of the magnetic field disturbed by the vehicle. Each of the integral value Ai and the shape factor Pi includes information representing the characteristics of the waveform.

  In step 7, two first evaluation quantities S1 / T1 and A1 / S1 are calculated from the evaluation elements for the measurement axis 3 (i = 1). Since the measuring axis 3 of the magnetic sensor 1 forms an angle close to orthogonal to the line connecting the magnetic sensor 1 and the vehicle on the left lane, the magnetic field generated by the vehicle on the left lane is detected small. The magnetic field generated by the vehicle on the right lane is not so small. That is, the absolute value of the measurement value H1 of the measurement shaft 3 varies depending on the lane through which the vehicle passes. For this reason, a lane can be roughly distinguished by the magnitude | size of 1st evaluation amount S1 / T1. The other first evaluation amount A1 / S1 is obtained by quantifying the waveform characteristic on the measurement axis 3. By combining the two first evaluation quantities, the lanes can be distinguished with higher accuracy.

  As a means for performing this distinction (also referred to as classification), a correlation map between the two first evaluation quantities may be created. As shown in FIG. 4, points based on a plurality of waveform data measured in the past are plotted on two-dimensional coordinates in which the horizontal axis is S1 / T1 and the vertical axis is A1 / S1. The dark point with the own lane and the annotation is the point by the waveform data when the vehicle passes the right lane where the magnetic sensor 1 is installed, and the thin point with the adjacent lane and the annotation has passed the left lane It is a point according to the waveform data. Looking at this correlation map, the points of the own lane and the points of the adjacent lane are distributed in an uneven manner. Based on this distribution, it is possible to demarcate the own lane region 9 having only the point of the own lane and the adjacent lane region 10 having only the point of the adjacent lane. Further, for a portion where the points of the own lane and the points of the adjacent lane are mixed, the high density shared area 11 having a relatively large number of points on the own lane and the low density shared area 12 having a relatively small number of points on the own lane are demarcated. Can do.

  When the two first evaluation quantities based on the measured waveform data are applied to the correlation map of FIG. 4, the waveform data can be classified into any of the four types of regions 9 to 12. When the vehicle is classified into the own lane region 9 and the vehicle is classified into the adjacent lane region 10, the lane determination is immediately determined from this classification.

  In step 8, two second evaluation quantities P2 / (S2T2) and (S1-S2) / (S1 + S2) are calculated from the evaluation elements related to the measurement axes 3, 4 (i = 1, 2). The second evaluation amount P2 / (S2T2) is obtained by quantifying the waveform feature on the measurement axis 4. Another second evaluation amount (S1−S2) / (S1 + S2) is a comparison value of the strength of the magnetic field of the measurement axes 3 and 4. That is, if the difference between the magnetic fields of the two measurement axes 3 and 4 is small, the value of the second evaluation amount is small. If the difference between the magnetic fields of the two measurement axes 3 and 4 is large, the second evaluation amount has a large value.

  As a means for making a determination from a combination of two second evaluation amounts, a correlation map between the two second evaluation amounts may be created. FIG. 5 is a correlation map corresponding to the high-density shared region 11 in FIG. 4, and FIG. 6 is a correlation map corresponding to the low-density shared region 12 in FIG. As shown in FIG. 5 and FIG. 6, points based on a plurality of waveform data measured in the past are plotted on two-dimensional coordinates with the horizontal axis P2 / (S2T2) and the vertical axis (S1-S2) / (S1 + S2). Plot. The dark point with the own lane and the annotation is the point by the waveform data when the vehicle passes the right lane where the magnetic sensor 1 is installed, and the thin point with the adjacent lane and the annotation has passed the left lane It is a point according to the waveform data. Looking at these correlation maps, the points in the own lane and the points in the adjacent lane are distributed in each correlation map. Based on these distributions, it is possible to draw a high-density boundary line 13 and a low-density boundary line 14 that divide a region having only a point on the own lane and a region having only a point on the adjacent lane on each correlation map.

  In the case of the high-density shared region 11 in the classification of step 7, two second evaluation quantities based on the measured waveform data are applied to the correlation map of FIG. In the case of the low density shared region 12 in the classification of step 7, two second evaluation quantities based on the measured waveform data are applied to the correlation map of FIG. The lane can be determined based on which side of the high-density boundary line 13 or the low-density boundary line 14 is located.

  Since the first evaluation amount S1 / T1 and the second evaluation amount P2 / (S2T2) are divided by the pulse length T1 or T2, it is suitable for evaluating waveform data having different vehicle speeds on the same scale. . Similarly, the first evaluation amount A1 / S1 is obtained by dividing the integral values obtained in the same integration range, and the second evaluation amount (S1-S2) / (S1 + S2) is also a ratio. Is suitable for evaluating waveform data with different scales on the same scale.

  Next, in step 9, the judgment value (judgment result) is output to the outside. Repeat these steps periodically.

  As can be seen from the above description of the algorithm, in the present invention, since the evaluation element having various information is extracted from the time series of the measurement values of a certain time width, rather than simply comparing the instantaneous values of the threshold values and the measurement values, The passing lane can be accurately determined. Further, since the first and second evaluation quantities do not depend on the vehicle speed, it is easy to compare with many past waveform data. Furthermore, by using the correlation map, the determination result can be directly read from the two evaluation amounts.

  In the embodiment of FIG. 1, two magnetic sensors having one measurement axis are provided at intervals, but a multi-axis magnetic sensor having two or three detection axes in one sensor housing. Even if it installs in one place of the road longitudinal direction, this invention can be implemented.

  The magnetic sensor 91 shown in FIG. 9 is a multi-axis magnetic sensor having two detection axes orthogonal to each other inherent to the magnetic sensor 91. Only one magnetic sensor 91 is used in place of the magnetic sensors 1 and 2 in FIG. 1, and the magnetic sensor 91 is buried under a predetermined lane. At this time, the two detection axes are oriented in the x direction parallel to the road surface and the z direction perpendicular to the road surface. Thereby, the magnetic sensor 91 can output the x direction component Hx and the z direction component Hz of the magnetic field when the vehicle passes.

  By applying the following rotation calculation to these output values Hx and Hz, the magnetic field of the desired measurement axis, that is, the measurement values Hi of the two measurement axes 3 and 4 obtained by the two magnetic sensors 1 and 2 in FIG. (I = 1, 2) can be obtained by one magnetic sensor 91.

  Here, assuming that the inclination of the measurement axis 3 with respect to the z-axis is θ1, and the inclination of the measurement axis 4 with respect to the z-axis is θ2, the virtual magnetic sensor (corresponding to the magnetic sensor i) having the measurement axes 3 and 4 as the detection axes. The output Hi is expressed by the following equation by rotation calculation.

It is a layout view of a magnetic sensor showing an embodiment of the present invention. It is a flowchart of the vehicle detection algorithm which shows one Embodiment of this invention. It is an output waveform diagram of a magnetic sensor showing an embodiment of the present invention. It is a figure of the correlation map used for this invention. It is a figure of the correlation map used for this invention. It is a figure of the correlation map used for this invention. It is a flowchart of the conventional vehicle detection algorithm. It is an output waveform figure of a magnetic sensor. It is a layout view of a magnetic sensor showing an embodiment of the present invention.

Explanation of symbols

1, 2 Magnetic sensor 3, 4 Measuring axis

Claims (7)

As an evaluation element from the magnetic sensor that measures the magnetic field by the vehicle passing on the road with multiple axes with different directions, the memory device that stores the time series of these measured values as waveform data, and the waveform data of each measurement axis,
The pulse starts when the absolute value of the time differential value of the measured value exceeds a preset threshold value, and then the pulse ends when the absolute value of the time differential value exceeds the preset threshold value again. Pulse length Ti, which is the time from the beginning to the end ,
An absolute value integral value Si, which is a time integral value at the time Ti of the absolute value of the measured value ,
An integrated value Ai, which is a time integrated value of the measured value at time Ti ,
The calculation means for calculating the shape factor Pi (i is the measurement axis number) that is the time integral value at the time Ti of the product of the measurement value and the time, the vehicle passing based on these evaluation factors, and the vehicle A vehicle detection apparatus comprising: determination means for determining a lane that has passed. Each time integral value is a time integral value in a time range from -Ti / 2 to + Ti / 2, where time t when the measured value at the center of the pulse length Ti is 0 is 0. The vehicle detection device according to claim 1. The determination means calculates two first evaluation amounts S1 / T1 and A1 / S1 from evaluation elements related to an arbitrary one measurement axis (i = 1), and a waveform is obtained by combining these two first evaluation amounts. 3. The vehicle detection device according to claim 1, wherein the data is classified and the lane is determined based on the classification. 4. The vehicle detection apparatus according to claim 3 , wherein the determination unit has a correlation map between the two first evaluation quantities in advance, and performs the classification using the correlation map. The determination means calculates two second evaluation amounts P2 / (S2T2) and (S1-S2) / (S1 + S2) from evaluation elements related to any two measurement axes (i = 1, 2), The vehicle detection device according to claim 3, wherein the determination is made based on the second evaluation amount. The vehicle detection according to claim 5 , wherein the determination unit has a correlation map between the two second evaluation amounts for each of the classifications in advance, and performs determination using a correlation map according to the classification. apparatus. A magnetic sensor having two or three detection axes is installed at one location as the magnetic sensor, and a magnetic field of a desired measurement axis is obtained by performing rotation calculation on the output value of each detection axis. vehicle detection device in accordance with claim 1-6.

Images