Vehicle user big data portrait analysis method and system based on damage model

Abstract

The invention discloses a vehicle user big data portrait analysis method and system based on a damage model, and the method comprises the steps: pre-building an operation feature classification matrix of a vehicle, carrying out the feature extraction based on the classification matrix in combination with the actual operation condition of the vehicle, and taking the feature as a user tag portrait; performing damage classification definition, damage key factor definition and damage correlation parameter definition according to a failure scene of the engine, and extracting damage correlation parameters according to the damage key factor matched with the user tag portrait; and quantitative calculation and cumulative distribution normalization processing are carried out according to the collected key damage correlation parameters to obtain damage data, and user tag portraits and representative users representing different damage classifications of the engine are clustered according to damage data sorting. According to the method, the mathematical damage model converted from the fault failure mechanism of the engine is applied to the big data of the vehicle-mounted terminal for statistics and analysis, and the data is used as bench test and design simulation input, so that the unification of development verification and market application is realized.

Description

Technical Field

[0001] The present invention relates to the field of engine technology, and in particular to a vehicle user big data portrait analysis method and system based on a damage model. Background Art

[0002] User portrait refers to a method of summarizing a series of user usage information, such as driving area, usage duration, usage frequency and other information tags. User portrait is an effective tool to associate target user usage with product development goals. By outlining the user portrait of historical products, we can obtain accurate knowledge of the application of a certain product segment, quantitatively describe user characteristics, and provide targeted services for product design and verification.

[0003] At present, the mainstream user portraits for whole vehicle applications have some shortcomings. For example, the patent publication number "CN115221234A" discloses a method and system for user profiling based on powertrain data. It mainly clusters the analysis based on power and economy, but does not fall to the engine level. In addition, there is a lack of quantitative evaluation methods for reliability, which is of concern to customers. It is difficult to accurately describe it from the user side. The data utilization rate of engine terminal operation is insufficient, the identified user portraits are inaccurate, and it is impossible to subdivide the engine application scenarios and working characteristics from the perspective of reliability through user portraits. How to improve the big data clustering evaluation of user portraits is also an urgent problem that needs to be solved in the industry.

[0004] The information disclosed in the above background technology section is only intended to increase the understanding of the overall background of the present invention, and should not be regarded as admitting or suggesting in any form that the information constitutes the prior art already known to ordinary technicians in this field. Summary of the invention

[0005] In view of the above-mentioned existing defects, the purpose of the present invention is to provide a vehicle user big data portrait analysis method and system based on a damage model, collect the operating data of the engine under different vehicle models and usage areas, combine the failure and damage mechanism of components, conduct user portrait cluster analysis, determine the typical user scenarios and extremely harsh user scenarios related to component reliability, and carry out targeted engine component development verification and reliability improvement work.

[0006] The technical solution of the present invention is as follows:

[0007] A vehicle user big data portrait analysis method based on a damage model, comprising:

[0008] S1: Pre-build the vehicle's operating feature classification matrix, combine the vehicle's actual operating conditions, and extract features based on the classification matrix as a user label profile;

[0009] S2: According to the failure scenario of the engine, damage classification, damage key factors and damage association parameters are defined, and damage association parameters are extracted according to the damage key factors matched by the user tag profile;

[0010] S3: Quantitative calculation and cumulative distribution normalization are performed based on the collected key damage association parameters to obtain damage data. The damage data is sorted and clustered to obtain user label portraits and representative users that represent different damage classifications of the engine.

[0011] Specifically, the operating characteristics of the vehicle in S1 include: user operation, external use environment, road characteristics, vehicle driving task profile, and load characteristics.

[0012] Specifically, the damage classification definition and damage association parameter definition in S2 include: defining the damage classification and damage association parameters according to the reliability failure mechanism of engine parts and the failure mechanism of engine functions not meeting the requirements, and associating the damage classification with the user label.

[0013] Specifically, the quantitative calculation and cumulative distribution normalization processing are performed in S3, including the following steps:

[0014] S301: Applying the component damage model formula, calculating the component failure pseudo damage data value according to the collected key damage association parameters;

[0015] S302: performing cumulative distribution calculation on all data belonging to the same damage classification according to the calculated pseudo damage data value, and obtaining a cumulative distribution value;

[0016] S303: define relative weights of different damage cumulative distribution values, and calculate damage weighting factors of different users as damage data.

[0017] The present invention also discloses a vehicle user big data portrait analysis system based on a damage model, which adopts the above-mentioned method and includes an engine ECU, a T-BOX vehicle-mounted terminal, a cloud platform, and a client computer. The engine ECU is connected to the T-BOX vehicle-mounted terminal by wired communication, and the cloud platform is connected to the T-BOX vehicle-mounted terminal and the client computer by wireless communication, respectively, wherein: the engine ECU is used to record the actual operating status data of the vehicle, and transmit it to the cloud platform for storage through the T-BOX vehicle-mounted terminal; the client computer is used to obtain the vehicle operating status data stored in the cloud platform, and perform feature extraction according to the classification matrix to form a user label portrait; the client computer is also used to extract the damage-related parameters of the engine according to the damage key factors of the engine matched by the user label portrait; the client computer is also used to perform quantitative calculation and cumulative distribution normalization processing on the damage-related parameters to obtain damage data; the client computer is also used to sort and cluster user label portraits and representative users that characterize different types of engine failures according to the damage data.

[0018] Beneficial effects of the present invention:

[0019] The invention associates the engine failure with the user portrait. The innovation lies in taking the engine failure mechanism as the prototype, applying the mathematical damage model of physical failure conversion to the vehicle terminal big data, performing data statistics and analysis, sorting and clustering the user label portraits and representative users that characterize different types of engine failures. The actual operation data of such users is used as the input for bench testing and design simulation, so as to achieve the unity of development verification and market application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 The present invention is a flow chart of a vehicle user big data portrait analysis method based on a damage model.

[0021] Figure 2 This is a schematic diagram of the architecture of a vehicle user big data portrait analysis system based on a damage model according to the present invention. DETAILED DESCRIPTION

[0022] In order to explain the technical content, purpose and effect of the present invention in detail, the following is an explanation in combination with embodiments and accompanying drawings. In the description of the embodiments, it should be understood that the terms indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present embodiment and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention.

[0023] Example 1

[0024] like Figure 1As shown, according to a specific implementation method of this scheme, a vehicle user big data portrait analysis method based on a damage model is implemented by the following steps:

[0025] S1: Create a vehicle operation feature classification matrix, combine the actual vehicle operation status, and extract features based on the classification matrix as a user label profile. Specifically, it includes the following steps:

[0026] User matrix classification: User classification is performed based on user operations, external usage environment, road characteristics, vehicle driving task profile, load characteristics, etc., as shown in Table 1.

[0027]

[0028] User label definition: The user matrix classification is combined with the actual operation characteristics to extract as the basis of user portrait; as shown in Table 2, abc1d4e3 represents the user portrait of overloaded and speeding driving on mountain roads, which can be subsequently classified as high-load reliability damage.

[0029]

[0030] S2: According to the failure scenario of the engine, damage classification, damage key factors and damage association parameters are defined, and damage association parameters are extracted according to the damage key factors matched by the user tag profile; specifically, the following steps are included:

[0031] Reliability failure classification: Based on the reliability failure mechanism of the main engine components, the function or performance of the components will decline due to the load changes during normal use, and the damage classification definition and damage association parameter definition are performed; as shown in Table 3.

[0032]

[0033] Functional failure classification: According to the failure mechanism that the main functions of the engine do not meet, damage classification and damage association parameter definition are performed; as shown in Table 4.

[0034]

[0035] Damage data extraction: Based on the extracted damage-related parameters and users divided according to user tag definitions, T-BOX data collection and key parameter extraction analysis are performed. T-BOX is a vehicle-mounted remote data collection terminal, which is partially popular in vehicles and can directly read GPS and engine ECU data streams on the cloud platform. Data analysis is to extract the proportion distribution and amplitude distribution of damage-related parameters according to the key damage factors.

[0036] S3: Perform quantitative calculation and cumulative distribution normalization processing based on the collected key damage correlation parameters to obtain damage data, sort and cluster the damage data to obtain user label portraits and representative users that represent different damage classifications of the engine. Specifically, quantitative calculation and cumulative distribution normalization processing include the following processes:

[0037] User damage calculation: Apply the component damage model formula and perform component failure pseudo damage calculation and analysis based on the collected key damage association parameters. For example, for low-cycle fatigue calculation of component load amplitude, the amplitude change size and number within the time history of the collected data can be calculated by rain flow counting, and the torque amplitude damage is calculated as the fifth power of the torque amplitude * number of times / collection time, which is used to represent low-cycle fatigue pseudo damage; the calculation results are shown in Table 5, and Table 6 is a schematic diagram of the damage model formula for some key components.

[0038]

[0039]

[0040] in, is the temperature variation amplitude of the parts; k is the failure factor; PFP(t) is the burst pressure; n(t) is the engine speed; N(t) is the frequency under the corresponding load; N sj The frequency of engine shutdown ≥ 2s.

[0041] Damage distribution sorting: Based on the pseudo-damage data of N users calculated in step 6, all data of similar failures are cumulatively distributed. For example, you can use EXCEL to calculate the mean and variance respectively, and then use the NORM.DIST function to find the cumulative distribution value. The cumulative distribution value can represent the degree of damage; as shown in Table 6, the distribution is calculated using low-cycle fatigue torque fluctuations. The role of cumulative distribution is to non-dimensionalize the data into a value with a probability between 0 and 1, which is used for data sorting and value selection. In engineering, it is generally believed that the characteristic users represented by the values ​​of 0.9 to 0.95 are typical and can be used as user representatives of this type of label.

[0042]

[0043] Damage weighted calculation: It can usually be divided into three categories for calculation. Reliability failure can be classified according to the damage ranking of single failure mode; the relative weights of different failure damages can also be defined to meet the low cycle weight Z1 + high cycle weight Z2 + friction and wear weight Z3 + aging weight Z4 = 1; calculate the relative reliability failure weighting factors of different users; functional failure parameters are calculated and ranked separately for each failure; as shown in Table 7.

[0044]

[0045]

[0046] User clustering: Based on the reliability damage weighted calculation factor or the cumulative distribution classification of a reliability failure mode, 0.8 to 1 is classified as high reliability failure risk, 0.6 to 0.8 is classified as medium reliability failure risk, and <0.6 is classified as low failure risk. Based on this principle, user clustering is performed and combined with user label classification to determine the distribution characteristics of user portraits based on reliability failure.

[0047] In summary, this embodiment proposes a user big data portrait analysis method based on a damage model, which is applicable to all engines. By collecting T-BOX data operated by users and combining common reliability damage and functional damage parameters of engines, the characteristic user portraits of common reliability and functional failure modes of engines are refined through quantitative calculation of damage parameters and cumulative distribution normalization processing, and the user's use of characteristic parameters is quantitatively described. Based on the determined reliability damage and functional damage user distribution, user label features are clustered to provide target input for design simulation and experimental verification, thereby achieving precise development and verification.

[0048] Example 2

[0049] This embodiment is based on the first embodiment and discloses a vehicle user big data portrait analysis system based on a damage model. Figure 2 As shown, it includes engine ECU, T-BOX vehicle terminal, cloud platform, and client computer. The engine ECU is connected to the T-BOX vehicle terminal by wired communication, and the cloud platform is connected to the T-BOX vehicle terminal and the client computer by wireless communication, wherein:

[0050] The engine ECU is used to record the actual operating status data of the vehicle and transmit it to the cloud platform for storage through the T-BOX vehicle terminal;

[0051] The client computer is used to obtain the vehicle operation status data stored in the cloud platform and extract features according to the classification matrix to form a user label portrait;

[0052] The client computer is also used to extract engine damage-related parameters based on the engine damage key factors matched by the user tag portrait;

[0053] The client computer is also used to perform quantitative calculations and cumulative distribution normalization on damage-related parameters to obtain damage data;

[0054] The client computer is also used to sort and cluster the damage data to produce user label portraits and representative users that characterize different types of engine failures.

[0055] The parameters required for user label profiling are transmitted to the cloud platform by the data stream recorded by the engine ECU through the T-BOX vehicle terminal. The corresponding data is downloaded internally by the enterprise to perform pseudo-damage calculations according to various reliability / functional failure models, and the damage data is dimensionless. Then, weighted calculations or direct sorting are performed, and users represented by 0.9 to 0.95 data are selected as representatives of this type of user labels. The failure characteristics they exhibit serve as input for subsequent engine bench test objectives and design objectives.

[0056] Although the present invention has been described in detail above with specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A vehicle user big data portrait analysis method based on a damage model, characterized in that: include: S1: Pre-build the vehicle's operating feature classification matrix, combine the vehicle's actual operating conditions, and extract features based on the classification matrix as a user label profile; S2: According to the failure scenario of the engine, damage classification, damage key factors and damage association parameters are defined, and damage association parameters are extracted according to the damage key factors matched by the user label portrait; S3: Quantitative calculation and cumulative distribution normalization are performed based on the collected key damage association parameters to obtain damage data. User label portraits and representative users representing different damage classifications of the engine are obtained based on the damage data sorting and clustering.

2. The vehicle user big data portrait analysis method based on damage model according to claim 1 is characterized in that: The operating characteristics of the vehicle in S1 include: user operation, external use environment, road characteristics, vehicle driving task profile, and load characteristics.

3. The vehicle user big data portrait analysis method based on damage model according to claim 1 is characterized in that: The definition of damage classification and damage association parameters in S2 include: defining damage classification and damage association parameters according to the reliability failure mechanism of engine parts and the failure mechanism of engine functions not meeting the requirements, and associating damage classification with user tags.

4. The vehicle user big data portrait analysis method based on damage model according to claim 1 is characterized in that: The quantitative calculation and cumulative distribution normalization processing are performed in S3, including the following steps: S301: Applying the component damage model formula, calculating the component failure pseudo damage data value according to the collected key damage association parameters; S302: performing cumulative distribution calculation on all data belonging to the same damage classification according to the calculated pseudo damage data value, and obtaining a cumulative distribution value; S303: define relative weights of different damage cumulative distribution values, and calculate damage weighting factors of different users as damage data.

5. A vehicle user big data portrait analysis system based on a damage model, using the method according to claims 1-4, characterized in that: It includes an engine ECU, a T-BOX vehicle terminal, a cloud platform, and a client computer. The engine ECU is connected to the T-BOX vehicle terminal by wired communication, and the cloud platform is connected to the T-BOX vehicle terminal and the client computer by wireless communication, wherein: The engine ECU is used to record the actual operating status data of the vehicle and transmit it to the cloud platform for storage through the T-BOX vehicle terminal; The client computer is used to obtain the vehicle operation status data stored in the cloud platform, and perform feature extraction according to the classification matrix to form a user label portrait; The client computer is also used to extract engine damage-related parameters according to the engine damage key factors matched by the user tag portrait; The client computer is also used to perform quantitative calculation and cumulative distribution normalization processing on the damage-related parameters to obtain damage data; The client computer is also used to sort and cluster the damage data to obtain user label portraits and representative users that characterize different types of engine failures.

Images