Vehicle data management device, vehicle data management program, and vehicle data management method

Abstract

To provide a vehicle data management device which can manage vehicle data likely to be necessary for later analysis, in distinction from the other vehicle data.SOLUTION: The vehicle data management device comprises: a processing unit which executes preliminarily set processing on vehicle data collected from a vehicle; an acquisition unit which acquires an observed value varying by the execution of the processing, from an observation object set in accordance with a type of vehicle data; a calculation unit which calculates an evaluation value evaluating abnormality of the vehicle data on the basis of the observed value; and a distinction unit which distinguishes the vehicle data on the basis of the evaluation value.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a vehicle data management device, a vehicle data management program, and a vehicle data management method.

Background Art

[0002] Recently, communication devices have been installed in vehicles, and various vehicle data acquired by sensors and the like have come to be transmitted to a server of a vehicle manufacturer or the like. By analyzing the received vehicle data, the server predicts, for example, a failure of the vehicle or provides driving support for the driver. In addition, the vehicle data collected in this way is stored in a storage device and may be used for cause analysis or the like when a failure or the like occurs later.

[0003] By the way, the number of vehicles supplied by vehicle manufacturers to the market is enormous. If all the various vehicle data generated by each vehicle were to be stored in the server, the costs and resources spent on data collection would increase explosively. Therefore, attempts have been made to set collection conditions for the vehicle data generated by the vehicle and to reduce the amount of vehicle data to be stored (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Collecting vehicle data only from vehicles that meet specific criteria may result in the collection of data only from vehicles whose causes of malfunctions have been identified. This could lead to a situation where data necessary for investigating unknown causes in subsequent analysis is not collected. While measures may be taken to initially store all vehicle data and then periodically delete data that has expired or has not been used, there is still a possibility that the relevant vehicle data may be lost when it is needed.

[0006] This invention was made to solve these problems and provides a vehicle data management device, etc., that can manage vehicle data that may be necessary for later analysis separately from other vehicle data. [Means for solving the problem]

[0007] A vehicle data management device in a first aspect of the present invention comprises a processing unit that performs a pre-set process on vehicle data collected from a vehicle, an acquisition unit that acquires observed values ​​that change as a result of the execution of the process from observation targets set according to the type of vehicle data, a calculation unit that calculates an evaluation value for evaluating the abnormality of the vehicle data based on the observed value, and a sorting unit that sorts the vehicle data based on the evaluation value.

[0008] Furthermore, the vehicle data management program in the second aspect of the present invention causes a computer to perform the following steps: a processing step of performing a pre-set process on vehicle data collected from a vehicle; an acquisition step of obtaining observed values ​​that change as a result of the execution of the process from observation targets set according to the type of vehicle data; a calculation step of calculating an evaluation value for evaluating the abnormality of the vehicle data based on the observed value; and a sorting step of sorting the vehicle data based on the evaluation value.

[0009] Furthermore, the vehicle data management method in the third aspect of the present invention includes a processing step in which a computer performs a pre-set process on vehicle data collected from a vehicle; an acquisition step in which the computer obtains observed values ​​that change as a result of the execution of the process from observation targets set according to the type of vehicle data; a calculation step in which the computer calculates an evaluation value for evaluating the abnormality of the vehicle data based on the observed value; and a sorting step in which the computer sorts the vehicle data based on the evaluation value. [Effects of the Invention]

[0010] The present invention provides a vehicle data management device, etc., that can manage vehicle data that may be necessary for later analysis, separately from other vehicle data. [Brief explanation of the drawing]

[0011] [Figure 1] This diagram illustrates the configuration of the management server according to this embodiment and the overall environment in which the management server is used. [Figure 2] This figure shows an example of a vehicle data processing log. [Figure 3] This figure shows an example of the processing log for evaluation values. [Figure 4] This figure shows an example of an anomaly estimation log edited by the sorting department. [Figure 5] This diagram explains the calculation concept described in the evaluation value. [Modes for carrying out the invention]

[0012] The present invention will be described below through embodiments, but the claims are not limited to the following embodiments. Furthermore, not all of the configurations described in the embodiments are necessarily essential for solving the problem.

[0013] Figure 1 is a diagram illustrating the configuration of a management server 100, which is one embodiment of the vehicle data management device according to this embodiment, and the overall environment in which the management server 100 is used. The management server 100 is, for example, a server operated by a vehicle manufacturer, and collects vehicle data from vehicles supplied to the market by the vehicle manufacturer.

[0014] Specifically, each vehicle targeted for vehicle data collection is equipped with a communication device. This device converts detection signals from on-board sensors that monitor the vehicle's status and behavior into vehicle data, which is then transmitted to the management server 100 via a communication network. The management server 100 is not limited to collecting vehicle data directly from the target vehicles; for example, it may also collect vehicle data from sensor equipment installed along the roadside that has detected and processed the target vehicles. The communication network can be the internet or a 5G network.

[0015] Vehicle data includes vehicle-specific identification information (Vehicle ID) and detection data adjusted from detection signals from on-board sensors that monitor the vehicle's target. This data is generated for each on-board sensor, for example, at regular intervals, when a specific condition is detected, or upon request from the management server 100. The detection data includes time information when the detection signal was detected.

[0016] The management server 100 is a computer and includes a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), memory, storage devices, a network unit, etc. The CPU works in cooperation with the GPU and other components to perform various processes by loading programs stored in the storage devices into memory and executing them. The memory consists of volatile memory such as RAM (Random Access Memory), and the storage devices consist of non-volatile memory such as HDD (Hard Disk Drive) or SSD (Solid State Drive). The network unit is responsible for connecting to the communication network and exchanging data with vehicles, and is composed of a LAN unit, for example.

[0017] The CPU serves as a functional arithmetic unit that executes various operations according to the processes instructed by the vehicle data management program. The CPU can mainly function as a data processing unit 110, an observed value acquisition unit 120, an evaluation value calculation unit 140, a classification unit 160, and a recording control unit 190. When the vehicle data management device is composed of multiple servers, the CPUs provided in each server may share and perform the functions as the above functional arithmetic unit.

[0018] Also, the storage device serves as a storage unit that stores various types of data respectively. The storage device can mainly function as an observed value storage unit 130, an evaluation value storage unit 150, an abnormal estimation log storage unit 170, and a processing log storage unit 180. Note that the storage device may have a configuration independent of the management server 100, or may be composed of multiple storage devices.

[0019] The data processing unit 110 executes various service processes which are preset processes for the collected vehicle data. For example, when accelerator operation data, brake operation data, and steering wheel operation data are collected from a certain vehicle, a behavior analysis process related to abnormal driving is set as a service process to be executed, and the data processing unit 110 executes the service process. As a result of the behavior analysis process, for example, when it is determined that there is a sudden steering or a wobbly drive, the data processing unit 110 performs response processes such as confirming the occurrence of an accident and issuing a drowsiness warning to the vehicle.

[0020] Also, when GPS data indicating the current position of the target vehicle is collected as vehicle data, a search process for searching for an optimal route and a route confirmation process for supporting safe driving are set as service processes to be executed, and the data processing unit 110 executes those service processes. As a result of the route confirmation process, for example, when it is detected that a natural disaster has occurred in the route direction, the data processing unit 110 performs response processes such as advising the vehicle to change the route.

[0021] Also, when collecting surrounding image data during driving as vehicle data, an image confirmation process for checking whether the camera unit is functioning properly is set as a service process to be executed, and the data processing unit 110 executes the service process. When the data processing unit 110 determines, as a result of the image confirmation process, for example, a defect in exposure adjustment or focus adjustment, it performs a response process such as imposing a function limit on a pedestrian detection function or a sign detection function that uses the surrounding image data, for example.

[0022] After the data processing unit 110 executes such a service process for each vehicle data and executes a response process as necessary, it delivers the processing result of the service process to the recording control unit 190 together with the vehicle data. Also, a processing log is generated as a record of the service process and accumulated in the processing log storage unit 180.

[0023] Here, an example of a processing log table in which the processing logs of vehicle data are integrated will be described. FIG. 2 is a diagram showing an example of a processing log table of vehicle data. The processing log table of vehicle data is an execution date and time for each executed service process, a service process ID, a main layer that executes the service process, a target vehicle ID, and a recording file for recording a service log.

[0024] The execution date and time of the process is date and time information when the service process is executed. If it is a service process that uses vehicle data acquired over a certain period or a plurality of vehicle data over a certain period, that period is described. The service process ID is identification information indicating which service process was executed. The main layer that executes the service process is, for example, a layer that is preset and mainly generates a load for the service process. A layer is defined as a hierarchical unit that constitutes a function for processing vehicle data, and specifically, each layer of the OSI reference model can be adopted. When each layer of the OSI reference model is adopted, it is divided into a physical layer, a network layer, an application layer, etc., and the application layer may be further divided into a middleware layer, an individual application layer, etc.

[0025] The target vehicle ID is a unique identification information (Vehicle ID) included in the vehicle data, representing the vehicle on which the service process was performed. The service log describes the processing status and information related to the service process. The processing status can be described as normal, abnormal, error, unknown, etc. For example, for an "abnormal" status, related information such as the abnormal value resulting from the processing is described.

[0026] Returning to Figure 1, the explanation continues. The data processing unit 110 generates a load on the aforementioned layer when it performs a pre-configured service process on the collected vehicle data. The observation value acquisition unit 120 observes the state of the layer, which is set as the observation target according to the type of vehicle data, and acquires the observation value that changes as a result of the execution of the service process.

[0027] Observations indicating the state of a layer can be obtained as metrics, which are aggregated data over a specific period. For example, the processing time, error rate, and resource utilization rate of the layer being observed can be observed values. More specifically, for example, the utilization rate of hardware resources (CPU utilization rate and memory utilization rate), the time spent using hardware resources, the amount of input / output data, and the frequency of errors until a certain service process is completed can be observed values ​​to be obtained. The observation value acquisition unit 120 collects the execution date and time for each executed service process, the service process ID, the main layer executing the service process, the metric ID indicating the target of observation, and the acquired observed values, and stores them in the observation value storage unit 130.

[0028] The evaluation value calculation unit 140 reads observed data from the observed data storage unit 130 at a preset timing and calculates an evaluation value to assess the anomaly of the vehicle data targeted for the service processing based on the observed data. "Vehicle data anomaly" refers to the anomaly of the detection data itself or the processing result of the service processing included in the collected vehicle data. However, in this embodiment, the evaluation value calculation unit 140 does not calculate an evaluation value that directly evaluates such detection data or processing results, but rather calculates an evaluation value that estimates the anomaly of the vehicle data using observed values ​​that indicate the state of the set layer when the target service processing is executed. The evaluation value in this embodiment is a so-called anomaly score, and the calculation formula is defined such that a larger value is assigned the more anomaly is estimated.

[0029] The evaluation value calculation unit 140 calculates an evaluation value by comparing the actually observed values ​​with at least one of the predicted values ​​and actual values ​​that would be calculated if the vehicle data were normal when the service processing is performed. Predicted values ​​are generated, for example, from simulation results or experimental results of service processing performed on selected normal vehicle data. Actual values ​​are generated by collecting observed values ​​observed when service processing is performed on vehicle data that was judged to be normal from past vehicle data. The specific concept of calculating the evaluation value will be described later.

[0030] The evaluation value calculation unit 140 compiles the calculated evaluation values ​​into a processing log and stores them in the evaluation value storage unit 150. Here, an example of a processing log table that aggregates the processing logs of evaluation values ​​will be described. Figure 3 is a diagram showing an example of an evaluation value processing log table. The evaluation value processing log table is a record file that records the execution date and time for each executed service process, the service process ID, the main layer that executes the service process, the metric ID indicating the observed target, the acquired observed value, and the evaluation value (anomaly score).

[0031] The execution date and time of the process, the service processing ID, and the main layer executing the service processing are the same as in the example of the vehicle data processing log table. The metric ID indicating the observed object represents the hardware resources or processing blocks of the main layer that are being observed. The observed value is as described above and is the value observed on the observed object. The observed value is described, for example, as CPU usage being 11.11%. The evaluation value is as described above and is a value calculated from the observed value. The evaluation value is described, for example, as a normalized value between 0 and 1.

[0032] Returning to Figure 1, the explanation continues. The sorting unit 160 reads the evaluation value processing log table from the evaluation value storage unit 150 at a predetermined timing and sorts each processing log into at least logs estimated to be abnormal and logs estimated to be normal based on the evaluation value. Specifically, the sorting unit 160 can sort processing logs based on the magnitude of the evaluation value. For example, a threshold is set for each service process, and processing logs with an evaluation value greater than that threshold are sorted into logs estimated to be abnormal.

[0033] The sorting unit 160 further reads the vehicle data processing log table from the processing log storage unit 180, extracts only the processing logs corresponding to the processing logs sorted into anomaly estimation logs, and edits them into an anomaly estimation log table. The sorting unit 160 stores the edited anomaly estimation log table in the anomaly estimation log storage unit 170.

[0034] Here, we will describe an example of an anomaly estimation log table, which is an aggregate of anomaly estimation logs. Figure 4 shows an example of an anomaly estimation log table compiled by the sorting unit 160. Specifically, it extracts processing logs in which anomalies are estimated from the vehicle data processing log table shown in Figure 2, based on the evaluation values ​​in the evaluation value processing log table shown in Figure 3.

[0035] For example, if the threshold for the evaluation value of "spark_012," which is the observed target, is set to 0.5 for a service process represented by the service process ID "service003," the processing log for the service process "service003" executed at the execution date and time "2022 / 03 / 30 15:01:30-15:02:30" as shown in Figure 3 will be classified as an anomaly estimation log because the evaluation value of "spark_012" is 0.6. Then, the processing log for "service003" executed at the execution date and time "2022 / 03 / 30 15:01:30-15:02:30" is extracted from the processing log table shown in Figure 2. The extracted processing log is added to the end of the anomaly estimation log table. In this way, the anomaly estimation log table is updated each time an anomaly estimation log is classified.

[0036] Returning to Figure 1, the explanation continues. The sorting unit 160 passes the extracted abnormality estimation log information to the recording control unit 190. The recording control unit 190 selects vehicle data corresponding to the abnormality estimation log from the vehicle data for which the data processing unit 110 has performed service processing, and stores only the selected vehicle data, along with the processing results of that service processing, in the processing data storage device 200. The processing data storage device 200 is a storage device connected to the management server 100 and is composed of non-volatile memory such as an HDD or SSD.

[0037] Furthermore, a display device 300 is connected to the management server 100, and the display device 300 displays, for example, an anomaly estimation log table stored in the anomaly estimation log storage unit 170, so that the administrator can check the processing status of the management server 100. The display device 300 may also be a display on a portable terminal carried by the administrator.

[0038] By configuring the system to select vehicle data corresponding to anomaly estimation logs and store it in the processing data storage device 200, the number of vehicle data entries to be stored can be significantly reduced compared to the enormous amount of vehicle data sent to the management server 100. This contributes to reducing the costs and resources spent on data collection. Furthermore, since the detection data itself and the processing results of the service processing contained in the collected vehicle data are not directly used as material for anomaly determination, it is expected that situations will be prevented where vehicle data necessary for investigating unknown causes in later analysis is not collected. In other words, vehicle data that may be necessary in later analysis can be managed separately from other vehicle data.

[0039] In the above embodiment, only vehicle data corresponding to the anomaly estimation log and its processing results were stored in the processing data storage device 200. However, the handling of vehicle data and its processing results is not limited to this. Vehicle data other than the vehicle data corresponding to the anomaly estimation log and its processing results may be stored in another processing data storage device, for example, which is deleted after a certain period of time. Alternatively, vehicle data corresponding to the anomaly estimation log may be treated in a way that distinguishes it from other vehicle data, for example, by adding tag information indicating this, and then all vehicle data and its processing results may be stored in the processing data storage device 200.

[0040] Next, we will explain in detail the concept of calculating the evaluation value. Figure 5 is a diagram illustrating the calculation concept for calculating the evaluation value. The upper figure shows the time course of the CPU usage rate being observed, with the horizontal axis representing elapsed time and the vertical axis representing the usage rate (%). The lower figure shows the time course of the calculated evaluation value, with the horizontal axis representing the elapsed time which is the same as the horizontal axis in the upper figure, and the vertical axis representing the evaluation value.

[0041] In the figure above, the solid line shows the trend of actually observed values, and the dotted line shows the trend of predicted values. As can be seen in the figure, there are occasional discrepancies between the trends of observed values ​​and predicted values. When the actually observed values ​​deviate from the predicted values ​​that should be calculated for normal vehicle data, it is thought that a different processing method than the normal processing was required to process the acquired vehicle data, or that a processing method that should have been performed was omitted, and it is presumed that such vehicle data contains some kind of anomaly.

[0042] Comparing the trends in observed and predicted values ​​with the trends in evaluation values ​​shown in the figure below, it can be seen that the evaluation value increases when there is a discrepancy between the observed and predicted values. In other words, it can be seen that the calculation formula is set so that the evaluation value increases when there is a discrepancy between the observed and predicted values. For example, if the threshold for the evaluation value is set to 0.6 (dotted line in the figure below), the processing logs of vehicle data corresponding to the time period when the evaluation value is above the dotted line will be classified as anomaly estimation logs.

[0043] In the embodiment described above, the management server 100 is operated by a vehicle manufacturer, and the target of vehicle data collection is assumed to be vehicles supplied to the market by that vehicle manufacturer. However, the operation of the management server 100 is not limited to this form. The management server 100 may be managed by multiple vehicle manufacturers, or it may be managed by a telecommunications carrier or government agency independent of the vehicle manufacturers. Furthermore, the vehicles are not limited to private cars, but may also include public vehicles such as buses and trucks. [Explanation of symbols]

[0044] 100...Management server, 110...Data processing unit, 120...Observation value acquisition unit, 130...Observation value storage unit, 140...Evaluation value calculation unit, 150...Evaluation value storage unit, 160...Sorting unit, 170...Anomaly estimation log storage unit, 180...Processing log storage unit, 190...Recording control unit, 200...Processing data storage device, 300...Display device

Claims

1. A processing unit that performs pre-configured processing on vehicle data collected from the vehicle, An acquisition unit that acquires observed values ​​that change as a result of the execution of the process from an observation target set according to the type of vehicle data, A calculation unit that calculates an evaluation value for evaluating the abnormality of the vehicle data based on the observed values, A sorting unit that sorts the vehicle data based on the evaluation value. A vehicle data management device equipped with the following features.

2. The vehicle data management device according to claim 1, further comprising a recording control unit that determines whether or not to store the vehicle data in a storage device based on the sorting performed by the sorting unit.

3. The vehicle data management device according to claim 1 or 2, wherein the acquisition unit acquires at least one of the usage rate, usage time, and error frequency of the hardware resources that perform the processing as the observed value.

4. The vehicle data management device according to claim 1 or 2, wherein the calculation unit calculates the evaluation value by comparing the observed value with at least one of the predicted value and the actual value.

5. The vehicle data management device according to claim 1 or 2, further comprising a log creation unit that creates a log relating to the vehicle data that is determined to be highly abnormal based on the sorting by the sorting unit.

6. A processing step that performs pre-configured processing on vehicle data collected from the vehicle, An acquisition step of acquiring observed values ​​that change as a result of the execution of the process from an observation target set according to the type of vehicle data, A calculation step of calculating an evaluation value for evaluating the abnormality of the vehicle data based on the observed values, A sorting step of sorting the vehicle data based on the evaluation value, A vehicle data management program that causes a computer to execute a command.

7. A processing step that causes a computer to perform pre-configured processing on vehicle data collected from a vehicle, An acquisition step in which the computer obtains observed values ​​that change as a result of the execution of the process from an observation target set according to the type of vehicle data, A calculation step in which a computer is instructed to calculate an evaluation value for evaluating the abnormality of the vehicle data based on the observed values, A sorting step in which the vehicle data is sorted by a computer based on the evaluation value. A vehicle data management method having the following characteristics.

Images