KR102242554B1 - Inspection target selection apparatus and method for evaluating the performance of vehicle detectors - Google Patents

Abstract

The present invention relates to an inspection target selection method for evaluating performance of a vehicle detector. The inspection target selection method for evaluating performance of a vehicle detector includes the steps of: (a) generating evaluation reference data, which is a criterion for performance evaluation, based on traffic data collected from at least one of a plurality of vehicle detectors; and (b) selecting, as an inspection target, a vehicle detector determined to be in need of inspection among the plurality of vehicle detectors by using the evaluation reference data.

Description

Inspection target selection device and method for evaluating the performance of vehicle detectors {INSPECTION TARGET SELECTION APPARATUS AND METHOD FOR EVALUATING THE PERFORMANCE OF VEHICLE DETECTORS}

The present application relates to an apparatus and method for selecting an inspection object for evaluating the performance of a vehicle detector.

In the Intelligent System, data collected through vehicle detectors (for example, collected data such as traffic volume, speed, occupancy time, etc.) are processed to provide traffic information such as section travel time, moving speed, congestion, and traffic accidents. It is possible to generate. At this time, if there is an error in the primary data such as traffic volume, speed, occupancy time, etc., which are data collected by the vehicle detector, the traffic information produced based on this will of course also include the error, and the wrong traffic information is transmitted to the driver and the public. . Therefore, in order to generate reliable traffic information, it can be said that the accuracy of the collected data, which is the primary data collected by the vehicle detector, is very important.

In general, the installation interval of vehicle detectors is about 2.0km for general national highways, about 1km for highways, and is installed and operated on major roads for urban roads.

Table 1 below shows, as an example, the installation status of video-type vehicle detectors of five regional homeland management offices as of 2019. The accuracy of a large number of vehicle detectors operated in this way is not always kept constant. Therefore, from the viewpoint of managing this, it is very important to automatically select the equipment whose performance is suspicious or the equipment whose performance is degraded from among the large number of vehicle detectors and tune it to exhibit its original accuracy. .

[Table 1]

However, the basis for establishing regular evaluation criteria for conventional intelligent transportation systems (ITS) equipment is insufficient, and some local governments do not perform regular evaluation for reasons such as budget. In addition, conventionally, since only one lane is evaluated (the vehicle detector performance evaluation standard notified by the Ministry of Land, Infrastructure and Transport), the performance evaluation for the remaining lanes is insufficient, and there is a problem that the performance evaluation standard equipment cannot be installed depending on the site conditions.

Accordingly, there is a demand for the development of a technology capable of effectively selecting equipment (vehicle detectors) suspected of deteriorating performance among a large number of vehicle detectors and tuning them to maintain the original performance.

The technology behind the present application is disclosed in Korean Utility Model Publication No. 20-0375368.

The present application is to solve the problems of the prior art described above, a vehicle detector that can effectively select equipment (vehicle detectors) that are suspected of deteriorating performance among a large number of vehicle detectors and tune them to maintain the original performance. The purpose of this study is to provide an inspection target selection device and method for evaluating the performance of the product.

The present application is intended to solve the problems of the prior art described above, and the problem of the conventional vehicle detector performance evaluation technology (that is, the basis for setting the regular evaluation criteria of ITS equipment is insufficient, and some local governments are due to budget, etc.) The performance of vehicle detectors that can solve the problem of not performing regular evaluation, and because of the conventional evaluation of only one lane, the performance evaluation of the remaining lanes is insufficient, and the problem of not being able to install performance evaluation standard equipment depending on the site conditions. Its purpose is to provide an apparatus and method for selecting inspection targets for evaluation.

However, the technical problem to be achieved by the embodiments of the present application is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, the method for selecting an inspection object for evaluating the performance of a vehicle detector according to an embodiment of the present application includes: (a) based on traffic data collected from at least one of a plurality of vehicle detectors. Generating evaluation criterion data that becomes a criterion for performance evaluation; And (b) selecting, among the plurality of vehicle detectors, a vehicle detector that is determined to require inspection, as an inspection target, using the evaluation criterion data.

In addition, the plurality of vehicle detectors may be vehicle detectors included in the same target road.

In addition, in the step (a), the evaluation criterion data includes data on attribute classification given to any one of the plurality of vehicle detectors based on a past measurement value of one of the plurality of vehicle detectors, The above attribute classification is a first attribute type, which is a road type with a peak time in the morning, a second attribute type, which is a road type with a peak time in the afternoon, a third attribute type that is a road type with morning and afternoon peak hours, and a road without peak hours. It includes an attribute type classification for each time period including a fourth attribute type that is a type, and the step (b) is obtained from the one vehicle detector in comparison with the attribute type donated to the one vehicle occupant. When the attribute type assigned based on the evaluation target measurement value is at least partially different, the one vehicle detector may be selected as the inspection target.

In addition, the attribute classification is a fifth attribute type in which one of the two average daily traffic volumes is a large road type, and a road type in which the other of the two average daily traffic volumes is large by comparing the average daily traffic volume during the week and the average daily traffic volume on the weekend. It may further include classification of attribute types for each day of the week including the sixth road type and the seventh road type, which is a road type with the same average daily traffic amount.

In addition, the attribute classification is performed by comparing the average daily traffic volume of some months of a year and 12 months of the year and the average daily traffic volume of the remaining months excluding the partial months of the 12 months of a year. A monthly attribute type classification including an eighth attribute type, a ninth road type in which the other of the two average daily traffic volumes is a large road type, and a tenth road type in which the two average daily traffic amounts are the same road type may be further included.

In addition, the past measurement value corresponds to the past accumulated data accumulated during the (N-1) year, and the evaluation target measurement value corresponds to the recent accumulated data accumulated during the last year excluding the period of (N-1). It can be.

In addition, in the step (b), based on the attribute classification, an inspection object is first selected among the plurality of vehicle detectors, and among the plurality of vehicle detectors, the unselected vehicle detectors that are not first selected are targeted, the Among the unselected vehicle detectors, the inspection target is secondarily selected based on the traffic volume estimation model to determine the inspection target, and at the second screening, based on the accumulated data of any unselected vehicle detector among the unselected vehicle detectors. Secondary screening may be performed in consideration of a difference between the traffic volume estimation value calculated through the traffic volume estimation model and the traffic volume observation value collected from one of the unselected vehicle detectors.

In addition, the evaluation criterion data includes data on an annual average traffic volume change pattern for each time slot for a daily traffic volume, for any one of the plurality of vehicle detectors, and the step (b) includes: When the similarity between the traffic volume change pattern and the traffic volume change pattern obtained from any one of the plurality of vehicle detectors is evaluated, and if the similarity is out of the allowable range set in advance, one of the vehicle detectors is selected as the inspection object. I can. Here, the similarity may be evaluated by a correlation coefficient that is an index that quantifies the difference between two patterns to be compared.

In addition, the similarity between the two patterns may be evaluated in consideration of an increase/decrease rate of the traffic volume for each time slot obtained from any one of the vehicle detectors compared to the annual average traffic volume for each time slot.

In addition, in step (b), when determining the degree of similarity, a correlation coefficient calculated using traffic data belonging to a predetermined time zone among the entire day time zone is considered as a similarity evaluation index, and the predetermined time zone is The traffic volume variation may include a time zone corresponding to a predetermined variation amount or more, or a time zone corresponding to the traffic volume of each time zone equal to or greater than a predetermined traffic volume.

In addition, the evaluation criterion data includes data on an average annual increase/decrease rate of traffic volume of each of the plurality of vehicle detectors, and the step (b) includes data on an annual average increase/decrease rate of traffic volume of any one of the plurality of vehicle detectors. If the similarity between the data on the average annual increase/decrease in traffic volume of the other vehicle detector among the plurality of vehicle detectors is evaluated and the similarity is out of the allowable range, one of the vehicle detectors is likely to fall under the inspection target. I can judge.

In addition, the step (a) generates the average annual increase/decrease rate of the traffic volume for each group of a plurality of predetermined time zones in consideration of the effect of the shadow due to the movement of the sun during the daytime and the effect of the absence of the sun at night, and the step (b) , When the difference between the average annual increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones generated for any one of the plurality of vehicle detectors is out of a preset allowable range, the one of the vehicle detectors is selected as the inspection object. I can.

In addition, the evaluation criterion data includes data on an estimate of the traffic volume in n time zones estimated using a traffic volume estimation model in the adjacent time zone with respect to any one of the plurality of vehicle detectors, and the step (b) Is, when the traffic volume estimate in the n time zone is compared with the traffic volume observation value in the n-time zone collected by the any one vehicle detector, and if it is out of a preset confidence tolerance range, the one vehicle detector may be selected as the inspection object. .

On the other hand, the inspection target selection device for evaluating the performance of a vehicle detector according to an embodiment of the present application is generated to generate evaluation reference data that is a criterion for performance evaluation based on traffic data collected from at least one of a plurality of vehicle detectors. part; And a selection unit that selects a vehicle detector determined to need to be inspected among the plurality of vehicle detectors using the evaluation criterion data as an inspection target.

In addition, the plurality of vehicle detectors may be vehicle detectors included in the same target road.

In addition, the evaluation criterion data includes data on an average annual traffic volume change pattern for each time slot with respect to the daily traffic volume, with respect to any one of the plurality of vehicle detectors, and the selection unit comprises: When the similarity between the traffic volume change pattern for each time period obtained from any one of the plurality of vehicle detectors is evaluated and the similarity is out of the allowable range, any one of the vehicle detectors may be selected as the inspection object. Here, the similarity may be evaluated by a correlation coefficient that is an index that quantifies the difference between two patterns to be compared.

In addition, the similarity between the two patterns may be evaluated in consideration of an increase/decrease rate of the traffic volume for each time slot obtained from any one of the vehicle detectors compared to the annual average traffic volume for each time slot.

In addition, when determining the similarity, the selection unit considers a correlation coefficient calculated using traffic data belonging to a predetermined time zone among the entire day as a similarity evaluation index. A time zone corresponding to a predetermined amount of fluctuation or more, or a time zone corresponding to a predetermined amount of traffic or more may be included.

In addition, the evaluation criterion data includes data related to an average annual increase/decrease rate of traffic volume of each of the plurality of vehicle detectors, and the selection unit includes data related to an annual average increase/decrease rate of traffic volume of any one of the plurality of vehicle detectors, and the plurality of If the similarity between the data on the annual average increase/decrease rate of traffic volume of the other vehicle detector of is evaluated and the degree of similarity is out of the allowable range, it can be determined that one of the vehicle detectors is likely to fall under the inspection target. have.

In addition, the generation unit generates the average annual traffic volume increase/decrease for each group of a plurality of predetermined time zones in consideration of the influence of the shadow due to the movement of the sun during the daytime and the effect of the absence of the sun at night, and the selection unit includes the plurality of vehicle detectors. When the difference between the average annual increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones generated for any one of the vehicle detectors exceeds a preset allowable range, the one of the vehicle detectors may be selected as the inspection object.

In addition, the evaluation criterion data includes data on an estimate of a traffic volume in n time zones estimated using a traffic volume estimation model in a predetermined adjacent time zone with respect to any one of the plurality of vehicle detectors, and the selection unit comprises: When the traffic volume estimate in n time zones is compared with the traffic volume observation value in n-time zones collected by any one of the vehicle detectors, and the vehicle detector is outside a preset allowable range, the one vehicle detector may be selected as the inspection object.

The above-described problem solving means are merely exemplary and should not be construed as limiting the present application. In addition to the above-described exemplary embodiments, additional embodiments may exist in the drawings and detailed description of the invention.

According to the above-described problem solving means of the present application, by providing an inspection target selection device and method for evaluating the performance of the vehicle detector, effectively selecting the equipment (vehicle detector) suspected of deteriorating performance among a large number of vehicle detectors, It can be tuned to maintain its original performance.

According to the above-described problem solving means of the present application, by providing an inspection target selection apparatus and method for evaluating the performance of the vehicle detector, the problem of the conventional vehicle detector performance evaluation technology (that is, the basis for setting the periodic evaluation criteria of the ITS equipment is It is insufficient, and some local governments do not conduct regular evaluation for reasons such as budget, and performance evaluation for the remaining lanes is insufficient due to the conventional evaluation of only one lane, and performance evaluation standard equipment can be installed depending on the site conditions. Problems, etc.) can be solved.

However, the effect obtainable in the present application is not limited to the above-described effects, and other effects may exist.

1 is a block diagram showing a schematic configuration of an inspection target selection apparatus for evaluating the performance of a vehicle detector according to an embodiment of the present application.
2A is a diagram schematically illustrating an installation example of a plurality of vehicle detectors considered in the inspection target selection apparatus for evaluating the performance of the vehicle detector according to an exemplary embodiment of the present disclosure.
2B is a diagram illustrating a process of selecting an inspection object using data on attribute classification in the inspection object selection apparatus for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure.
2C is a diagram illustrating a process of selecting an inspection object using data on attribute classification in the inspection object selection apparatus for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure.
3 is a view showing an example of a pattern of changes in average annual traffic volume for each time period in the inspection target selection device for evaluating the performance of a vehicle detector according to an embodiment of the present application.
4 is a view showing an example of comparing the similarity of two change patterns with respect to the same point as an example in the inspection target selection apparatus for evaluating the performance of a vehicle detector according to an embodiment of the present application.
5 is a diagram schematically illustrating a process of selecting an inspection object by an inspection object selection device for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present application.
6 is a flowchart illustrating an operation of a method for selecting an inspection object for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure.

Hereinafter, embodiments of the present application will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present application. However, the present application may be implemented in various different forms and is not limited to the embodiments described herein. In addition, in the drawings, parts irrelevant to the description are omitted in order to clearly describe the present application, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the present specification, when a part is said to be "connected" with another part, it is not only the case that it is "directly connected", but also "electrically connected" or "indirectly connected" with another element interposed therebetween. "Including the case.

Throughout the present specification, when a member is positioned "on", "upper", "upper", "under", "lower", and "lower" of another member, this means that a member is located on another member. This includes not only the case where they are in contact, but also the case where another member exists between the two members.

In the entire specification of the present application, when a certain part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

1 is a block diagram showing a schematic configuration of an inspection target selection apparatus 10 for evaluating the performance of a vehicle detector according to an embodiment of the present application.

Hereinafter, an inspection target selection device 10 for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure will be referred to as the present device 10 for convenience of description.

Referring to FIG. 1, the apparatus 10 may be a device that selects a vehicle detector to be inspected for performance evaluation among a plurality of vehicle detectors that collect traffic data. The device 10 may include a generating unit 11 and a sorting unit 12.

The generator 11 may generate evaluation reference data, which is a criterion for performance evaluation, based on traffic data collected from at least one of the plurality of vehicle detectors.

Here, the evaluation criterion data is data generated based on past accumulated data on traffic data collected from vehicle detectors, and whether the vehicle detector is an abnormal vehicle detector with a performance abnormality or a normal vehicle detector with no performance abnormality. It may refer to comparison reference data that is a criterion for determining recognition, that is, past-based (reference) data generated based on past accumulated data.

Traffic data is data collected (acquired) by a vehicle detector, and may, for example, refer to data (primary data, collected data) such as traffic volume, speed, and occupancy time.

The selection unit 12 may select, as an inspection target, vehicle detectors that are determined to require inspection (check for performance evaluation) among a plurality of vehicle detectors, using the evaluation reference data generated by the generation unit 11. .

In this case, a plurality of vehicle detectors considered by the present apparatus 10 (ie, a plurality of vehicle detectors considered to identify an inspection target) may be vehicle detectors included in the same target road.

Here, the same target road may refer to a road having the same road name. However, it is not limited thereto, and it is desirable to broadly understand the same target road as a concept encompassing interconnected roads so that vehicle driving can be continuously performed even if the road names are the same as well as different road names. In other words, the same target road may mean a road through which a vehicle having a degree of similarity of a predetermined or more can be driven by being interconnected with each other.

In general, vehicle detectors are installed at intervals along the road extension direction (vehicle traveling direction). For example, vehicle detectors are installed at intervals of about 2.0km in the case of general national highways, about 1km in the case of expressways, and are installed on major roads in the case of urban roads (city roads).

FIG. 2A is a diagram schematically illustrating an installation example of a plurality of vehicle detectors considered in the inspection target selection apparatus 10 for evaluating the performance of the vehicle detector according to an exemplary embodiment of the present disclosure. In particular, FIG. 2A schematically shows an example in which a plurality of vehicle detectors (for example, four vehicle detectors including vehicle detector 1 to vehicle detector 4) are installed on an urban road as an example for grasping road conditions.

Referring to FIG. 2A, a plurality of vehicle detectors considered in the present apparatus 10 may be vehicle detectors included in a preset section even within the same target road. For example, the plurality of vehicle detectors may be vehicle detectors included in a preset section (a pre-set partial section) within the same target road (for example, an urban road, etc.) as shown in FIG. 2A. In other words, as an example, the preset section may be set by specifying a partial section of an urban road as shown in FIG. 2A.

As another example, in the case of a road having a predetermined long section between the access road and the exit road, such as a highway, the preset section within the same target road is, for example, a section that does not include the access road and the exit road connected to the target road (entrance and exit roads). It can be set as a section that is not included as a.

In addition, in the case where the target road is divided into an up and down road, a plurality of vehicle detectors considered in the device 10 may be set as a vehicle detector belonging to one of the up and down roads (any one road). I can.

In addition, the target road considered in the present application may be classified into a plurality of road types including a general national road (general road), an express road, and an urban road (city road). It may refer to vehicle detectors belonging to any one road type (ie, a corresponding road recognized as the same road type) among road types of.

In the present application, as an example, the road type of the target road is exemplified as being divided into three types of roads including general national highways, highways, and urban roads, but this is only an example to aid understanding of the present application, and is limited thereto. no. As another example, the road type of the target road considered in the present application may be classified (classified) into various types according to a usage type, scale, function, and the like. For example, the type of road according to the usage type may include, for example, a general road, an underpass, and an elevated road. Types of roads according to scale may include, for example, optical roads, highways, middle roads, and lanes. Types of roads according to functions may include, for example, interstate roads, auxiliary arterial roads, local roads, and special roads.

As such, the apparatus 10 may select an inspection object among the plurality of vehicle detectors for a plurality of vehicle detectors included in the same target road.

As a process for selecting an inspection object by the present apparatus 10, as will be described later, a number of embodiments may exist (that is, there may be a number of embodiments regarding a method of evaluating traffic data for selecting an inspection object). , A more detailed description is as follows.

In addition to selecting the inspection object according to the first to third embodiments of the present application to be described later, the sorting unit 12 may select the inspection object based on attribute classification as an example. A detailed description of the inspection target selection process based on attribute classification is as follows.

Since traffic is a result of reflecting social and economic activities, it must be maintained even when the year changes, as there is no change in traffic patterns (change patterns in traffic volume) unless there is a special reason. In other words, the traffic volume change pattern using roads has a characteristic that repeats periodically. If there is no reason such as a change in the geometry of the road or a change in traffic purpose, the traffic volume change pattern for a specific road is similar to that of the past traffic volume change pattern. It can be said that it is reasonable to see it as having.

In consideration of these points (that the traffic volume change pattern will have a pattern similar to the past), the selection unit 12, for example, makes the traffic volume change pattern for each time zone, day of the week, and month different from each of the past traffic volume change patterns. Vehicle detectors that appear can be selected for inspection. That is, the selection unit 12 may define the attribute of the traffic volume change pattern as a plurality of groups (eg, three groups) including, for example, by time slot, by day of the week, and by month, in order to select an inspection target based on attribute classification.

In other words, the evaluation reference data generated by the generation unit 11 is based on the past measurement value of any one vehicle detector (for example, the first vehicle detector) among the plurality of vehicle detectors (for example, the first vehicle detector). It may include data on the classification of attributes assigned to the vehicle detector). In this case, in the description of the attribute classification in the present application, hereinafter, for convenience of description, any one vehicle detector will be exemplified as the first vehicle detector.

Here, the attribute classification includes a first attribute type that is a road type with a peak time in the morning, a second attribute type that is a road type with a peak time in the afternoon, and a third attribute type that is a road type with a peak time in the morning and afternoon, and the peak time. An attribute type classification for each time period including a fourth attribute type that is a road type that is not present may be included. According to this, the attribute type classification for each time slot defines the property of the time traffic volume (district/hour) change pattern for each time slot, and may mean, for example, information divided into a first group among a plurality of groups.

In this case, when data on attribute type classification by time zone is considered as data on attribute classification, the time zone may be from 06:00 to 20:00, for example. That is, the data on the classification of attribute types by time slots assigned to any one vehicle detector (first vehicle detector) based on the past measurement values of any one vehicle detector (first vehicle detector) is, for example, from 06:00 to 20:00. It can be classified as the above-described first attribute type to fourth attribute type as a pattern of change over time of the traffic volume (Vol ₆ , Vol ₇ , …, Vol _{20) for each time zone.}

The selection unit 12 is an evaluation target obtained from any one vehicle detector (first vehicle detector) in comparison with the attribute type donated to any one vehicle detector (first vehicle detector) based on past measurement values. When the attribute type assigned based on the measured value is at least partially different, any one vehicle detector (the first vehicle detector) may be selected as an inspection target.

Here, the past measured value may correspond to past accumulated data accumulated during (N-1) years. In particular, past measurement values correspond to past accumulated data accumulated during (N-1) years after any one of the vehicle detectors (for example, the first vehicle detector) was installed (for example, it corresponds to the average data of the past accumulated data). It can be. The measured value to be evaluated may correspond to the recent accumulated data accumulated during the last year excluding the period of (N-1) year (especially, the period from the installation of the vehicle detector to (N-1) year). In other words, the measured value to be evaluated may correspond to the data accumulated in the N year.

As an example, assuming that any one vehicle detector (the first vehicle detector) was installed in 2005, the past measurement values of the first vehicle detector were accumulated from 2005 to 2018 (about 14 years, the past 14 years). The measured value to be evaluated obtained from the first vehicle detector corresponds to the accumulated data in the past, for example, for the last year, and may correspond to the accumulated data in 2019. In this case, the year corresponding to the measurement value to be evaluated is illustrated as 2019 as an example, but this is only an example to aid understanding of the present application, and this may be variously set by user input such as 2020.

At this time, the selection unit 12 is at least partially different from the attribute type donated based on the past measurement value to the attribute type given based on the measurement value to be evaluated for any one vehicle occupant (first vehicle detector). If the judgment is made and the result of the judgment is different, any one of the vehicle detectors (the first vehicle detector) may be selected as the object to be inspected.

As an example, suppose that the attribute type (past attribute type) donated to the first vehicle detector based on the past measurement value was the first attribute type, which is a road type with a peak time in the morning. This is a result of analyzing the past accumulated data acquired and accumulated over the past 14 years by the first vehicle detector, and the pattern of changes in traffic volume at the point (or the road at that point) where the first vehicle detector observes the traffic volume was found in the morning. This may mean that the vehicle appears as a first attribute type (that is, a first attribute type, which is a road type with a peak time in the morning), which tends to exhibit a high road occupancy rate during the time period.

In addition, in order to determine whether the first vehicle detector is a target for inspection, the selection unit 12 is an evaluation target measurement value, for example, the latest accumulated data accumulated for the last year (for example, as the accumulated data for fiscal year 2019, for fiscal year 2019). (Meaning the accumulated data of the traffic data collected in) can be obtained, and an attribute type (recent attribute type) can be assigned based on the obtained measured value to be evaluated. That is, the selection unit 12 may determine and assign an attribute type of the last 2019 year of the corresponding point where the first vehicle detector performs traffic volume observation through analysis of the obtained measurement value to be evaluated. Accordingly, the attribute type donated based on the past measurement value may be referred to differently as a past attribute type, and the attribute type assigned based on the evaluation target measurement value may be referred to differently as the recent attribute type.

At this time, if the recent attribute type is the first attribute type, the selection unit 12 determines the first vehicle detector as a normal vehicle detector because it matches the donated attribute type (past attribute type), and may not select it as an inspection target. have. If the attribute type (that is, the latest attribute type) assigned based on the evaluation target measurement value is any one of the second attribute type to the fourth attribute type, this is different from the first attribute type, which is a donated attribute type, The selection unit 12 may determine that the first vehicle detector is an abnormal vehicle detector whose performance is suspected, and select the corresponding first vehicle detector as an inspection target.

As another example, the second vehicle detector is installed to perform traffic volume observation at a point on a road in the road (city road), and the road in the road is a road that tends to be congested during rush hour. Suppose that a vehicle detector has previously assigned a third attribute type (a third attribute type, which is a road type with peak hours in the morning and afternoon) as an attribute type (past attribute type). In this case, as a result of determining the attribute type (recent attribute type) based on the evaluation target measurement value obtained from the second vehicle detector, suppose that the recent attribute type is, for example, a fourth attribute type, which is a road type without peak time. In this case, since the selection unit 12 differs from the attribute type (ie, the third attribute type) donated based on the past measurement value and the attribute type (ie, the fourth attribute type) based on the measurement value to be evaluated. , The second vehicle detector can be selected as an inspection target. That is, the sorting unit 12 found that the traffic volume at the observation point observed by the second vehicle detector had a peak time in the morning/afternoon in the past, but recently, the second vehicle detector was found to have no peak time. It is suspected that there is a problem in the performance of the vehicle, and the corresponding second vehicle detector can be selected as an inspection target.

In the above-described example, the attribute classification mainly includes the attribute type classification for each time period, but is not limited thereto, and the attribute classification includes at least one of attribute type classification for each time period, attribute type classification for each day of the week, and attribute type classification for each month. can do.

Specifically, the attribute classification is a comparison of the average daily traffic volume on weekdays and the average daily traffic volume on weekends, and the fifth attribute type, which is a road type in which one of the two average daily traffic volumes is larger, and any other other than any one of the two average daily traffic volumes. The attribute type classification for each day of the week including a sixth road type, one of which is a large road type, and a seventh road type, which is a road type with the same average daily traffic amount, may be included. According to this, the attribute type classification for each day of the week is defined as an attribute of a daily traffic volume (unit/day) change pattern or a weekly traffic volume change pattern for each day of the week, and for example, may mean information divided into a second group among a plurality of groups. In the present application, weekdays may mean Monday to Friday, and weekends may mean Saturday and Sunday.

As an example, the fifth attribute type may refer to a road type when the average daily traffic on weekdays is greater than the average daily traffic on weekends (average daily traffic on weekdays> average daily traffic on weekends). The sixth attribute type may mean the type of road when the average daily traffic on weekends is larger than the average daily traffic on weekdays (average daily traffic on weekdays <average daily traffic on weekends). The seventh attribute type may refer to a road type when the average daily traffic on weekdays and the average daily traffic on weekends are the same (average weekly traffic = average daily traffic on weekends).

Here, with respect to the seventh attribute type, the fact that the average daily traffic volume during the week and the average daily traffic volume on the weekend are the same means that the value itself has the exact same (same) value, rather than narrowly interpreted, but a preset allowable range. It can be broadly understood (interpreted) as the concept that the average daily traffic volume on weekdays and the average daily traffic on weekends are identical to each other if they do not exceed the permissible range within the (error range).

As an example, the attribute type previously assigned to the first vehicle detector (past attribute type) is the fifth attribute type, and the attribute type (last attribute type) is the sixth attribute based on the evaluation target measurement value obtained from the first vehicle detector. When indicated as a type, the selection unit 12 may select the first vehicle detector as an object to be inspected because the recently assigned attribute type is different from that of the previously assigned attribute type.

In addition, the attribute classification is based on comparing the average daily traffic volume in some months out of 12 months a year and the average daily traffic volume in the remaining months excluding the above some months out of 12 months in a year. Includes a monthly attribute type classification including the 8 attribute type, the ninth road type in which one of the two average daily traffic volumes other than the above one is the largest road type, and the 10 road type in which the two average daily traffic amounts are the same road type. I can. According to this, the monthly attribute type classification is defined as an attribute of a monthly average daily traffic volume (unit/day) change pattern or a monthly traffic volume change pattern, and may mean, for example, information classified into a third group among a plurality of groups.

Here, some months are illustratively July and August, and the remaining months are remaining months excluding July and August, which may mean January to June and September to December. However, this is only an example for aiding understanding of the present application, and is not limited thereto, and some months may be set in various ways.

As an example, the 8th attribute type is when the average daily traffic volume in the rest of the month is larger than the average daily traffic volume in some months (for example, the average daily traffic volume in July and August <average daily traffic volume in the remaining months excluding July and August) It can mean the type of road when it is. The ninth attribute type is when the average daily traffic in some months is greater than the average daily traffic in the remaining months (e.g., average daily traffic in July and August> average daily traffic in the remaining months excluding July and August). It can mean the type of road. The tenth attribute type is when the average daily traffic volume in some months and the average daily traffic volume in the remaining months are the same (e.g., average daily traffic volume in July and August = average daily traffic volume in the remaining months excluding July and August). It can mean the type of road.

Here, with respect to the tenth attribute type, the fact that the average daily traffic volume for some months and the average daily traffic volume for the remaining months is the same means that the value itself is exactly the same (same) as mentioned in relation to the seventh attribute type. Rather than being narrowly interpreted as a concept that means when to have, it can be broadly understood (interpreted) as a concept that two values are considered to be the same if they do not deviate from the allowable range within a preset allowable range (error range).

As an example, the attribute type previously assigned to the first vehicle detector (past attribute type) is the ninth attribute type, and the attribute type (last attribute type) is the eighth attribute based on the evaluation target measurement value obtained from the first vehicle detector. When indicated as a type, the selection unit 12 may select the first vehicle detector as an object to be inspected because the recently assigned attribute type is different from that of the previously assigned attribute type. On the other hand, if the attribute type (past attribute type) previously assigned to the first vehicle detector is the eighth attribute type, and the attribute type (recent attribute type) is the eighth attribute type based on the measured value to be evaluated, the selection unit 12 Since the recent attribute type has not been different from the previously assigned attribute type, the first vehicle detector may not be selected as an inspection target.

As an example, when analyzing the traffic data (traffic data) collected through an arbitrary vehicle detector installed on the road (for example, the first vehicle detector as any one vehicle detector), according to the analysis result, it is illustratively applicable. The traffic volume change pattern at the observation point where the traffic volume is observed by the vehicle detector (the first vehicle detector) belongs to the first attribute type in the case of the time traffic volume change pattern by time, and the sixth attribute type in the case of the weekly traffic volume change pattern. And, in the case of the monthly traffic volume change pattern, it can be defined as belonging to the ninth attribute type.

In other words, the generation unit 11 may analyze a traffic volume change pattern by time slot/day of the week/month based on past measurement values (past accumulated data), which is traffic data collected in the past from the first vehicle detector. Based on the analysis result, the generation unit 11 may exemplarily assign an attribute classification to the first vehicle detector. In this case, the generation unit 11 may pre-assign a plurality of attribute types including a first attribute type, a sixth attribute type, and a ninth attribute type as data on attribute classification with respect to the first vehicle detector. The generator 11 may assign an attribute type related to each of the attribute type classification by time slot, the attribute type classification by day of the week, and the monthly attribute type classification to the first vehicle detector based on past measurement values. Accordingly, a plurality of attribute types (eg, a first attribute type, a sixth attribute type, and a ninth attribute type) may be previously assigned to the first vehicle detector.

At this time, as mentioned above, unless there is a special reason, it is reasonable to see the past traffic volume change pattern and the recent (current) traffic volume change pattern as being maintained even when the year changes (there are similar forms to each other). When at least one attribute type of a plurality of attribute types previously assigned to an arbitrary vehicle detector (for example, a vehicle detector as a first vehicle detector) is changed (if different), the corresponding arbitrary Vehicle detectors can be selected as targets for inspection.

That is, the selection unit 12 includes, for example, a plurality of past attribute types previously given to the first vehicle detector (for example, a past attribute type donated by a plurality of groups, a past first attribute type, a past sixth attribute type, If any one of them is different from the plurality of attribute types (plural latest attribute types) given based on the evaluation target measured value obtained from the first vehicle detector compared to the past 9th attribute type), the first vehicle detector is used as a target for inspection. Can be selected.

For example, if the selection unit 12 is assigned a'first attribute type, a sixth attribute type, and a ninth attribute type' as a plurality of latest attribute types, all of the previously assigned attribute types are the same ( Since the latest attribute type is not different from the assigned attribute type), the first vehicle detector can be selected as not subject to inspection.

If, the selection unit 12 is assigned a'first attribute type, a fifth attribute type, a ninth attribute type' as a plurality of latest attribute types, or'a second attribute type, a sixth attribute type, and a ninth attribute type' Is given, or if '3rd attribute type, 7th attribute type, 10th attribute type' is given, at least some of the previously assigned attribute types are different from the latest attribute type, so check the first vehicle detector. Can be selected as a target.

The selection unit 12 first selects an inspection object among a plurality of vehicle detectors based on the above-described attribute classification, targets unselected vehicle detectors that are not first selected among the plurality of vehicle detectors, and is an example of unselected vehicle detectors. Based on the traffic volume estimation model, the inspection object can be confirmed by secondly selecting the inspection object. At this time, at the time of the second screening, the selection unit 12 includes a traffic volume estimation value calculated through a traffic volume estimation model based on the accumulated data of any unselected vehicle detector among the unselected vehicle detectors and the unselected vehicle detector. Secondary screening can be performed by considering the difference between the collected traffic volume observations.

In this case, as a traffic volume estimation model considered herein, a deep learning model previously learned to enable traffic volume estimation (prediction) may be used as an example. Such a deep learning model may be a model trained to take a plurality of past traffic data as input data and use traffic volume information matching each of the plurality of past traffic data as an output value.

For example, the deep learning model may mean an artificial intelligence (AI) algorithm model, a machine learning (machine learning) model, a neural network model (artificial neural network model), a neuro fuzzy model, and the like. In addition, various neural network models such as a convolution neural network (CNN, convolutional neural network), a recurrent neural network (RNN), and a deep neural network may be applied as an example of the deep learning model.

In the present application, as an example of the traffic volume estimation model, a previously learned deep learning model is used, but this is only an example to aid understanding of the present application, and is not limited thereto. In this application, various pre-learned models that enable traffic volume estimation can be applied as a traffic volume estimation model, and as a traffic volume estimation model, various factors (means) capable of estimating traffic volume, such as previously known or developed traffic volume estimation equations (equations). Can be applied.

Meanwhile, the selection unit 12 compares the attribute type donated to any one vehicle occupant among the plurality of vehicle detectors (a plurality of past attribute types donated) and the evaluation target measurement value obtained from any one vehicle detector. When the attribute type (latest attribute type) assigned as the basis is at least partially different, one vehicle detector may be first selected as an inspection target. The selection unit 12 may primarily select an inspection object among a plurality of vehicle detectors based on such attribute classification (comparison of attribute types). Thereafter, the selection unit 12 secondly selects the inspection targets based on the traffic volume estimation model among the unselected vehicle detectors, targeting the unselected vehicle detectors, which are vehicle detectors that have not been first selected among the plurality of vehicle detectors. Can be confirmed. The selection unit 12 may finally determine that, as an example of a plurality of vehicle detectors, both of the vehicle detectors that are the firstly selected inspection target and the vehicle detectors that are the secondly selected inspection target are the inspection targets that are determined to require inspection.

According to this, the selection unit 12 targets only vehicle detectors that were not selected in the first screening process (that is, targets unselected vehicle detectors), as an example, and calculates the predicted value and the observed value (year N) predicted through the traffic volume estimation model. After comparing each other, unselected vehicle detectors that show that the difference is out of the preset allowable range (i.e., unselected vehicle detectors that appear to be above a certain level) are secondarily selected as targets for inspection that are vehicle detectors with suspected performance. I can. According to this device 10, the first selected vehicle detector and the second selected vehicle detector among the plurality of vehicle detectors are determined to be inspection targets, and detailed inspection and detailed inspection for performance evaluation are performed only for the determined inspection objects. Can be done.

As an example, a description of the primary selection process of the inspection object may be more easily understood with reference to FIG. 2B.

FIG. 2B is a view for explaining a process for selecting an inspection object (especially, a process for selecting a primary inspection object) using data on attribute classification in the inspection object selection device 10 for evaluating the performance of a vehicle detector according to an embodiment of the present application to be.

Referring to FIG. 2B, the selection unit 12 may select an inspection object by using data on attribute classification as evaluation reference data. Hereinafter, as an example, a process of selecting the first vehicle detector as one of the plurality of vehicle detectors as a target for inspection will be described as an example.

In step S31, the generation unit 11 may obtain a past measurement value of the first vehicle detector in order to determine whether the first vehicle detector is a target for inspection. Here, as the past measurement value of the first vehicle detector, for example, data accumulated during (N-1) years after the first vehicle detector was installed (that is, accumulated data during (N-1) years after the first vehicle detector was installed. Data accumulated in the past) can be used, and this can be used as a reference value.

Thereafter, in step S32, the generation unit 11 may analyze the past measurement value obtained in step S31 to provide data regarding attribute classification to the first vehicle detector. At this time, in step S32, the generation unit 11 may assign a plurality of attribute types (a plurality of past attribute types) as data regarding attribute classification.

Here, the plurality of attribute types include, for example, an attribute type (attribute I) related to the time traffic volume change pattern by time zone, an attribute type (attribute J) related to the daily traffic volume (car/day) change pattern for each day of the week, and the average daily traffic change pattern by month. The attribute type (attribute K) for may be included.

At this time,'attribute I'means information classified by the above-described attribute type classification by time slot (i.e., information classified into the first group), and'attribute J'is information classified by the above-described attribute type classification by day of the week (i.e. , Information classified into a second group), and'attribute K'may mean information classified by the above-described monthly attribute type classification (ie, information classified into a third group). These'attribute I','attribute J', and'attribute K'may be referred to differently as donated past attribute information. According to this, as'attribute I', any one of a past first attribute type to a past fourth attribute type may be assigned. As'attribute J', any one of the past 5th attribute type to the past 7th attribute type may be assigned. As the'attribute K', any one of the past eighth attribute type to the past tenth attribute type may be assigned.

Next, in step S33, the generation unit 11 may obtain a measurement value to be evaluated from the first vehicle detector. Here, as an evaluation target measurement value, as an example, accumulated data of the (N) year as an evaluation target (that is, the recent accumulated data collected and accumulated in the N year) may be used, and this may be used as an evaluation target value.

Next, in step S34, the generation unit 11 may analyze the measurement value to be evaluated acquired in step S33 to provide data on attribute classification to the first vehicle detector. At this time, in step S34, the generation unit 11 may assign a plurality of attribute types (a plurality of latest attribute types) as data related to attribute classification.

In step S34, the plurality of attribute types include, by way of example, an attribute type (attribute i) related to a time traffic volume change pattern for each time slot, an attribute type (attribute j) related to a daily traffic volume (unit/day) change pattern for each day of the week, and an average daily traffic volume per month. The attribute type (attribute k) for the pattern of change may be included.

In this case,'attribute i','attribute j', and'attribute k'may be referred to differently as the latest attribute information. According to this, any one of the latest first attribute type to the latest fourth attribute type may be assigned as the'attribute i'. As'attribute j', any one of the latest fifth attribute type to the latest seventh attribute type may be assigned. As'attribute k', any one of the latest eighth attribute type to the latest tenth attribute type may be assigned.

Thereafter, in step S35, the selection unit 12, the attribute type donated in step S32 (that is, attribute I, attribute J, attribute K, which are a plurality of donated past attribute types) and attribute type (plurality) given in step S34. By comparing the attribute i, attribute j, and attribute k), which are the latest attribute types, it is possible to check (determine) whether or not the attribute type given in step S34 is at least partially different from the donated attribute type. That is, the selection unit 12 is the attribute type (attribute I, attribute J, attribute K) donated in step S32 and the attribute type (attribute i, attribute j) given in step S34 corresponding to each of the attribute types contributed in step S32. , Attribute k) can be compared to each other to see if they match.

At this time, if it is confirmed that any one of the donated attribute types does not match the attribute type assigned in step S34 (that is, it is confirmed that any one of the donated attribute types is different from the latest attribute type), in step S36 The sorting unit 12 may recognize the first vehicle detector as a vehicle detector with suspected performance, and may first select the first vehicle detector as an inspection target. The sorting unit 12 may perform such a primary sorting process for each of the plurality of vehicle detectors, thereby first selecting an object to be inspected among the plurality of vehicle detectors.

In the above description, steps S31 to S36 may be further divided into additional steps or may be combined into fewer steps, depending on the embodiment of the present application. In addition, some steps may be omitted as necessary, or the order between steps may be changed.

2C is for explaining a process for selecting an inspection object (especially, a process for selecting a secondary inspection object) using data on attribute classification in the inspection object selection device 10 for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present invention. It is a drawing.

Referring to FIG. 2C, the selection unit 12 is a vehicle detector that is not first selected as an inspection object among a plurality of vehicle detectors (by using a traffic volume estimation model for unselected vehicle detectors as an example, the inspection object can be secondarily selected). Hereinafter, as an example, when the second vehicle detector among a plurality of vehicle detectors is an unselected vehicle detector, a process of selecting the second vehicle detector as an inspection target by the present apparatus 10 will be described as an example. same.

In step S41, the selection unit 12 may acquire a past measurement value and an evaluation target measurement value of the second vehicle detector in order to determine whether the second vehicle detector is an inspection target. Here, as a past measurement value, for example, accumulated data for (N-1) years after the second vehicle detector is installed may be used. As the measurement value to be evaluated, data accumulated in the N year to be compared (that is, the latest accumulated data collected and accumulated in year N) can be used.

Next, in step S42, the selection unit 12 compares the predicted value (estimated model calculated value) and the observed value (observed value in year N) predicted through the traffic volume estimation model, and the difference is set in a preset tolerance (error range). ) Can be checked. At this time, in step S42, the selection unit 12 may evaluate the estimated error for each of the time traffic volume, the daily traffic volume for each day of the week, and the average daily traffic volume for each month.

That is, the selection unit 12 may check whether a difference between the predicted value (model estimate value) and the observed value predicted through the traffic volume estimation model is out of the error range in relation to the data on the classification of attribute types by time slot. In addition, the selection unit 12 may check whether a difference between the predicted value and the observed value predicted through the traffic volume estimation model is out of an error range with respect to data on the classification of attribute types for each day of the week. In addition, the selection unit 12 may check whether a difference between the predicted value and the observed value predicted through the traffic volume estimation model is out of the error range with respect to the data on the classification of the attribute type by month.

In this case, if the difference is out of the error range in any of the three cases described above, in step S43, the selection unit 12 may secondarily select the third vehicle detector, which was an unselected vehicle detector, as an inspection target. The sorting unit 12 performs such a secondary sorting process for each of the unselected vehicle detectors, so that the object to be inspected among the unselected vehicle detectors can be secondarily selected.

Thereafter, the selection unit 12 may determine that the vehicle detector, which is the first selected inspection object, and the vehicle detector, which is the secondly selected inspection object, is an inspection object that is judged to be necessary for evaluation (inspection) because the performance is suspicious.

In the above description, steps S41 to S43 may be further divided into additional steps or may be combined into fewer steps, depending on the embodiment of the present application. In addition, some steps may be omitted as necessary, or the order between steps may be changed.

Hereinafter, a process for selecting an inspection object according to the first to third embodiments of the present application will be described in detail.

In the inspection object selection process according to the first embodiment of the present application, the selection unit 12 determines the inspection object through the similarity evaluation of the traffic volume change pattern for each time period (i.e., the similarity evaluation on the traffic volume change pattern for each time period with respect to daily traffic volume). Can be selected.

Specifically, the traffic volume change pattern using the road has a characteristic that is periodically repeated. Unless there are reasons for changes in road geometry and purpose of traffic, it is reasonable to assume that the pattern of changes in traffic volume by time has a pattern similar to that of the past. In other words, if there is no reason for changes in road geometry or change in traffic purpose, it is reasonable to assume that the current traffic volume change pattern by time zone has a similar pattern (shape, shape) to the past traffic volume change pattern by time zone. have. In addition, the change in the size of the traffic volume and speed (vehicle speed) shows a change of increase or decrease by an annual average increase rate, whereas the change pattern for each time period has a characteristic that is maintained, and a description thereof can be more easily understood with reference to FIG. 3. have.

Consider these points (i.e., the traffic volume change pattern by time slot has a pattern similar to the past, that is, the traffic volume change pattern by time zone acquired from a normal vehicle detector has a pattern similar to the annual average traffic volume change pattern per time zone for daily traffic volume). Thus, the selection unit 12 may select the inspection object based on the similarity evaluation of the traffic volume change pattern for each time period.

3 is a view showing an example of a pattern of changes in average annual traffic volume for each time period in the inspection target selection device 10 for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present application.

At this time, the solid line graph in FIG. 3 represents a graph of the annual average increase rate (ie, the annual average traffic volume change pattern for each time period) for the past two years (ie, N-2 year) for any one vehicle detector. In FIG. 3, the dotted line graph represents a graph of the annual average increase rate (ie, the annual average traffic volume change pattern for each time period) for the past year (ie, year N-1) for any one vehicle detector.

Referring to FIG. 3, the current traffic volume change pattern for each time zone (for example, a graph corresponding to a solid line graph) is a pattern similar to the past traffic volume change pattern for each time zone (for example, a graph corresponding to a dotted line graph) ( Shape, shape).

On the other hand, there may exist various types of roads such as roads used for commuting, roads used for tourism, etc. (that is, there are various types of roads). The percentage of traffic over time can be different. That is, the ratio of traffic volume according to time of a road used for commuting and a road used for tourism may be different from each other.

In consideration of these points (that is, the traffic volume change patterns obtained from vehicle detectors belonging to the same road type may have similar patterns to each other), the selection unit 12 is based on the similarity evaluation of the traffic volume change patterns by time zone. It is possible to perform the screening of one inspection object.

FIG. 4 shows two change patterns (i.e., a traffic volume change pattern by time zone, a traffic volume rate change by time zone) for the same point as an example in the inspection target selection device 10 for evaluating the performance of a vehicle detector according to an embodiment of the present application. A diagram showing an example of comparing similarity of pattern). That is, FIG. 4 shows an example of a traffic volume change pattern for each of two time periods acquired based on traffic data collected for a certain point (a specific same point) on a road from a vehicle detector.

In this case, among the two change patterns in FIG. 4, the change pattern corresponding to the dotted line graph shows the traffic pattern at peak times in the morning and afternoon. That is, the dotted line graph shows an example of a traffic volume change pattern for each time zone for daily traffic in which peak time zones (for example, peak time zones in the morning and afternoon) appear in two places. Meanwhile, among the two change patterns in FIG. 4, the change pattern corresponding to the solid line graph shows a traffic pattern that gradually increases in the afternoon time period without the peak time period. That is, the solid line graph shows an example of a traffic volume change pattern for each time zone with respect to the daily traffic volume, which appears to have no peak time zone. In the present application, the peak time may mean a time during which the vehicle occupies the highest occupancy of the road.

Referring to FIG. 4, when traffic volume observation is performed for the same point (same observation point) on the road through a vehicle detector, the level of traffic volume fluctuation is in a specific time period including the peak time (e.g., 07:00 to 18:00). ) Appears larger (i.e., differs more clearly) than in other time zones. In consideration of this point, the selection unit 12 selects the inspection target through similarity evaluation of the traffic volume change pattern for each time period, as in the description below, for example, some preset time periods (07:00-18) among the entire day time period. The correlation coefficient calculated using traffic data belonging to the city) can be considered as an index for evaluating similarity.

In the present apparatus 10, in order to perform the similarity evaluation of the traffic volume change pattern by time slot, as an example, the similarity evaluation index of the traffic volume change pattern may be set as follows.

The correlation coefficient is one of the indicators that quantify that there is a difference with the naked eye when each of the traffic volume data or speed data is plotted as traffic data according to the time zone to be compared.

At this time, if the traffic data considered when selecting the inspection target is traffic data (i.e., traffic volume related data), if the correlation coefficient is calculated using all data corresponding to 24 hours a day, it may be analyzed that there is no discrimination power. , In the present apparatus 10, it is possible to analyze using only data between 7 o'clock and 18 o'clock, which can best express the change in traffic volume. The apparatus 10 may apply a correlation coefficient as an index for evaluating the similarity of the traffic pattern graph (that is, as an index for evaluating the similarity between two patterns to be described later). In the present apparatus 10, the criterion for evaluating the similarity between two patterns (that is, a preset similarity tolerance range) may be set to 0.95 or more as an example, but is not limited thereto, and the numerical value may vary according to user input. Can be set.

According to the above, when the operation of the equipment (vehicle detector) that measures traffic volume and speed is one example, it is biased or inaccurate due to an error, the similarity of the aforementioned traffic patterns for each time zone (traffic volume change patterns for each time zone). And, the change in size as much as the annual average growth rate is not maintained. Therefore, when such a case occurs, it becomes possible to doubt the performance of the vehicle detector. In other words, the device 10 is suspected of deteriorating the performance of the vehicle detector in which such a case occurs (that is, a vehicle detector that does not have similar traffic patterns for each time slot or does not maintain a change in size as much as the average annual increase rate) and needs to be checked. It can be selected as the subject of inspection that is judged to be.

The process of selecting an inspection object according to the first embodiment of the present application will be described again as follows.

The evaluation reference data generated by the generation unit 11 includes data on an average annual traffic volume change pattern for each time period with respect to the daily traffic volume for any one vehicle detector (for example, a first vehicle detector) among a plurality of vehicle detectors. can do. Hereinafter, for convenience of explanation, data on the annual average traffic volume change pattern for each time period with respect to the daily traffic volume will be referred to as first evaluation reference data. In addition, hereinafter, for convenience of description, any one vehicle detector will be described as an example as the first vehicle detector (or vehicle detector 1).

When the data on the annual average traffic volume change pattern for each time period for daily traffic (i.e., the first evaluation standard data) is considered as the evaluation reference data, the selection unit 12 determines the annual average traffic volume change pattern for each time period (i.e., the first evaluation standard data). If the similarity between the pattern corresponding to the reference data) and the traffic volume change pattern for each time period acquired from any one of the plurality of vehicle detectors is evaluated and the similarity is out of the allowable range, any one vehicle detector (i.e., for each time slot) Any one vehicle detector in which the traffic volume change pattern has been acquired) can be selected as an inspection target.

Here, the preset similarity allowable range may be set to 0.95 (in other words, 95%) as an example, but is not limited thereto, and the range may be variously set and changed by the user.

At this time, the two patterns subject to similarity evaluation (i.e., the annual average traffic volume change pattern by time slot for daily traffic volume, which is the pattern corresponding to the first evaluation criterion data, and traffic volume change pattern by time slot) are among a plurality of vehicle detectors, for example. It may be data obtained from any one vehicle detector (for example, a first vehicle detector) (that is, data obtained from one and the same vehicle detector). However, this is only an example to aid the understanding of the present application, but is not limited thereto. As another example, each of the two patterns subject to similarity evaluation is data acquired by different vehicle detectors among a plurality of vehicle detectors (i.e., the average annual traffic volume per time period for daily traffic volume, which is a pattern corresponding to the first evaluation criterion data). The change pattern is data acquired from the first vehicle detector, which is one of the plurality of vehicle detectors, and the traffic volume change pattern by time is one vehicle detector different from the first vehicle detector among the plurality of vehicle detectors. Data obtained from a vehicle detector).

In addition, at this time, when the selection unit 12 evaluates the similarity between the two patterns, the similarity may be evaluated by a correlation coefficient, which is an index that quantifies the difference (difference between the two patterns) between two patterns to be compared.

In addition, the similarity between the two patterns may be evaluated in consideration of the increase/decrease rate of the traffic volume for each time zone obtained from any one vehicle detector (for example, the first vehicle detector) compared to the annual average traffic volume for each time zone.

In general, from year to year (yearly) traffic can increase (or decrease or remain unchanged). According to this, in the present application, the pattern similarity (pattern similarity) means that even if the traffic volume generally increases or decreases throughout the time zone, the shape (shape) of the pattern is a preset similarity allowance range (the similarity preset by the user is allowed. In the case where it appears to be maintained within the range) (for example, peak times substantially coincide with each other, etc.), it may mean that it is determined to be similar. In other words, as a result of evaluating the similarity between the two patterns, the selection unit 12 determines that the two patterns are similar if the difference between the shapes (shapes) of the two patterns falls within the preset similarity tolerance range, while the difference between the shapes of the two patterns If is out of the preset similarity tolerance range, it can be determined that the two patterns are not similar to each other.

For example, based on the drawing of FIG. 4 described above, the dotted line graph in FIG. 4 is an example of a pattern of changes in the average annual traffic volume for each time period for daily traffic in which peak times appear in two places (i.e., a pattern corresponding to the first evaluation reference data). ) Shows an example. At this time, when the overall pattern shape (shape) is different, such as a change in the peak time period with respect to the traffic volume change pattern obtained from any one vehicle detector (for example, the first vehicle detector) for each time period, the two The correlation coefficient, which is an index that quantifies the difference between patterns, may be calculated as a low value (eg, 0.01). In this case, the selection unit 12 may evaluate the degree of similarity between the two patterns by applying the correlation coefficient.

In other words, as a result of comparing (contrast) the two patterns (the shapes of the two patterns) with each other, as an example, in the pattern corresponding to the first evaluation reference data, two peak time zones appeared, and one vehicle detector (as an example). In the traffic volume change pattern by time zone acquired from the first vehicle detector), the peak time zone does not appear in the two places in which peak time zones appear in the pattern corresponding to the first evaluation criterion data, and in the corresponding part, or a preset time zone as another example. In the pattern corresponding to the first evaluation reference data for (as a specific time zone, for example, the time zone between 1:00 and 2:00 pm), the pattern corresponding to the first evaluation criterion data appeared as a rising curve, but any one vehicle detector (for example, the first vehicle detector) In the case where the shape of the pattern is generally different, such as when it is indicated by a descending curve, in the traffic volume change pattern obtained from the time period, the correlation coefficient, which is an index that quantifies the difference between the two patterns, may be calculated low.

In addition, when determining the degree of similarity, the selection unit 12 may consider a correlation coefficient calculated using traffic data belonging to a predetermined time zone among the entire time zone (24 hours) as an index for evaluating the similarity. In this case, some of the preset time zones may include a time zone in which the amount of change in traffic volume for each time zone is equal to or greater than a predetermined amount of change amount or a time zone in which the traffic volume per time zone corresponds to a predetermined amount of traffic or more. For example, some of the preset time zones are time zones including commuting time, and may mean a time zone between 07:00 and 18:00, for example.

As such, the selection unit 12 may select an inspection object from among a plurality of vehicle detectors by using data on a change pattern of an average annual traffic volume for each time of day as evaluation reference data.

In the inspection object selection process according to the second embodiment of the present application, the selection unit 12 may select the inspection object through verification (evaluation) of the increase or decrease in traffic volume. At this time, the selection unit 12 evaluates the identity of the increase/decrease rate of traffic volume for each partial cumulative time slot and the verification through a yearly average increase/decrease comparison with adjacent vehicle detectors (i.e., comparative verification with adjacent vehicle detectors) in relation to the verification (evaluation) of the traffic volume increase/decrease rate. (In other words, comparison of partial cumulative results at the same point in time) can be performed.

In this case, in the present application, as an example, the verification process by comparing the annual average increase/decrease rate with adjacent vehicle detectors is referred to as the first sub-inspection target selection process among the inspection target selection process according to the second embodiment, and evaluates the identity of the traffic volume increase/decrease rate by partial cumulative time The process will be referred to as a second sub-inspection object selection process in the inspection object selection process according to the second embodiment.

First, a description of the process of selecting an inspection object (ie, the first sub-inspection object selection process) by verification through a comparison of the annual average increase/decrease rate (that is, the annual average increase/decrease rate of traffic volume) with adjacent vehicle detectors is as follows.

In general, in order to grasp the traffic condition of the road, several (multiple) vehicle detectors are installed and operated at certain intervals (for example, 1km or 2km, etc.) on the road. At this time, depending on the road, a certain level of traffic volume may increase or decrease every year, or there may be no change in traffic volume.

In the case of an image-type vehicle detector based on image analysis, which is a type of vehicle detector, the shadow of a vehicle traveling on an adjacent lane may be counted or the shaking of a leaf of a roadside tree may be misrecognized as a vehicle. In addition, if the vehicle cannot be classified due to super proximity driving, it may cause an error in traffic volume counting. According to these misrecognitions and errors, there was a limit to obtaining accurate traffic data.

Therefore, taking into account that there is a high possibility that the increase or decrease in traffic volume varies greatly depending on the morning time, afternoon time, and night time, the device 10 is based on this focus on a method of selecting equipment (vehicle detectors) whose performance is suspected. I propose about it.

In this regard, if the traffic volume increase or decrease of an arbitrary road is investigated over several years, the traffic volume increase or decrease pattern of the road can be found, and it is considered that the increase rate is similar in the case of vehicle detectors installed and operated in close proximity to each other. It can be said to be reasonable.

In other words, for an arbitrary road (especially the same target road), if you look at the increase/decrease rate (i.e., average annual increase/decrease rate of traffic volume) for that road over several years, the pattern of the increase/decrease rate of traffic volume for that road (i.e., the annual average increase/decrease rate of traffic volume) Pattern). In addition, it is reasonable to consider that the traffic volume increase/decrease (annual average increase/decrease rate of traffic volume) corresponding to each of the vehicle detectors close to each other is similar to each other. Here, the average annual increase/decrease rate (in other words, the annual average increase/decrease rate of traffic volume) of the road on which the vehicle detector is installed may be calculated as an example, such as the annual average increase/decrease rate for N-2 years after the vehicle detector is installed. It can be calculated on the basis (through analysis of the collected traffic data).

In consideration of these points (that is, the annual average increase/decrease rate of traffic corresponding to each of the vehicle detectors adjacent to each other on the same target road may be similar to each other), the selection unit 12 provides an annual average of a specific vehicle detector and the adjacent vehicle detectors adjacent thereto. Inspection targets can be selected based on verification through comparison of the increase/decrease rate (that is, the annual average increase/decrease rate of traffic volume).

That is, in the process of selecting the inspection object according to the second embodiment of the present application, a process of selecting an inspection object using an annual average increase/decrease comparison with an adjacent vehicle detector will be described again as follows.

The evaluation reference data generated by the generation unit 11 may include data on an average annual increase/decrease rate of the traffic volume of each of the plurality of vehicle detectors. Hereinafter, for convenience of explanation, data on the annual average increase/decrease rate of traffic volume will be referred to as second evaluation reference data. In addition, in the present application, it may be referred to differently as an annual average increase/decrease rate of traffic volume and the like.

When data on the annual average increase/decrease rate of traffic volume (i.e., the second evaluation standard data) is considered as the evaluation standard data, the selection unit 12 If the similarity between the data on the annual average increase/decrease rate of traffic volume and the data on the average annual increase/decrease rate of traffic volume of the other vehicle detector (for example, the second vehicle detector) among the plurality of vehicle detectors is evaluated, and the degree of similarity is out of the allowable range, one of the two It may be determined that there is a possibility that the vehicle detector of (i.e., one vehicle detector and the other vehicle detector) may correspond to the inspection target. That is, the selection unit 12 may determine that one of the two vehicle detectors is an inspection target candidate that may correspond to the inspection target.

Here, the pre-set similarity allowable range to be considered may be set to 0.95 (in other words, 95%) as an example, but is not limited thereto, and the range may be variously set and changed by the user.

In addition, the other vehicle detector (i.e., the second vehicle detector) may be, for example, a vehicle detector immediately adjacent to any one vehicle detector (the first vehicle detector), but is not limited thereto, and the same target road Any one of the remaining vehicle detectors excluding any one of the plurality of vehicle detectors belonging to the (first vehicle detector) may be applied.

In addition, as an example, the similarity evaluation between the two second evaluation criteria data in the present application (i.e., between data on the annual average increase/decrease rate of the traffic volume of one of the plurality of vehicle detectors and the data on the annual average increase/decrease rate of the traffic volume of the other vehicle detector. Similarity evaluation) can be performed by comparing the similarity (similarity) of the two scalar values based on a scalar value such as'The annual average increase/decrease rate of traffic volume in 2020 is 10.3%' as an example of the average annual increase/decrease rate of the traffic volume. As another example, the similarity evaluation between the two second evaluation criteria data is based on the pattern of the average annual increase/decrease rate of traffic volume by time slot (or the annual average increase/decrease rate of traffic volume by time slot), and compare the similarity between the two patterns (i.e., the similarity between the shapes of the patterns). Can be achieved by comparing). At this time, the similarity evaluation between the two second evaluation criteria data according to another example uses a correlation coefficient, etc., similar to the similarity comparison between the two patterns in the process of selecting an inspection object according to the first embodiment of the present application described above. It can be done by doing. As another example, the similarity evaluation between the two second evaluation criteria data is based on the increase/decrease rate of the cumulative traffic volume (i.e., the traffic volume increase/decrease rate for each partial cumulative time zone), which partially accumulates the traffic volume for some time periods, as described later, This can be achieved by comparing the similarity of the increase/decrease rate value (scalar value) of the two cumulative traffic volumes.

In addition, as described above, the selection unit 12 detects one of the two vehicle detectors (i.e., one vehicle detector and the other vehicle detector) through the similarity evaluation (comparison) between the data on the annual average increase/decrease rate of the two traffic volumes. It can be determined that there is a possibility that it may correspond to the inspection target (ie, it is determined as a candidate for inspection).

In this case, the selection unit 12 may determine that one of the two vehicle detectors is a candidate to be inspected through 1:1 contrast. That is, the selection unit 12 may determine that one of the vehicle detectors is a candidate to be inspected by performing 1:1 contrast among a plurality of vehicle detectors (that is, performing 1:1 contrast of the vehicle detectors). . Here, the selection unit 12 may finally select a vehicle detection target, which is an inspection target, from among the candidates to be checked by combining a plurality of vehicle detectors in various ways (by combining them in various ways).

In other words, the selection unit 12 evaluates (comparison) the similarity between the data on the annual average increase/decrease rate of the traffic volume of one vehicle detector and the data on the average annual increase/decrease rate of the traffic volume of the other vehicle detector. It can be repeated for 1:1 combinations (combinations). According to this, the selection unit 12 can select (identify, recognize) candidates to be inspected among the two vehicle detectors through a 1:1 contrast first among the plurality of vehicle detectors. By repeatedly performing various 1:1 combinations, it is possible to finally select the inspection object among the candidates to be inspected.

For example, as a plurality of vehicle detectors, there are a first vehicle detector, a second vehicle detector, and a third vehicle detector, wherein the first vehicle detector is a vehicle detector that is suspected of deteriorating performance and is determined to require inspection, and Suppose that the 2nd vehicle detector and the 3rd vehicle detector are not suspected of deterioration in performance and therefore are judged not to need inspection. In this case, the first vehicle detector may be referred to differently as an abnormal vehicle detector, and the second vehicle detector and the third vehicle detector may be referred to differently as a normal vehicle detector.

In this case, the selection unit 12 may, for example, evaluate the degree of similarity between the data on the average annual increase/decrease rate of two traffic volumes corresponding to each of the first vehicle detector and the second vehicle detector. As a result, the first vehicle detector and the second vehicle detector Any one of the vehicle detectors may be selected as an inspection target candidate (in other words, any one of the first vehicle detector and the second vehicle detector is identified as an abnormal vehicle detector). Next, the selection unit 12 may, for example, evaluate the degree of similarity between the data on the annual average increase/decrease rate of two traffic volumes corresponding to each of the second vehicle detector and the third vehicle detector, and as a result, the second vehicle detector and the third vehicle detector Both can be selected (identified) as normal vehicle detectors. Next, the selection unit 12 may, for example, evaluate the similarity between the data on the average annual increase/decrease rate of two traffic volumes corresponding to each of the third vehicle detector and the first vehicle detector. As a result, the third vehicle detector and the first vehicle detector Any one of the vehicle detectors may be selected as an inspection target candidate (in other words, any one of the third vehicle detector and the first vehicle detector is identified as an abnormal vehicle detector). In this way, by repeatedly performing similarity evaluation for various 1:1 combinations between the first vehicle detector and the third vehicle detector, the selection unit 12 based on the result, the first vehicle detector among the first to third vehicle detectors. Is finally determined to be an abnormal vehicle detector, and the corresponding first vehicle detector can be selected as an inspection target.

Next, a description of the process of selecting an inspection object through the uniformity evaluation of the traffic volume increase/decrease rate for each partial cumulative time slot (that is, the second sub-inspection object selection process) is as follows.

For example, as a method of checking the increase/decrease rate (annual average increase/decrease rate of traffic volume) based on the possibility of error of the video-type vehicle detector, the accuracy of the vehicle detector is when the increase rate of traffic volume is abnormally high or low only during the time zone affected by a specific shadow or only in the night time zone. It can be said that it is possible to doubt. Therefore, it is possible to evaluate the annual average increase/decrease rate of traffic volume by grouping several times in consideration of the influence of the sun's location and street tree shadows during the day (i.e., considering the influence of the sun's position during the day and street tree shadows, etc.) Through an evaluation method that combines time and compares the annual average increase/decrease rate of traffic volume, it will be possible to select vehicle detectors whose performance is suspected of deteriorating or whose accuracy is suspicious), and this device 10 selects inspection targets using this comparative evaluation method. I want to.

Table 2 below shows, as an example, values of the average annual increase/decrease in traffic volume generated for each group of a plurality of predetermined time zones for each of a plurality of vehicle detectors including the first vehicle detector (vehicle detector 1) to the third vehicle detector (vehicle detector 3). An example of (element) is schematically shown.

[Table 2]

Referring to Table 2, in the case of a normal vehicle detector whose performance is not suspected of deterioration, the values of the annual average increase/decrease rate of traffic volume for each group of a plurality of time zones generated for the corresponding normal vehicle detector (i.e., R _morning , R _afternoon , R _night ) These should have similar values.

In other words, if the values of the average annual increase/decrease rate of traffic volume for each group in a plurality of time zones corresponding to the same vehicle detector (for example, the first vehicle detector) _{have similar values (i.e., R morning} , R _afternoon , R _night ), the selection unit (12) may identify the vehicle detector (ie, the first vehicle detector) as a normal vehicle detector.

On the other hand, if the values of the average annual increase/decrease rate of traffic volume for each group in a plurality of time zones corresponding to the same vehicle detector (for example, the first vehicle detector) (i.e., R _morning , R _afternoon , R _night ) do not have similar values (i.e., plural If at least one of the values of the average annual increase/decrease rate of traffic volume for each group of time periods is different from the remaining values), the selection unit 12 uses the vehicle detector (i.e., the first vehicle detector) for an abnormality in which performance deterioration is suspected (abnormal). Vehicle detector), and as a result, the corresponding vehicle detector (ie, the first vehicle detector) can be selected as an inspection target.

At this time, the selection unit 12 is, for example, the value of the average annual increase/decrease rate of traffic volume for each group in a plurality of time zones corresponding to the same vehicle detector (for example, the first vehicle detector) (i.e., R _morning , R _afternoon , R _night ). The criterion that is determined not to have a value may be determined to be not similar when a difference between two values among _{three values (ie, R morning} , R _afternoon , and R _{night) is out of a preset allowable range.} Here, the preset allowable range may be set to a range of ± 10% as an example, but this is only an example to aid understanding of the present application, and is not limited thereto, and the numerical value may be set in various ways. have. In other words, the selection unit 12 may set a criterion for determining the similarity between the annual average increase/decrease rate of traffic volume for each group in a plurality of time zones as an initial value of ± 10% as an example.

That is, the process of selecting the inspection object through the uniformity evaluation of the increase/decrease rate of the traffic volume for each partial cumulative time period in the inspection object selection process according to the second embodiment of the present application will be described again as follows.

The generator 11 may generate an average annual increase/decrease rate of traffic volume for each of the plurality of vehicle detectors for each of a plurality of predetermined time zone groups, taking into account the influence of the shadow due to the movement of the sun during the day and the absence of the sun at night. Thereafter, the selection unit 12 determines the difference between the average annual increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones generated for any one vehicle detector (for example, the first vehicle detector) among the plurality of vehicle detectors. , ± 10%), any one vehicle detector (for example, the first vehicle detector) can be selected as an inspection target.

Here, a plurality of predetermined time zone groups may be classified (set) as follows, as an example. The plurality of predetermined time zone groups may be classified into a first time zone, morning, a second time zone, afternoon, and a third time zone, night. In this case, the morning refers to a time zone after sunrise among the time zones in which the sun exists, and may exemplarily mean a zone between 7 o'clock and 12 o'clock. The afternoon refers to a time zone before sunset among the time zones in which the sun exists, and may, for example, mean a zone between 13:00 and 17:00. Nighttime refers to a time zone in which the sun does not exist (ie, a time zone in which the sun does not exist) as the sun sets, and may, for example, mean a zone between 19 o'clock and 21 o'clock.

As another example, a plurality of predetermined time zone groups may be classified (set) as follows. According to another example of the present application, a plurality of predetermined time zone groups may be set in consideration of changing a direction of a shadow based on a road (same target road) on which traffic data is acquired by a vehicle detector. That is, a plurality of predetermined time zone groups may be divided into a first time zone, a second time zone, and a third time zone. At this time, the first time zone is the corresponding time zone before the direction of the shadow changes when the road is based on the time zone in which the sun exists (for example, from the time the sun rises, the corresponding sun light is transmitted by the vehicle detector. It may mean a time zone until the sun is located at a position that illuminates the upper side of the vehicle on the road in which the acquisition is made). The second time zone is the time zone during which the sun exists, after the direction of the shadow is changed (e.g., when the sun is located at a position where the sun illuminates the top of the vehicle on the road, the shadow of the vehicle is not generated by the sun). It may mean a time zone until until). The third time zone may mean a time zone in which the sun does not exist because the sun sets.

In addition, when the difference between the average annual increase/decrease rate of the traffic volume with the adjacent vehicle detector is out of a preset range (i.e., the traffic volume between one vehicle detector among the plurality of vehicle detectors and another vehicle detector adjacent thereto. If the difference in the average annual increase/decrease is out of the preset range), the possibility that one of the vehicle detectors (i.e., one vehicle detector) to be compared with the adjacent vehicle detector (i.e., one vehicle detector) will fall under the inspection target. It can be judged (recognized, identified) as a candidate to be inspected. Here, the preset range may be set to a range of ± 5% as an example, but this is only an example for aiding understanding of the present application, and is not limited thereto, and the numerical value may be set in various ways.

More specifically, for example, the selection unit 12 may increase/decrease the average annual increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones generated for any one vehicle detector among a plurality of vehicle detectors (for example, R _AM , R of the first vehicle detector). Any one of (for example, R _morning of the first vehicle detector) and one of the annual average increase/decrease rate of traffic volume for each group in a plurality of predetermined time zones generated for the other vehicle detector among the plurality of vehicle detectors _{(for example, afternoon} , R _night) For example, it is determined whether the difference between the second vehicle detector R in the _morning or R _afternoon or R at _night is out of a preset range (for example, ± 5%), and if it is out of the preset range, one of the vehicle detectors ( 1 vehicle detector) and the other vehicle detector (second vehicle detector) can be identified as candidates for inspection. At this time, the description of the process of finally selecting the inspection target among candidates for inspection can be understood similarly to the process of selecting the final inspection target by repeatedly performing various 1:1 combinations for the plurality of vehicle detectors described above. , Hereinafter, a detailed description will be omitted.

As described above, the selection unit 12 may select an object to be inspected from among the plurality of vehicle detectors by using data on the average annual increase/decrease rate of the traffic volume of each of the plurality of vehicle detectors as evaluation reference data. In particular, the selection unit 12, as an example, as described above, in using data on the annual average increase/decrease rate of the traffic volume of each of the plurality of vehicle detectors, the verification through a comparison of the annual average increase/decrease rate with the adjacent vehicle detectors (ie, with adjacent vehicle detectors). The inspection target can be selected by using (considering) at least one of two methods, including the comparison verification) method and the uniformity evaluation (i.e., comparison of the partial cumulative results of the same point time period) of traffic volume increase/decrease by partial cumulative time zone.

In the inspection object selection process according to the third embodiment of the present application, the selection unit 12 may select the inspection object through the application of a traffic volume estimation model for an adjacent time zone.

Specifically, the selection unit 12 may estimate traffic volume data in the adjacent time zone by using past data (traffic data acquired in the past) for the same vehicle detector (for example, a first vehicle detector).

For example, the traffic volume estimate for each point in n time zones (that is, the traffic volume estimate for each observation point in which traffic data is acquired by the first vehicle detector) may be expressed as Equation 1 below. In other words, the selection unit 12 may obtain an estimated traffic volume (predicted value) using Equation 1 below, as an example. However, this is only an example to help the understanding of the present application, and is not limited thereto. As another example, the selection unit 12 calculates the traffic volume estimate (predicted value) using the deep learning model previously learned as described above. Can be obtained.

[Equation 1]

는 n 시간대의 교통량 예측치(대/단위시간)를 나타내고,

는 n-1 시간대의 모형계수를 나타내며,

는 n-1 시간대의 상수를 나타낸다.here,

Represents the traffic volume predicted value (unit/unit time) in n time zones,

Represents the model coefficient in the n-1 time zone,

Represents the constant of the n-1 time zone.

In this case, the value of the model coefficient for each time period and the constant value for each time period considered in Equation 1 may be exemplarily applied to the values shown in Table 3 below. In this case, the values (model coefficients or constants) shown in Table 3 may be calculated based on past data (traffic data acquired in the past) collected from the same vehicle detector (for example, a first vehicle detector) as an example. .

[Equation 2]

)(즉, 제1 차량검지기에 대한 n시간대의 교통량 예측치)와 평가 대상이 되는 차량검지기에서 수집된 교통량 관측치(

)(일예로, 제1 차량검지기로부터 수집된 현재의 n시간대의 교통량 측정치)를 비교하는 평가를 수행할 수 있다.The sorting unit 12 may select the object to be inspected using Equations 1 and 2 above. Specifically, the selection unit 12 predicts the traffic volume in n time zones (

) (I.e., the traffic volume predicted value for the n-hour period for the first vehicle detector) and the traffic volume observed value collected from the vehicle detector to be evaluated (

) (For example, a current measurement of traffic volume in n time zones collected from the first vehicle detector) may be evaluated.

At this time, the selection unit 12 suspects the performance of the vehicle detector (for example, the first vehicle detector) if the difference (difference between the two values) between the traffic volume predicted value and the collected traffic volume observation value in the n-hour period is large. Can be selected as the subject of inspection that is judged to be necessary due to the suspicion of performance degradation. In this case, the criterion for determining that the difference between the two values is large may be determined as, for example, whether the difference between the two values is out of a preset trust tolerance range. Here, the preset confidence tolerance range may be set to a range of ± 95% as an example, but this is only an example to aid understanding of the present application, and is not limited thereto, and the numerical value may be set in various ways. I can. The preset confidence allowance range may be referred to differently as a confidence interval.

That is, if the difference between the two values is large, the selection unit 12 may determine that the performance of the vehicle detector is suspicious and select the vehicle detector as an inspection target. In this case, as a criterion for the difference between the two values, an example Can be set to ± 95% of the confidence interval.

Exemplarily, the selection unit 12 calculates the difference (difference between the two values) (d _n ) between the traffic volume predicted value in n time zones and the collected traffic volume observation value for each time zone as shown in Equation 2 above, and between the two values calculated for each time zone. A difference value having a maximum value among the _{difference values d 1} , …, d _{24 may be obtained.} In this case, the selection unit 12 determines whether the difference value having the maximum value is out of the preset confidence allowable range, and as a result, if it is out of the preset confidence allowable range, selects the corresponding vehicle detector as an inspection object.

As an example, the apparatus 10 may create (generate) a traffic volume estimation model for each adjacent time zone for each installation point of the vehicle detector. For example, in relation to the adjacent time zone traffic volume estimation model, if only some of the values calculated due to the constraints of the ground are presented, it may be as shown in Table 3 below. That is, Table 3 below shows an example of a traffic volume estimation model in the time zone immediately after each point of the vehicle detector.

[Table 3]

That is, in the process of selecting a subject for inspection according to the third embodiment of the present application, a process for selecting a subject for inspection using a model for estimating a traffic volume for adjacent time will be described again.

The evaluation reference data generated by the generation unit 11 may include data on an estimate of the traffic volume in n time zones estimated using a traffic volume estimation model in the adjacent time zone with respect to any one of the plurality of vehicle detectors. have. Hereinafter, for convenience of explanation, data on the estimated traffic volume in the n-hour period will be referred to as third evaluation reference data.

When the data on the estimated traffic volume in n time zones (that is, the third assessment standard data) is considered as the assessment reference data, the selection unit 12 When the observations of traffic volume in n-hour periods collected by one vehicle detector are compared and out of the preset confidence tolerance range (for example, ± 95%), any one vehicle detector may be selected as an inspection target.

At this time, the preset traffic volume estimation model for adjacent time zones may be set as a relational expression between model coefficients and constants for n-1 time zones, traffic volume observations for n time zones, and traffic volume estimates for n time zones. Can be set.

According to the above, the selection unit 12 may select an inspection target using a traffic volume estimation model for an adjacent time zone.

That is, the device 10 is based on at least one of the above-described inspection target selection process according to the first embodiment of the present application to the inspection target selection process according to the third embodiment of the present application, based on the plurality of Among the vehicle detectors of, the vehicle detectors judged to be in need of inspection (or performance evaluation) can be selected as inspection targets. Here, the inspection target selection process according to the second embodiment of the present application may include a first sub-inspection target selection process and a second sub-inspection target selection process as described above.

This is only one example to aid the understanding of the present application, but is not limited thereto. As another example, the apparatus 10 may sequentially perform each of the above-described inspection object selection process according to the first embodiment of the present application to the inspection object selection process according to the third embodiment of the present application, and illustratively the present application A vehicle detector that satisfies all conditions for each of the inspection target selection process according to the first embodiment of the present invention to the inspection target selection process according to the third embodiment of the present application may be selected as the inspection target.

The present apparatus 10 may select only equipment (vehicle detectors) whose performance is suspicious among a large number of vehicle detectors, and perform performance evaluation on the selected inspection target. This device 10 may have an economical and temporal advantage compared to evaluating a vast amount of vehicle detectors (all vehicle detectors) one by one.

In addition, the device 10 is a problem related to the performance evaluation of existing vehicle detectors, in other words, problems with the conventional performance evaluation technology for vehicle detectors (that is, the basis for setting the regular evaluation criteria of ITS equipment is insufficient, and some local governments In this case, regular evaluation is not carried out for reasons of budget, etc., and performance evaluation for the remaining lanes is insufficient due to the conventional evaluation of only one lane, and the problem of not being able to install performance evaluation standard equipment depending on the site conditions) is effectively solved. As can be done, it is possible to effectively improve (increase) the reliability and efficiency of the system (intelligent system) compared to the prior art.

If the vehicle detector is regarded as a kind of measurement equipment, from the perspective of managing the system (intelligent system), the vehicle detector must maintain a certain level of accuracy to ensure that the data measured through the vehicle detector (i.e., traffic data) can be trusted. There will be. Accordingly, the device 10 selectively selects only vehicle detectors that are determined to require inspection due to suspected performance degradation among the plurality of vehicle detectors, and performs a performance test on the selected vehicle detectors (selected inspection targets). By performing, it is possible to provide so that all of the plurality of vehicle detectors can maintain a more certain level of accuracy at all times.

This device 10 is more accurate traffic data (for example, traffic volume, speed, occupancy time, etc.) from a plurality of vehicle detectors associated with the device 10 (i.e., a plurality of vehicle detectors evaluated and managed by the device). Collected data) can be collected, and traffic information generated based on this (for example, section travel time, moving speed, congestion status, traffic accident occurrence, etc.) can be provided. Since the present apparatus 10 enables accurate collection of traffic data, traffic information produced on the basis thereof may also be provided so that more accurate traffic information can be generated without errors. The present apparatus 10 can generate and provide accurate traffic data and accurate traffic information, and thus, it is possible to provide a more reliable traffic information service.

In addition, in describing the apparatus 10 herein, the terms of an annual average traffic volume and a traffic volume annual average may be used interchangeably.

This device 10 is a performance among multiple vehicle detectors in order to solve the existing difficulties that are difficult to regularly manually inspect (evaluate) numerous vehicle detectors installed and operated at intervals of about 2.0 km on general national roads and 1 km on highways. A technology to perform performance evaluation by selectively selecting vehicle detectors suspected of deterioration is proposed.

Focusing on the fact that the traffic volume change pattern using the road has a characteristic that is periodically repeated, the device 10 analyzes the traffic volume change pattern for each time period using a similarity evaluation index called a correlation coefficient. The vehicle detector can be selected as a vehicle detector suspected of deteriorating performance (ie, a vehicle detector subject to inspection). In addition, the present apparatus 10 introduces a traffic volume estimation model for selection of such inspection targets.

The device 10 can be said to be a performance evaluation (performance maintenance) automation technology that selects and evaluates only equipment whose performance is suspect among many vehicle detectors. This device 10 can be effectively applied to a wider variety of fields, considering that it is essential to predict a traffic volume with higher reliability or accuracy in the introduction of an autonomous vehicle, which is the latest trend.

Hereinafter, based on the details described above, the operation flow of the present application will be briefly described.

5 is a diagram schematically illustrating a process of selecting an inspection object by the inspection object selection apparatus 10 for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure.

In this case, the process of selecting the inspection object shown in FIG. 5 may be performed by the apparatus 10 described above. Therefore, even if the contents are omitted below, the contents described with respect to the device 10 may be equally applied to the description of the process of selecting the inspection object.

Referring to FIG. 5, in step S11, the apparatus 10 may select an inspection object using an evaluation of the similarity of a traffic volume change pattern for each time period with respect to the daily traffic volume. In this case, step S11 refers to the process of selecting an inspection object according to the first embodiment of the present application described above, and a detailed description thereof will be omitted.

Next, in step S12, the apparatus 10 may select an inspection target using the evaluation of the increase/decrease rate of the traffic volume. At this time, in step S12, the apparatus 10 may select an inspection object through comparison with an adjacent vehicle detector as an example, and, as another example, select an inspection object through a comparison of partial cumulative results at the same point in time. have.

In this case, step S12 refers to the process of selecting the inspection target according to the second embodiment of the present application, and the comparison process with the adjacent vehicle detector refers to the process of selecting the first sub-inspection target described above, and a partial cumulative result of the same point in time. The comparison process refers to the above-described second sub-inspection target selection process, and a detailed description thereof will be omitted.

Next, in step S13, the apparatus 10 may select an inspection object using an evaluation of a traffic volume estimation model for an adjacent time zone (a criterion judged as a normal vehicle detector should be within ±95%). In this case, step S13 refers to the process of selecting an inspection object according to the third embodiment of the present application, and a detailed description thereof will be omitted.

Next, in step S14, the apparatus 10 may select, among a plurality of vehicle detectors, equipment (vehicle detectors) whose performance of traffic volume accuracy is suspected as an inspection target through the process of steps S11 to S13.

In the above description, steps S11 to S14 may be further divided into additional steps or may be combined into fewer steps, depending on the embodiment of the present application. In addition, some steps may be omitted as necessary, or the order between steps may be changed.

6 is a flowchart illustrating an operation of a method for selecting an inspection object for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure.

The method of selecting an inspection object for evaluating the performance of the vehicle detector shown in FIG. 6 may be performed by the apparatus 10 described above. Therefore, even if omitted below, the description of the device 10 may be equally applied to the description of a method for selecting an inspection object for evaluating the performance of the vehicle detector.

Referring to FIG. 6, in step S21, the generation unit 11 may generate evaluation reference data, which is a criterion for performance evaluation, based on traffic data collected from at least one of a plurality of vehicle detectors.

In this case, the plurality of vehicle detectors may be vehicle detectors included in the same target road.

Here, the same target road may refer to a road having the same road name. However, it is not limited thereto, and it is desirable to broadly understand the same target road as a concept encompassing interconnected roads so that vehicle driving can be continuously performed even if the road names are the same as well as different road names. In other words, the same target road may mean a road through which a vehicle having a degree of similarity of a predetermined or more can be driven by being interconnected with each other.

Next, in step S22, the selection unit 12 may select, among the plurality of vehicle detectors, a vehicle detector that is determined to require inspection, as an inspection target, using the evaluation reference data generated in step S21.

In addition, the evaluation reference data generated in step S21 may include data on a change pattern of the average annual traffic volume for each time period for the daily traffic volume with respect to any one of the plurality of vehicle detectors.

In this case, in step S22, the selection unit evaluates the similarity between the annual average traffic volume change pattern for each time slot and the traffic volume change pattern for each time zone acquired from any one of the plurality of vehicle detectors, and when the similarity is out of the allowable range, any one The vehicle detector of can be selected as an inspection target.

In this case, the degree of similarity may be evaluated by a correlation coefficient, which is an index that quantifies the difference between the two patterns to be compared.

In addition, the similarity between the two patterns may be evaluated in consideration of the increase/decrease rate of the traffic volume for each time zone obtained from any one vehicle detector compared to the annual average traffic volume for each time zone.

In addition, in step S22, when determining the degree of similarity, the selection unit may consider a correlation coefficient calculated using traffic data belonging to a predetermined time period among the entire day as a similarity evaluation index. In this case, some of the preset time zones may include a time zone in which the traffic volume fluctuation per time zone corresponds to a predetermined fluctuation amount or more, or a time zone in which the traffic volume per time zone corresponds to a predetermined traffic volume or more.

In addition, the evaluation reference data generated in step S21 may include data on an average annual increase/decrease rate of the traffic volume of each of the plurality of vehicle detectors.

At this time, in step S22, the selection unit evaluates the similarity between the data on the annual average increase/decrease rate of the traffic volume of one of the plurality of vehicle detectors and the data on the average annual increase/decrease rate of the traffic volume of the other vehicle detector among the plurality of vehicle detectors. If the degree of similarity is out of the allowable range, it can be determined that there is a possibility that one of the two vehicle detectors may be subject to inspection.

In addition, in step S21, the generation unit may generate an annual average increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones in consideration of the effect of the shadow due to the movement of the sun during the daytime and the absence of the sun at night.

At this time, in step S22, the selection unit checks any one vehicle detector when the difference between the average annual increase/decrease rate of the traffic volume for each group of a plurality of predetermined time zones generated for any one vehicle detector among the plurality of vehicle detectors is out of a preset allowable range. Can be selected as a target.

In addition, the evaluation reference data generated in step S21 may include data on an estimate of the traffic volume of n-time zones estimated using a traffic volume estimation model for the adjacent time zone with respect to any one of the plurality of vehicle detectors.

At this time, in step S22, the selection unit compares the traffic volume estimate in n time zones with the traffic volume observation values in n time zones collected by any one vehicle detector, and selects any one vehicle detector as an inspection target if it is out of the preset confidence tolerance range. I can.

In the above description, steps S21 and S22 may be further divided into additional steps or may be combined into fewer steps, depending on the embodiment of the present application. In addition, some steps may be omitted as necessary, or the order between steps may be changed.

The method of selecting an inspection target for evaluating the performance of a vehicle detector according to an exemplary embodiment of the present disclosure may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those produced 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 described above may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

In addition, the above-described method of selecting an inspection object for evaluating the performance of the vehicle detector may be implemented in the form of a computer program or application executed by a computer stored in a recording medium.

The foregoing description of the present application is for illustrative purposes only, and those of ordinary skill in the art to which the present application pertains will be able to understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and are not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present application is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present application.

10: Inspection target selection device for evaluating the performance of vehicle detectors
11: generation
12: sorting unit

Claims (18)

As a method of selecting inspection targets for evaluating the performance of vehicle detectors,
(a) generating evaluation reference data serving as a criterion for performance evaluation based on traffic data collected from at least one of a plurality of vehicle detectors included in the same target road; And
(b) selecting a vehicle detector determined to need inspection among the plurality of vehicle detectors as an inspection target using the evaluation criteria data,
Including,
In the step (a), the evaluation criterion data includes data on attribute classification given to any one of the plurality of vehicle detectors based on a past measurement value of any one of the plurality of vehicle detectors,
The above attribute classification is a first attribute type, which is a road type with a peak time in the morning, a second attribute type, which is a road type with a peak time in the afternoon, a third attribute type that is a road type with morning and afternoon peak hours, and a road without peak hours. Includes attribute type classification by time period including a fourth attribute type that is a type,
In the step (b), when the attribute type assigned based on the evaluation target measurement value obtained from the one vehicle detector is at least partially different from the attribute type donated to the one vehicle detector, the Selecting one vehicle detector as the inspection target,
The step (b),
Firstly selects an inspection target among the plurality of vehicle detectors based on the attribute classification,
Targeting the unselected vehicle detectors among the plurality of vehicle detectors that are not first selected, the inspection object is secondarily selected based on a traffic volume estimation model among the unselected vehicle detectors to determine the inspection object,
At the time of the secondary selection, between the traffic volume estimation value calculated through the traffic volume estimation model based on the accumulated data of the unselected vehicle detector among the unselected vehicle detectors and the traffic volume observation value collected from the unselected vehicle detector. A method for selecting an inspection target for evaluating the performance of a vehicle detector, which is to perform a secondary selection in consideration of the difference. delete delete The method of claim 1,
In the attribute classification, the average daily traffic volume on weekdays and the average daily traffic on weekends are compared with each other, and the fifth attribute type is a road type in which one of the two average daily traffic volumes is larger, and a road type in which one of the two average daily traffic volumes is larger. 6 The method of selecting an inspection target for evaluating the performance of a vehicle detector, further comprising classifying attribute types for each day of the week including the type of road 7 and the type of road with the same road type with the same average daily traffic. The method of claim 1,
In the attribute classification, the average daily traffic volume in some months of 12 months a year and the average daily traffic volume in the remaining months excluding the some months out of 12 months of a year are compared with each other, and one of the two average daily traffic volumes is the eighth type of road. The vehicle detector further includes a monthly attribute type classification including an attribute type, a ninth road type that is a road type in which the other of the two average daily traffic volumes is a large road type, and a 10 road type in which the two average daily traffic amounts are the same road type. How to select inspection targets to evaluate the performance of The method of claim 1,
The above past measurements correspond to past accumulated data accumulated over (N-1) years,
The evaluation target measurement value corresponds to the recent accumulated data accumulated for the last year excluding the period of (N-1). delete The method of claim 1,
The evaluation criterion data includes data on an average annual traffic volume change pattern for each time slot with respect to the daily traffic volume, with respect to any one of the plurality of vehicle detectors,
The step (b) may include evaluating the similarity between the average annual traffic volume change pattern for each time slot and the traffic volume change pattern for each time zone acquired from any one of the plurality of vehicle detectors, and when the similarity is out of the allowable range. Selecting one vehicle detector as the inspection target,
The similarity is evaluated by a correlation coefficient that is an index that quantifies the difference between two patterns to be compared. The method of claim 8,
The similarity between the two patterns is evaluated by considering the increase/decrease rate of the traffic volume for each time period obtained from the any one vehicle detector compared to the annual average traffic volume for each time period. The method of claim 8,
In the step (b), when determining the degree of similarity, a correlation coefficient calculated using traffic data belonging to a predetermined time period among the entire day time period is considered as a similarity evaluation index,
Some of the preset time zones include a time zone in which the traffic volume fluctuation per time zone corresponds to a predetermined fluctuation amount or more or a time zone in which the traffic volume per time zone corresponds to a predetermined traffic volume or more. . The method of claim 1,
The evaluation reference data includes data on an average annual increase/decrease rate of traffic volume of each of the plurality of vehicle detectors,
In the step (b), the similarity between the data on the average annual increase/decrease rate of the traffic volume of the vehicle detector among the plurality of vehicle detectors and the data on the average annual increase/decrease rate of the traffic volume of the other vehicle detector among the plurality of vehicle detectors is evaluated in advance. In the case where the set similarity is out of the allowable range, it is determined that one of the two vehicle detectors is likely to correspond to the inspection object. The method of claim 11,
In the step (a), the average annual increase/decrease rate of the traffic volume is generated for each group of a plurality of predetermined time zones in consideration of the effect of the shadow due to the movement of the sun during the day and the absence of the sun at night,
In the step (b), when the difference between the average annual increase/decrease rate of traffic volume for each group of a plurality of predetermined time zones generated for any one vehicle detector among the plurality of vehicle detectors is out of a preset allowable range, the one vehicle detector is performed. The method of selecting an inspection object for evaluating the performance of a vehicle detector, which is to be selected as the inspection object. The method of claim 1,
The evaluation criterion data includes data on an estimate of traffic volume in n time zones estimated using a traffic volume estimation model for adjacent time zones with respect to any one of the plurality of vehicle detectors,
In the step (b), when the traffic volume estimate in the n time zone is compared with the traffic volume observation value in the n-time zone collected by the any one vehicle detector, if it is out of a preset confidence tolerance range, the one vehicle detector is subjected to the inspection. The method of selecting an inspection target for evaluating the performance of a vehicle detector. As an inspection target selection device for evaluating the performance of vehicle detectors,
A generator configured to generate evaluation reference data serving as a criterion for performance evaluation based on traffic data collected from at least one of a plurality of vehicle detectors included in the same target road; And
A selection unit that selects vehicle detectors determined to need inspection among the plurality of vehicle detectors as inspection targets using the evaluation criterion data,
Including,
The evaluation criterion data includes data on attribute classification given to any one of the plurality of vehicle detectors based on a past measurement value of any one of the plurality of vehicle detectors,
The above attribute classification is a first attribute type, which is a road type with a peak time in the morning, a second attribute type, which is a road type with a peak time in the afternoon, a third attribute type that is a road type with morning and afternoon peak hours, and a road without peak hours. Includes attribute type classification by time period including a fourth attribute type that is a type,
The selection unit, when the attribute type assigned based on the evaluation target measurement value obtained from the one vehicle detector is at least partially different from the attribute type donated to the one vehicle detector, the one vehicle Select the detector as the above inspection object,
The selection unit,
Firstly selects an inspection target among the plurality of vehicle detectors based on the attribute classification,
Targeting the unselected vehicle detectors among the plurality of vehicle detectors that are not first selected, the inspection object is secondarily selected based on a traffic volume estimation model among the unselected vehicle detectors to determine the inspection object,
At the time of the secondary selection, between the traffic volume estimation value calculated through the traffic volume estimation model based on the accumulated data of the unselected vehicle detector among the unselected vehicle detectors and the traffic volume observation value collected from the unselected vehicle detector. The inspection target selection device for evaluating the performance of a vehicle detector to perform the secondary selection in consideration of the difference. delete The method of claim 14,
The evaluation criterion data includes data on an average annual traffic volume change pattern for each time slot with respect to the daily traffic volume, with respect to any one of the plurality of vehicle detectors,
The selection unit evaluates the similarity between the average annual traffic volume change pattern for each time slot and the traffic volume change pattern for each time zone acquired from any one of the plurality of vehicle detectors, and when the similarity is out of the allowable range, any one of the vehicle Select the detector as the above inspection object,
The similarity is evaluated by a correlation coefficient that is an index quantifying a difference between two patterns to be compared. The method of claim 14,
The evaluation reference data includes data on an average annual increase/decrease rate of traffic volume of each of the plurality of vehicle detectors,
The selection unit evaluates the similarity between the data on the average annual increase/decrease rate of the traffic volume of one of the plurality of vehicle detectors and the data on the average annual increase/decrease rate of the traffic volume of the other vehicle detector among the plurality of vehicle detectors, allowing a preset similarity. If out of range, to determine that one of the two vehicle detectors are likely to correspond to the inspection target, inspection target selection device for the performance evaluation of the vehicle detector. The method of claim 14,
The evaluation criterion data includes data on an estimate of traffic volume in n time zones estimated using a traffic volume estimation model for adjacent time zones with respect to any one of the plurality of vehicle detectors,
The selection unit compares the estimated traffic volume of the n time zones with the observed traffic volume of the n time zones collected by the one of the vehicle detectors, and selects any one of the vehicle detectors as the inspection object when it is out of a preset allowable range. Inspection target selection device for performance evaluation of phosphorus and vehicle detectors.

Images