Method, device, computer program and computer-readable storage medium for detecting defective vehicles

Abstract

A method for identifying defective vehicles, wherein the defective vehicles are a subset of vehicles (2) and the vehicles (2) are divided into a plurality of vehicle types, comprising: - providing (S1) an expected value of a number of predefined messages to be sent for each vehicle type, - determining (S2) an actual value of a number of predefined messages sent for each vehicle (2), - determining (S3) a deviation value for each vehicle (2), wherein each deviation value is representative of a difference between the actual value and the expected value, - determining (S4) a defective subgroup of defective vehicles depending on the deviation values, wherein the actual value of the defective subgroup is different from the expected value and the defective vehicles of the defective subgroup each comprise at least one property, wherein the property is the same.

Description

A method for identifying defective vehicles is described. Furthermore, a device, a computer program, and a computer-readable storage medium are specified. The publication DE 10 2013 107 962 A1 describes a method for monitoring at least components (2, 3, 4, 5) of a system (1) which comprises cooperative driver assistance systems (2, 3) and / or stationary infrastructure facilities (4, 5) cooperating with the driver assistance systems (2, 3), wherein the components (2, 3, 4, 5) are equipped for automatic data communication with each other and have at least one communication unit (20, 30, 40, 50) for this purpose, wherein a first component (2, 3, 4, 5) monitors a second, other component (2, 3, 4, 5), in particular its communication unit (20, 30, 40, 50), for functional errors and generates corresponding diagnostic data in the event of a functional error or suspected functional error. The task at hand is to specify a method for identifying defective vehicles particularly easily and quickly. Furthermore, a device and a computer program capable of executing such a method should be specified. Additionally, a computer-readable storage medium containing such a computer program should be provided. These tasks are solved by the method and subject matter of the independent patent claims. Advantageous embodiments, implementations, and further developments are the subject of the respective dependent patent claims. First, the procedure for identifying defective vehicles is explained. The defective vehicles are a subset of vehicles. These vehicles form, for example, a total set, where the vehicles in this total set are, for instance, networked together. The vehicles are networked, for example, by means of a device. Each of the vehicles is connected to the device, in particular an external device, via a communication link. The device is, for example, a server, in particular a backend server. The vehicles are categorized into a variety of vehicle types. A vehicle type is, for example, a designation of the vehicle's type. Each vehicle type encompasses several different vehicles. Vehicles belonging to one vehicle type are not necessarily belonging to another. Vehicle types, for example, group the vehicles. According to at least one embodiment of the method, an expected value for the number of predefined messages to be sent per vehicle type is provided. This expected value is, for example, the average of the number of predefined messages to be sent per vehicle type. The expected values ​​are determined, for example, before the method is executed. The expected value is determined, for example, based on a short period of time, such as the last four or twelve weeks, before the method is executed. Advantageously, the expected value is not static but can change dynamically. According to at least one embodiment of the method, an actual value is determined for each vehicle based on a number of predefined messages sent. For example, the actual values ​​are determined as a function of a time interval. This time interval is, for example, at least 1 hour and at most 48 hours, in particular 24 hours. In particular, the time interval is representative of a weekday. For example, the predefined messages sent for each vehicle during operation are transmitted to the external device, where they are stored and processed. For instance, to determine the actual values, the predefined messages sent by each vehicle within the given time interval are processed. These actual values ​​could be, for example, the number of predefined messages sent within that time interval. The predefined messages to be sent and / or the predefined messages already sent are, for example, messages containing the status data of the respective vehicles. When a vehicle is in operation, it sends, for example, a predefined message containing its status data to the external device. This status data could be, for example, vehicle data displayed to a user in the vehicle and / or on a mobile device. According to at least one embodiment of the method, a deviation value is determined for each vehicle, with each deviation value being representative of a difference between the actual value and the expected value. For example, each vehicle has one deviation value. Furthermore, in addition to the deviation value, each vehicle has at least one other characteristic. This characteristic includes, for example, one of the following pieces of information: vehicle type, product update, software version, combination of control units, assigned home market, backend hub, production date. The vehicles, for example, share at least some of the same characteristics. For example, the information comprehensively specifies the product update, indicating what type of updates the vehicle's software, particularly its control software, includes. Similarly, the information comprehensively specifies the software version, indicating the current state of the vehicle's software, particularly its control software. The information comprehensively specifies the combination of control units, indicating, for example, which control units are installed in the vehicle and / or which control units communicate with each other. The information comprehensively specifies the assigned home market, indicating the geographic region in which the vehicle is primarily operated. The information comprehensively specifies the backend hub to which the vehicle is connected. According to at least one embodiment of the method, a defective subgroup of defective vehicles is identified based on the deviation values. This defective subgroup includes, for example, defective vehicles with comparatively large deviation values. According to at least one embodiment of the method, the actual value of the defective subgroup differs from the expected value. For example, the faulty subgroup comprises faulty vehicles that send a particularly small number of predefined messages compared to the total number of messages to be sent. This means that the faulty vehicles in the faulty subgroup send comparatively few messages. In this case, the actual value of the faulty subgroup is lower than the expected value. Alternatively, the actual value may be greater than the expected value. For example, in this case, the faulty vehicles send more of the predefined messages than expected. For example, the actual value of the faulty subgroup is at least two standard deviations greater or less than the expected value. Such a method allows for the particularly simple and rapid identification of a subgroup of defective vehicles. These defective vehicles share, for example, common characteristics that are responsible for the defect. If the error affects only a subset of vehicles, particularly a subset of networked vehicles in a fleet, the specified method can be used to identify the affected subgroup. This affected subgroup includes, for example, vehicles with the error that are specific to certain vehicle types from a backend hub with a particular software version and / or production date. This means that the characteristics of the defective vehicles are known and targeted corrective measures can be initiated. According to at least one embodiment of the method, when providing the expected value, a standard deviation of the number of predefined messages to be sent for each vehicle type is also provided. Specifically, when providing the expected value, both an expected value and a standard deviation of the number of predefined messages to be sent for each vehicle type are provided. Alternatively, when providing the expected value, a width corresponding to the number of predefined messages to be sent per vehicle type is also provided. This width, for example, corresponds to the distance between the 20th percentile and the 80th percentile of the expected value. According to at least one embodiment of the method, each deviation value is representative of a number of standard deviations from the expected value. According to at least one embodiment of the method, each expected value is determined as a function of several time intervals, and the time intervals are each representative of a day of the week. For example, each day of the week is assigned a single expected value, in particular a single expected value and a single standard deviation, for each vehicle type. According to at least one embodiment of the method, when determining the defective subgroup, a global median is calculated that is representative of all deviation values ​​of the vehicles. In particular, the global median is a median of all deviation values. According to at least one embodiment of the method, when determining the defective subgroup, at least a first subgroup is determined depending on a first property. According to at least one embodiment of the method, the at least one first subgroup has a first median that has a maximum difference to the global median. According to at least one embodiment of the method, when determining the defective subgroup, at least a second subgroup and at least a third subgroup are generated from the at least one first subgroup. According to at least one embodiment of the method, the at least one second subgroup has a second median depending on a second property. According to at least one embodiment of the method, the at least one third subgroup has a third median depending on a third property. According to at least one embodiment of the method, the difference between the second median and the third median is maximized. For example, further subgroups are generated from at least one second subgroup and / or at least one third subgroup, depending on further properties. According to at least one embodiment of the method, the defective subgroup is determined depending on a predetermined population size of the second subgroup and the third subgroup. For example, the predetermined population size is a termination criterion of the procedure. In this case, the subgroup containing errors is formed by the subgroup that is smaller than the predetermined population size. The predetermined population size of the subgroup containing errors is specified based on statistical relevance. For example, the predetermined population size is large enough to prevent overfitting. For instance, the predetermined population size of the subgroup containing errors comprises at least 500 vehicles and at most 5000 vehicles. If the population size of the first subgroup and / or the second subgroup is larger than the specified population size, further subgroups are generated from the second subgroup and / or the third subgroup. The process terminates, for example, if the population size of at least one of the further subgroups is smaller than the specified population size. According to at least one embodiment of the process, the first, second, and third attributes each include at least one of the following pieces of information: vehicle type, product update, software version, combination of control units, assigned home market, backend hub, and production date. If further subgroups are generated, the additional attributes also include one of the aforementioned pieces of information. According to at least one embodiment of the method, the first, second, and third properties differ from one another. In particular, the further properties also differ from one another when generating the further subgroups. Furthermore, a device for identifying defective vehicles is specified. The device is designed to perform the method described herein. Therefore, all features of the embodiment disclosed in connection with the method are also disclosed in connection with the device, and vice versa. Furthermore, a computer program is specified, comprising commands which, when executed by a computer, cause it to perform the procedure described herein. Furthermore, a computer-readable storage medium is specified on which the computer program described here is stored. Exemplary embodiments of the invention are explained in more detail below with reference to the schematic drawings. Figure 1 shows a flowchart of a method according to an embodiment, Figure 2 shows a schematic representation of a system with a device according to an embodiment, and Figure 3 shows a schematic representation for identifying a defective subgroup in the method according to an embodiment. In the flowchart of the procedure according to the embodiment of Fig. 1, a process step S1 is first carried out, in which an expected value of a number of predetermined messages to be sent per vehicle type is provided. For example, each vehicle 2 in a networked fleet is configured to send messages during operation. These messages are predefined and include, for example, status data for the respective vehicle 2. An expected value can be calculated for each vehicle type, representing the number of predefined messages to be sent, with each predefined message being representative of that vehicle type. Furthermore, the expected value includes the number of predefined messages to be sent within a single day of the week. The generation of expected values ​​can take place before the determination of a faulty subgroup of vehicles 2. In a subsequent process step S2, an actual value of a number of predefined messages sent is determined for each vehicle 2. For example, each vehicle 2 in operation sends the predefined messages to a device 1, in particular an external device, where these predefined messages are stored. For example, the number of predefined messages sent within a weekday is determined, corresponding to the actual value. In the subsequent process step S3, a deviation value is determined for each vehicle 2, where each deviation value is representative of a difference between the actual value and the expected value. That is, the number of predefined messages to be sent within a weekday is determined by the number of predefined messages sent within a weekday, specifically on the same weekday. In subsequent process step S4, a defective subgroup of defective vehicles is determined based on the deviation values, where the actual value of the defective subgroup is smaller than the expected value. The defective vehicles each exhibit at least one characteristic, and this characteristic is identical. If the number of predefined messages sent within a weekday is significantly smaller than the number of predefined messages to be sent within that weekday, particularly on the same weekday, the probability that a defect will be induced due to this characteristic is increased. The system according to the embodiment shown in Fig. 2 comprises a device 1, in particular an external device, which is configured to communicate with a vehicle 2 via a communication link 3. The communication link 3 is configured to transmit the predefined messages to the device 1. The device 1 is designed to carry out the method according to Fig. 1. The device is, for example, integrated into a backend server. The device 1 comprises, in particular, a processing unit, a program and data memory, and, for example, one or more communication interfaces. The program and data memory and / or the processing unit and / or the communication interfaces can be integrated into a single unit and / or distributed across multiple units. The program and data storage of device 1 contains, in particular, a program for identifying defective vehicles, which executes the procedure described above. As shown in Fig. 3, a first subgroup SG1 is initially determined based on a first property. This first property is, for example, information regarding a first product update. All vehicles with this property are included in the first subgroup SG1. In particular, the first subgroup SG1 has a first median that exhibits a maximum difference to a global median, where the global median is representative of all deviation values ​​of vehicles 2. For example, the first subgroup SG1 comprises 600,000 vehicles. Subsequently, a second subgroup SG2 and a third subgroup SG3 are generated from the first subgroup SG1. The second subgroup SG2, for example, includes the vehicles from the first subgroup SG1 where the actual value equals the expected value. The third subgroup SG3, for example, includes the vehicles from the first subgroup SG1 where the actual value differs from the expected value. The second subgroup, SG2, comprises vehicles depending on a second attribute. This second attribute is, for example, information regarding a second product update. Therefore, all vehicles in the second subgroup, SG2, include vehicles with both the first and second attributes. Furthermore, the second subgroup, SG2, has a second median. The third subgroup, SG3, comprises vehicles depending on a third property. This third property is, for example, information regarding a home market. All vehicles in the third subgroup, SG3, therefore include vehicles with both the first and the third property. Furthermore, the third subgroup, SG3, has a third median. Here, the second subgroup SG2 and the third subgroup SG3 are chosen such that the difference between the second median and the third median is maximized. Subsequently, further subgroups are generated from the third subgroup SG3 based on additional properties, namely a fourth subgroup SG4 and a fifth subgroup SG5. The fourth subgroup SG4, for example, comprises the vehicles from the third subgroup SG3 where the actual value matches the expected value. The fifth subgroup SG5, for example, comprises the vehicles from the first subgroup SG1 where the actual value differs from the expected value. For instance, the fifth subgroup SG5 contains only 15,000 vehicles. This fifth subgroup SG5 corresponds to the defective subgroup of defective vehicles according to the method in conjunction with the embodiment of Fig. 1. Reference character list 1 Device 2 Vehicle 3 Communication device SG1 first subgroup SG2 second subgroup SG3 third subgroup SG4 fourth subgroup SG5 fifth subgroup S1..S4 Procedure step

Claims

A method for identifying defective vehicles, wherein the defective vehicles are a subset of vehicles (2) and the vehicles (2) are divided into a plurality of vehicle types, comprising: - providing (S1) an expected value of a number of predefined messages to be sent for each vehicle type, - determining (S2) an actual value of a number of predefined messages sent for each vehicle (2), - determining (S3) a deviation value for each vehicle (2), wherein each deviation value is representative of a difference between the actual value and the expected value, - determining (S4) a defective subgroup of defective vehicles depending on the deviation values, wherein the actual value of the defective subgroup is different from the expected value and the defective vehicles of the defective subgroup each comprise at least one property, wherein the property is the same. Method according to claim 1, wherein, when providing the expected value, an additional standard deviation of the number of predetermined messages to be sent is provided for each vehicle type. The method according to claim 2, wherein each deviation value is representative of a number of standard deviations from the expected value. Method according to one of claims 1 to 3, wherein each expected value is determined as a function of several time intervals, and the time intervals are each representative of a day of the week. A method according to any one of claims 1 to 4, wherein, when determining the defective subgroup, a global median is determined which is representative of all deviation values ​​of the vehicles (2), at least a first subgroup (SG1) is determined depending on a first property, wherein the at least one first subgroup (SG1) has a first median which has a maximum difference to the global median, from the at least one first subgroup (SG1) at least a second subgroup (SG2) and at least a third subgroup (SG3) are generated, wherein the at least one second subgroup (SG2) has a second median depending on a second property, and the at least one third subgroup (SG3) has a third median depending on a third property, and the difference between the second median and the third median is maximized. Method according to claim 5, wherein the defective subgroup is determined depending on a predetermined population size of the second subgroup (SG2) and the third subgroup (SG3). Method according to claim 5 or 6, wherein the first, second and third property each comprise at least one of the following information: vehicle type, product update, software version, combination of control units, assigned home market, backend hub, production date, and the first, second and third property differ from each other. Device (1) for identifying defective vehicles, which is designed to perform the method according to any one of claims 1 to 7. A computer program comprising instructions which, when the computer program is executed by a computer, cause it to execute the method according to any one of claims 1 to 7. Computer-readable storage medium on which the computer program according to claim 9 is stored.

Images