Failure probability estimation method, device and storage medium based on finite generalization
Through the failure probability estimation method based on finite generalization, the interquartile range and extreme value theorem combined with maximum likelihood estimation, the problem of difficulty in estimating the failure probability of autonomous vehicles in the absence of fault data is solved, and the estimation is improved. Accuracy and efficiency reduce the risk of autonomous driving.
Patent Information
- Application Number
- CN202410700736.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-31
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-05-31
AI Technical Summary
The prior art is difficult to accurately estimate the probability of autonomous driving vehicles failing in data scenarios without failure or very few failures, resulting in the impact of the safety performance of autonomous driving systems and increasing traffic safety risks.
Using the failure probability estimation method based on finite generalization, the KPI value is determined by obtaining vehicle operating parameters, the interquartile range method is used to determine the threshold value, and the estimated distribution function is constructed based on the extreme value theorem and the maximum likelihood estimation method to calculate the failure probability. , ensure the accuracy and efficiency of the estimation.
It improves the accuracy of the failure probability estimation of the autonomous driving system in the absence of fault data, reduces the risk of autonomous driving, and ensures traffic safety.
Smart Images

Figure CN118503611B_ABST
Abstract
Description
Technical Field
[0001] This document relates to the technical field of failure probability estimation, and in particular to a failure probability estimation method, device and storage medium based on finite generalization. Background Art
[0002] With the accelerated development and application of autonomous driving technology, the various risks existing in the autonomous driving system can be tested through simulation tests. Various complex road conditions and traffic conditions can be simulated in a virtual environment to test the various functions of the autonomous driving car, thereby discovering potential safety hazards and ensuring the safety of autonomous driving cars.
[0003] Existing technologies mainly use reliability analysis algorithms or optimization algorithms to perform iterative sampling, so that the samples gradually converge to the scenario that causes the failure of the tested system or vehicle, and then estimate the failure probability through methods such as Monte Carlo.
[0004] Because the performance of vehicles in some scenarios in the existing technology may only fluctuate, that is, scenarios where they are about to fail but have not failed, it is difficult to use existing technology to estimate the failure probability of the tested system or vehicle for these vehicle operation data with no faults or very few faults, or the measured failure probability has low accuracy, which affects the safety performance of autonomous driving, increases the risk of autonomous driving, and affects traffic safety. Summary of the invention
[0005] In view of the above scheme, the present application aims to propose a failure probability estimation method, device and storage medium based on finite generalization to solve at least one of the above technical problems.
[0006] In a first aspect, one or more embodiments of this specification provide a failure probability estimation method based on limited generalization, including:
[0007] Obtaining operating parameters of the vehicle, and determining KPI values according to the operating parameters;
[0008] Based on the interquartile range method, a threshold is determined according to the KPI value;
[0009] Determine a sample data set according to the threshold and the KPI value;
[0010] Based on the extreme value theorem, determining an extreme value index and a proportion parameter according to the sample data set and the threshold;
[0011] Based on the maximum likelihood estimation method, determining an estimated distribution function according to the extreme value index and the proportion parameter;
[0012] The failure probability is calculated based on the estimated distribution function.
[0013] Further, based on the 1.5 times interquartile range method, the KPI values are sorted, and the upper quartile and lower quartile of the KPI values are calculated;
[0014] The threshold is determined based on the 1.5 times interquartile range method according to the upper quartile and the lower quartile.
[0015] Further, constructing a generalized Pareto distribution function and a vehicle performance distribution function according to the sample data set and the threshold;
[0016] Based on the extreme value theorem, the extreme value index and the proportional parameter are determined according to the generalized Pareto distribution function and the vehicle performance distribution function.
[0017] Furthermore, the calculation formula of the extreme value theorem is as follows:
[0018]
[0019] Among them, z F =inf{y,F Y (y)=1},F Y (y) represents the probability distribution function of KPI, F Y (y) = P(Y <= y);
[0020] ξ represents the extreme value index (EVI);
[0021] σ u represents the scale parameter;
[0022] z (i) =y (i) -u
[0023] z represents y (i) The value obtained by subtracting the threshold from the KPI value greater than the threshold; the KPI value includes multiple values; (i) Indicates the value of the KPI value that is greater than the threshold;
[0024] u represents the threshold;
[0025] The calculation formula of the generalized Pareto distribution function is as follows:
[0026]
[0027] The vehicle performance distribution function calculation formula is as follows:
[0028] G u (z)=P[Y≤u+z|Y>u]
[0029] Among them, Y represents the KPI in the logical scenario.
[0030] Further, a log-likelihood function of a generalized Pareto distribution function is calculated according to the extreme value index and the proportion parameter;
[0031] According to the log-likelihood function, an extreme value index estimator and a proportion parameter estimator are obtained;
[0032] Based on the maximum likelihood estimation method, an estimated distribution function is determined according to the extreme value index estimator and the proportion parameter estimator.
[0033] Further, the estimated distribution function is an estimate of the extreme value index and an estimate of the scale parameter;
[0034] The calculation formula of the failure probability is as follows:
[0035]
[0036] in, is expressed as an estimate of the probability of failure;
[0037] n represents the number of samplings with replacement performed on the kpi;
[0038] represents an estimator of the extreme value index;
[0039] u represents the threshold;
[0040] represents an estimator of the scale parameter;
[0041] k represents the number of KPIs that meet the conditions.
[0042] In a second aspect, an embodiment of the present application provides a failure probability estimation device based on limited generalization, the device comprising an acquisition module, a first determination module, a second determination module, a third determination module, an estimation module and a calculation module;
[0043] The acquisition module is used to acquire the operating parameters of the vehicle and determine the KPI value according to the operating parameters;
[0044] The first determination module is used to determine a threshold value according to the KPI value based on the interquartile range method;
[0045] The second determination module is used to determine a sample data set according to the threshold and the KPI value;
[0046] The third determination module is used to determine the extreme value index and the proportion parameter based on the extreme value theorem according to the sample data set and the threshold;
[0047] The estimation module is used to determine an estimated distribution function based on the extreme value index and the proportion parameter based on a maximum likelihood estimation method;
[0048] The calculation module is used to calculate the failure probability according to the estimated distribution function.
[0049] Furthermore, the device further comprises: a fourth determining module;
[0050] The fourth determination module is used to sort the KPI values based on the 1.5 times interquartile range method, calculate the upper quartile and the lower quartile of the KPI value; based on the 1.5 times interquartile range method, determine the threshold according to the upper quartile and the lower quartile.
[0051] Furthermore, the device further comprises: a fifth determining module;
[0052] The fifth determination module is used to construct a generalized Pareto distribution function and a vehicle performance distribution function according to the sample data set and the threshold; based on the extreme value theorem, the extreme value index and the proportional parameter are determined according to the generalized Pareto distribution function and the vehicle performance distribution function.
[0053] In a third aspect, an embodiment of the present application provides a storage medium, including:
[0054] Used to store computer executable instructions, which, when executed, implement the steps of any one of the failure probability estimation methods in the first aspect.
[0055] Compared with the prior art, this application can at least achieve the following technical effects:
[0056] This application can use the interquartile range method to determine the threshold value, thereby ensuring that the selected sample data set is close to 0 and sufficient. Due to the uncertainty of KPIs in real scenarios, and most of the KPI values are data in safety scenarios, the existing technology cannot use safety data to calculate its failure probability. In order to match the characteristics of data approaching 0, the extreme value theorem is used to determine the approximate distribution of the extreme value index and the proportion parameter. The maximum likelihood estimation method is then used to estimate the extreme value index and the proportion parameter to ensure the accuracy of the failure probability estimation. The failure probability is then calculated based on the estimated distribution function, which solves the problem that when there is no fault data in the existing data, the failure probability is difficult to estimate and the accuracy and efficiency are low, thereby reducing the risk of autonomous driving and ensuring traffic safety. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0058] Figure 1 A method flow chart of a failure probability estimation method based on limited generalization provided for one or more embodiments of this specification;
[0059] Figure 2 A schematic diagram of the structure of a failure probability estimation device based on limited generalization is provided for one or more embodiments of this specification. DETAILED DESCRIPTION
[0060] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.
[0061] At present, in order to measure the failure probability of autonomous driving, a large number of specific scenarios and data are required, or the scenario that causes the tested system or vehicle to fail is gradually converged through iterations to accurately estimate its failure probability. However, in certain specific scenarios, the data is limited and cannot be obtained further. At the same time, the data only contains vehicle data without safety accidents or vehicle data with very few safety accidents. In this case, the failure probability cannot be accurately estimated using existing technology, which affects the safety assessment of autonomous driving. For example: if only one piece of data among 10,000 vehicle driving data is fault data, then the Monte Carlo method can only estimate the probability of failure to be 1 / 10000, which is inaccurate. Even if there is no failure data, it is impossible to estimate the failure probability at this time.
[0062] In order to solve the above technical problems, this application proposes a method for estimating the failure probability of autonomous driving based on limited generalization, such as Figure 1 As shown, the specific steps are as follows:
[0063] Step S1, obtaining the operating parameters of the vehicle, and determining the KPI value according to the operating parameters.
[0064] In an embodiment of the present application, the vehicle operating parameters are obtained based on simulation data or real data, and the KPI value is determined according to the vehicle operating parameters; the KPI value is used to evaluate the performance of the vehicle in a specified scenario, and the KPI value includes two parts, KPI<0 indicates safety, and KPI≥0 indicates failure. In this solution, the obtained operating parameters are data under the same logical scenario by default. The logical scenario represents the expression of entity characteristics and the relationship between entities by defining the parameter range of variables in the state space; the logical scenario is a further detailed description of the functional scenario based on the state space variables, which is used to generate requirements in the project development stage; each logical scenario can derive any number of specific scenarios. Among them, the functional scenario represents: the most abstract level of operation scenario described by semantics, and the functional scenario is used for project definition, hazard analysis and risk assessment in the conceptual stage; during the testing process, it is often necessary to convert the functional scenario into a logical scenario, and convert it into a data format that can be used in the corresponding simulation environment.
[0065] For example, in the functional scenario, there are scenarios such as following a car and parking. For following a car, in the logical scenario, it includes the speed of the front car, the speed of the main car, visibility, etc.; it is expressed in numbers as the speed of the autonomous driving vehicle is 70km / h~90km / h, the distance between the autonomous driving vehicle and the front car is 100m~200m, etc. Then assign specific values to each scenario to form a specific scenario, the speed of the autonomous driving vehicle is 70km / h, and the distance between the autonomous driving vehicle and the front car is 100m.
[0066] Step S2, based on the interquartile range method, determine the threshold according to the KPI value.
[0067] In the embodiment of the present application, based on the 1.5 times interquartile range method, the upper quartile and the lower quartile of the KPI value are calculated according to the KPI value; and the threshold is determined according to the upper quartile and the lower quartile.
[0068] Among them, the interquartile range is an indicator in statistics that describes the degree of distribution of discrete data. It represents the 25%, 50%, and 75% positions in the data respectively. The values corresponding to these positions are the quartiles. The interquartile range represents the degree of dispersion of the data. The larger the interquartile range, the higher the degree of dispersion of the data. Selecting a value of 1.5 times the quartile can ensure that enough KPI values are reserved.
[0069] Specifically, in real life, most scenarios are safety scenarios, that is, KPIs usually reflect the normal driving status of the vehicle, and it is difficult to evaluate the probability of failure at this time.
[0070] In order to estimate the failure probability using normal data, a KPI value close to 0 is determined from the KPIs less than 0. The closer the KPI value is to 0, the more dangerous the vehicle is and the more likely it is to have a safety accident, so it can be used to evaluate the failure probability. Compared with the failure data, there are dozens or even hundreds of KPIs close to 0, so the failure probability calculated using the KPI close to 0 is more accurate, thereby ensuring the accuracy of the failure probability estimation. For example: if 1,000 existing data are obtained, 500 of which have KPI values around -10, 100 are around -5, 199 are around -3, 200 have KPI values around -1, and only 1 has a KPI value of 1. Then the KPI value at this time is around -1, which meets the condition of being close to 0 compared to the data with KPI values of -3, -5 and -10; at the same time, compared to the failure data with only 1 KPI=1, this application has 200 data with KPI values around -1. Therefore, the present application has enough KPI values to be used for calculating the failure probability, thereby ensuring the accuracy of calculating the failure probability.
[0071] In the embodiment of the present application, the accuracy of calculating the failure probability can be guaranteed only when there are enough normal data close to 0 (usually dozens of them are needed). It can be seen that the more KPI values on the left side are close to 0 and not equal to 0, the more accurate the calculated failure probability is.
[0072] Specifically, the 1.5 times IQR (inter quartile range) discrimination method is used to sort the KPI values, calculate the upper quartile, median and lower quartile; and obtain the threshold value based on the upper quartile and lower quartile. The formula is as follows:
[0073] u=1.5*(Q3-Q1)+Q3
[0074] Where u represents the threshold;
[0075] Q3 represents the upper quartile;
[0076] Q1 represents the lower quartile.
[0077] For example: intercept the operation data of autonomous driving vehicles on Highway A within one day, calculate the KPI value based on the operation data of each vehicle; and sort the KPI values of each vehicle, there are 500 KPI values around -10 (can be regarded as 500 KPI = -10), 300 around -7 (can be regarded as 300 KPI = -7), 50 KPI values around -2 (can be regarded as 50 KPI = -2), and 1 KPI value is 1. Based on the 1.5 times IQR discrimination method, the calculated position of Q1 is 213, and the position of Q3 is 639; then according to the positions of Q1 and Q3, it is determined that the value of Q1 is -10 and the value of Q3 is -7, then the threshold value in this logic scenario is -2.5.
[0078] Step S3, determining a sample data set according to the threshold and the KPI value.
[0079] In the embodiment of the present application, according to the threshold and the KPI value, the data greater than the threshold in the KPI value is extracted as the sample data set. If there are k KPI values that meet the conditions, the KPI sample data that meet the conditions is transformed and expressed by the following formula:
[0080] z (i) =y (i) -u;
[0081] z represents y (i) The value obtained by subtracting the threshold from the KPI value greater than the threshold; the KPI value includes multiple values; (i) Indicates the value of the KPI value that is greater than the threshold; i=1, 2, 3..., k;
[0082] u represents the threshold.
[0083] Step S4, based on the extreme value theorem, according to the sample data set and the threshold, determine the extreme value index and the proportion parameter.
[0084] In the embodiment of the present application, based on the sample data and the threshold, the approximate distribution of the extreme value index and the proportional parameter that satisfies the extreme value theorem is determined. Due to the uncertainty of the KPI value and the fact that most of the KPI data in real scenarios are security scenarios, this application uses data approaching 0 as samples. In order to match the characteristics of data approaching 0, the extreme value theorem is used to estimate its failure probability, thereby improving the efficiency of the estimation of the failure probability. For example, the extreme value theorem can use the POT (peaks-over-threshold) or PORT (peaks-over-random-threshold) method.
[0085] The extreme value theorem is expressed as follows:
[0086]
[0087] Among them, z F =inf{y,F Y (y)=1},F Y (y) represents the probability distribution function of KPI, F Y (y) = P(Y <= y);
[0088] ξ represents the extreme value index (EVI);
[0089] σ u Represents the scale parameter;
[0090] z (i) =y (i) -u
[0091] z represents y (i) The value obtained by subtracting the threshold from the KPI value greater than or equal to the threshold u; the KPI value includes multiple values; y (i) Indicates the value of the KPI value that is greater than the threshold;
[0092]
[0093] H ξ (z;σ u ) represents the generalized Pareto distribution function;
[0094] G u (z) = P[Y≤u+z|Y>u] represents the vehicle performance distribution function that theoretically satisfies the conditions;
[0095] Y represents the KPI in the logical scenario;
[0096] | represents conditional probability, whose z (i) , i=1,2,…,k is subject to the distribution function G u (z) is the independent sample data.
[0097] The generalized Pareto distribution function and the vehicle performance distribution function are introduced into the extreme value theorem, and the approximate distribution of the extreme value index and the proportional parameter is determined according to the threshold and the parameters in the randomly selected sample data set. This is equivalent to, z (i) , i=1,2,...,k approximately obeys the unknown parameters ξ and σ u distribution.
[0098] Step S5, based on the maximum likelihood estimation method, according to the approximate distribution, determine the estimated distribution function.
[0099] In the embodiment of the present application, the estimated distribution function is determined based on the approximate distribution probability using the maximum likelihood estimation method. The maximum likelihood estimation can accurately estimate the extreme value index and the proportion parameter, ensuring the accuracy of the failure probability estimation.
[0100] First, by calculating H ξ (z;σ u )
[0101]
[0102] Then the parameter estimator makes L(ξ,σ u )maximum,
[0103]
[0104] represents the estimator of the extreme value index;
[0105] represents the estimator of the scale parameter;
[0106] Then use the maximum likelihood estimation method to estimate, the specific formula is as follows:
[0107]
[0108] in, Represents G u (z) is estimated.
[0109] Step S6, calculating the failure probability according to the estimated distribution function.
[0110] In the embodiment of the present application, the conditional probability formula can be used to obtain
[0111] P[Y≥0]=(1-P[Y≤uu|Y>u])·P[Y>u].
[0112] Among them, P[Y≤uu|Y>u]=G u (-u) can be used Estimate; P[Y>u] can be estimated by k / n. Therefore, we can estimate the failure probability of the tested system in this logical scenario as
[0113]
[0114] in, Expressed as an estimate of the probability of failure;
[0115] n represents the number of samplings with replacement conducted on the kpi;
[0116] k represents the number of KPIs that meet the conditions.
[0117] Step S7, based on the bootsrap sampling method, the variance of the failure probability is estimated according to the failure probability.
[0118] In an embodiment of the present application, the bootsrap sampling method is used to perform sampling with replacement on the sample data set to calculate its failure probability; based on the failure probability, the mean of each failure probability is calculated, and then the variance of the failure probability is calculated using the mean and the number of samplings; then the confidence interval of the failure probability is determined based on the mean of the failure probability, and whether the variance of the failure probability is within the confidence interval is estimated based on the confidence interval and the variance of the failure probability; if so, the estimate of the failure probability is accurate; if not, the failure probability needs to be recalculated.
[0119] Specifically, since the number of failure probabilities calculated by this scheme based on the sample data set is fixed, the bootsrap method is used to extract the sample data set and calculate the failure probability through the replacement sampling method; then, based on the failure probability, its mean and variance are calculated, and the calculation formula of its mean is:
[0120]
[0121] in, represents the estimated mean;
[0122] n represents the number of samplings with replacement conducted on the kpi;
[0123] It represents the result of failure probability calculated based on random sampling, and the mean calculated based on the result of failure probability.
[0124] The formula for calculating variance is as follows:
[0125]
[0126] in, Represents variance.
[0127] The present application embodiment provides a failure probability estimation device based on limited generalization, such as Figure 2 As shown, the device includes an acquisition module 101, a first determination module 102, a second determination module 103, a third determination module 104, an estimation module 105 and a calculation module 106;
[0128] The acquisition module 101 is used to acquire the operating parameters of the vehicle and determine the KPI value according to the operating parameters;
[0129] The first determination module 102 is used to determine a threshold value according to the KPI value based on the interquartile range method;
[0130] The second determination module 103 is used to determine a sample data set according to the threshold and the KPI value;
[0131] The third determination module 104 is used to determine the extreme value index and the proportion parameter based on the extreme value theorem according to the sample data set and the threshold;
[0132] The estimation module 105 is used to determine the estimated distribution function based on the extreme value index and the proportion parameter based on the maximum likelihood estimation method;
[0133] The calculation module 106 is used to calculate the failure probability according to the estimated distribution function.
[0134] In the embodiment of the present application, the apparatus further includes: a fourth determining module;
[0135] The fourth determination module is used to sort the KPI values based on the 1.5 times interquartile range method, calculate the upper quartile and the lower quartile of the KPI value; based on the 1.5 times interquartile range method, determine the threshold according to the upper quartile and the lower quartile.
[0136] In the embodiment of the present application, the apparatus further includes: a fifth determining module;
[0137] The fifth determination module is used to construct a generalized Pareto distribution function and a vehicle performance distribution function according to the sample data set and the threshold; based on the extreme value theorem, the extreme value index and the proportional parameter are determined according to the generalized Pareto distribution function and the vehicle performance distribution function.
[0138] A storage medium is provided in an embodiment of the present application, including: a storage medium for storing computer executable instructions, wherein the computer executable instructions, when executed, implement the steps of any one of the failure probability estimation methods described in the embodiments.
[0139] It should be noted that the embodiments of the storage medium in this specification and the embodiments of the blockchain-based service provision method in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can refer to the aforementioned corresponding implementation of the blockchain-based service provision method, and the repeated parts will not be repeated.
[0140] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0141] In the 1930s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.
[0142] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.
[0143] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0144] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0145] It should be understood by those skilled in the art that one or more embodiments of this specification may be provided as a method, system or computer program product. Therefore, one or more embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0146] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0147] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0149] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0150] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0151] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0152] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0153] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.
[0154] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0155] The above description is only an embodiment of this document and is not intended to limit this document. For those skilled in the art, this document may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this document should be included in the scope of the claims of this document.
Claims
1. A failure probability estimation method based on finite generalization, characterized in that: include: Obtaining the operating parameters of the vehicle, and determining the KPI value according to the operating parameters; wherein the KPI value is used to evaluate the performance of the vehicle in a specified scenario, and the KPI value includes two parts, KPI<0 indicates safety, and KPI≥0 indicates failure; Based on the interquartile range method, a threshold is determined according to the KPI value; According to the threshold and the KPI value, extract the data in the KPI value that is greater than the threshold as a sample data set; Based on the extreme value theorem, determining an extreme value index and a proportion parameter according to the sample data set and the threshold; Based on the maximum likelihood estimation method, an estimated distribution function is determined according to the extreme value index and the proportion parameter; wherein, the log-likelihood function of the generalized Pareto distribution function is calculated according to the extreme value index and the proportion parameter; the estimated distribution function is obtained by calculating H ξ (z;σ u ) According to the log-likelihood function, the extreme value exponent estimator and the proportional parameter estimator are obtained; then the parameter estimator makes L(ξ,σ u )maximum, represents the estimator of the extreme value index; represents the estimator of the scale parameter; Based on the maximum likelihood estimation method, the estimated distribution function is determined according to the extreme value index estimator and the proportion parameter estimator; and then the maximum likelihood estimation method is used for estimation. The specific formula is as follows: in, Represents G u (z) estimates; According to the estimated distribution function, the failure probability is calculated; wherein the calculation formula of the failure probability is as follows: in, is expressed as an estimate of the probability of failure; n represents the number of samplings with replacement performed on the kpi; H represents the generalized Pareto distribution function; represents an estimator of the extreme value index; u represents the threshold; represents an estimator of the scale parameter; k represents the number of KPIs greater than the threshold.
2. The method according to claim 1, characterized in that: The method based on the interquartile range method and determining the threshold value according to the KPI value includes: Based on the 1.5 times interquartile range method, the KPI values are sorted and the upper quartile and lower quartile of the KPI values are calculated; The threshold is determined based on the 1.5 times interquartile range method according to the upper quartile and the lower quartile.
3. The method according to claim 1, characterized in that: The step of determining the extreme value index and the proportion parameter based on the extreme value theorem according to the sample data set and the threshold value includes: According to the sample data set and the threshold, a generalized Pareto distribution function and a vehicle performance distribution function are constructed; wherein the vehicle performance distribution function calculation formula is as follows: G u (z)=P[Y≤u+z|y>u] Wherein, Y represents the KPI in a logical scenario; wherein, the logical scenario represents the expression of entity features and the relationship between entities by defining the parameter range of variables in the state space; Based on the extreme value theorem, the extreme value index and the proportional parameter are determined according to the generalized Pareto distribution function and the vehicle performance distribution function.
4. The method according to claim 3, characterized in that: The calculation formula of the extreme value theorem is as follows: Among them, z F = inf{y,F Y (y)=1},F Y (y) represents the probability distribution function of KPI, F Y (y) = P(Y <= y); ξ represents the extreme value index (EVI); σ u represents the scale parameter; z (o) =y (o) -u z represents y (i) The value obtained by subtracting the threshold from the KPI value greater than the threshold; the KPI value includes multiple values; (i) Indicates the value of the KPI value that is greater than the threshold; u represents the threshold; The calculation formula of the generalized Pareto distribution function is as follows:
5. A failure probability estimation device based on finite generalization, characterized in that: The device comprises an acquisition module, a first determination module, a second determination module, a third determination module, an estimation module and a calculation module; The acquisition module is used to acquire the operating parameters of the vehicle, and determine the KPI value according to the operating parameters; wherein the KPI value is used to evaluate the performance of the vehicle in a specified scenario, and the KPI value includes two parts, KPI<0 indicates safety, and KPI≥0 indicates failure; The first determination module is used to determine a threshold value according to the KPI value based on the interquartile range method; The second determination module is used to extract data greater than the threshold value from the KPI value as a sample data set according to the threshold value and the KPI value; The third determination module is used to determine the extreme value index and the proportion parameter based on the extreme value theorem according to the sample data set and the threshold; The estimation module is used to determine the estimated distribution function based on the maximum likelihood estimation method according to the extreme value index and the proportion parameter; wherein the log-likelihood function of the generalized Pareto distribution function is calculated according to the extreme value index and the proportion parameter; the estimated distribution function is obtained by calculating H ξ (z;σ u ) According to the log-likelihood function, the extreme value exponent estimator and the proportional parameter estimator are obtained; then the parameter estimator makes L(ξ,σ u )maximum, represents the estimator of the extreme value index; represents the estimator of the scale parameter; Based on the maximum likelihood estimation method, the estimated distribution function is determined according to the extreme value index estimator and the proportion parameter estimator; and then the maximum likelihood estimation method is used for estimation. The specific formula is as follows: in, Represents G u (z) estimates; The calculation module is used to calculate the failure probability according to the estimated distribution function; wherein the calculation formula of the failure probability is as follows: in, is expressed as an estimate of the probability of failure; n represents the number of samplings with replacement performed on the kpi; H represents the generalized Pareto distribution function; represents an estimator of the extreme value index; u represents the threshold; represents an estimator of the scale parameter; k represents the number of KPIs greater than the threshold.
6. The device according to claim 5, characterized in that: The device further includes: a fourth determining module; The fourth determination module is used to sort the KPI values based on the 1.5 times interquartile range method, calculate the upper quartile and the lower quartile of the KPI value; based on the 1.5 times interquartile range method, determine the threshold according to the upper quartile and the lower quartile.
7. The device according to claim 5, characterized in that: The device further includes: a fifth determining module; The fifth determination module is used to construct a generalized Pareto distribution function and a vehicle performance distribution function according to the sample data set and the threshold; wherein the vehicle performance distribution function calculation formula is as follows: G u (z)=P[Y≤u+z|y>u] Wherein, Y represents the KPI in a logical scenario; wherein, the logical scenario represents the expression of entity features and the relationship between entities by defining the parameter range of variables in the state space; Based on the extreme value theorem, the extreme value index and the proportional parameter are determined according to the generalized Pareto distribution function and the vehicle performance distribution function.
8. A storage medium, characterized in that: include: Used to store computer executable instructions, which, when executed, implement the steps of the failure probability estimation method according to any one of claims 1 to 4.
