Vehicle full life cycle damage determination method and related equipment

Abstract

The invention discloses a vehicle full life cycle damage determination method and related equipment, and the method comprises the steps: obtaining the low-frequency and high-frequency operation data of a target vehicle, and dividing the operation times according to the operation parking time and mileage standard; based on the load, the operation area and the driving style, the operation trip is divided into a plurality of state spaces, and the mileage proportion of each state space is counted; load counting of the target part in each state space is determined, and unit mileage damage is calculated; performing distribution fitting on the low-frequency data to obtain an initial load boundary; for the high-frequency data, judging whether the number of state spaces exceeds a predetermined proportion, if so, merging the high-frequency data and the low-frequency data to fit to obtain a final load boundary, and otherwise, correcting the initial load boundary by using a similar state space deviation ratio; and finally, calculating the full-life-cycle damage of the target part according to the final load boundary. The invention aims to solve the problem of low evaluation reliability caused by neglecting user difference, rough working condition division and low-frequency data distortion in the existing load boundary calculation, and improve the accuracy of full-life-cycle damage evaluation.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of load spectrum compilation of users in the automobile industry, and particularly relates to a vehicle full-life-cycle damage determination method and related equipment. BACKGROUND

[0002] In the field of automobile design, performance evaluation and durability analysis, the load boundary refers to the threshold of the load or damage that a product bears under a specific working condition, and is used to represent the maximum expected load or damage level that the product may face under the working condition. It is a key indicator for evaluating the durability and reliability of the product. It is crucial to accurately obtain the load boundary covering a wide range of user groups, and the result is directly affected by the calculation method.

[0003] The current mainstream methods have obvious defects: the direct statistical fitting method mixes and fits the diversified working condition data of different users, ignoring the differences in load, driving area, driving habits, etc., resulting in poor representativeness of the load boundary; the single-dimensional division method only classifies through single dimensions such as load or driving area, ignoring key factors such as driving style, and the dimension is rough, with low calculation accuracy; the method of directly using high-frequency data to calculate the load boundary can accurately capture the high-frequency fluctuations of the load to improve the instantaneous accuracy of damage calculation, but it is limited by the high cost of high-frequency data acquisition and the narrow coverage, making it difficult to obtain large amounts of data of different users and different scenarios on a large scale; the method of directly applying low-frequency data to calculate the load boundary cannot capture high-frequency load fluctuations due to its low sampling rate, resulting in distorted damage calculation and failure to truly reflect the damage of components; the empirical formula estimation method relies too much on historical data and lacks pertinence, with poor applicability. These problems make the load boundary deviate from the actual working condition, and the reliability of the full-life-cycle damage evaluation is low, which cannot provide effective guidance for the fine design, cost optimization and maintenance strategy formulation of commercial vehicles. SUMMARY

[0004] In view of the problems in the prior art, the present application provides a vehicle full-life-cycle damage determination method and related equipment, which aims to solve the problem of low evaluation reliability caused by ignoring user differences, rough working condition division and distortion of low-frequency data in existing load boundary calculation, and to realize accurate association of the load boundary with the actual use conditions of users, and improve the accuracy of full-life-cycle damage evaluation.

[0005] To solve the above technical problems, the present application is implemented by the following technical scheme: According to a first aspect of the present application, a vehicle full-life-cycle damage determination method is provided, comprising: obtaining low-frequency running data and high-frequency running data of a target vehicle; wherein the low-frequency running data and the high-frequency running data are divided into a plurality of running laps, and each running lap is divided according to the standard that the vehicle runs for more than a predetermined time and the running mileage is more than a predetermined distance from running to stopping; For each trip in the low-frequency and high-frequency operation data, the trip is divided into multiple state spaces based on three dimensions: load status, operating area, and driving style, and the mileage percentage of each state space in the corresponding trip is calculated. For each run, determine the load count results of the target part in each state space, and calculate the unit mileage damage in each state space in each run based on the load count results and the mileage ratio. For low-frequency operating data, the distribution is fitted based on the damage per unit mileage to obtain the initial load boundary of the target part in each state space; For high-frequency operating data, determine whether the number of state spaces in each run under high-frequency operating data exceeds a predetermined proportion of the number of state spaces in each run under low-frequency operating data; if so, merge the unit mileage damage of each state space under high-frequency and low-frequency operating data and perform distribution fitting to obtain the final load boundary of the target part in each state space; if not, calculate the deviation rate of similar state spaces, and use the deviation rate to correct the initial load boundary to obtain the final load boundary of the target part in each state space; wherein, the similar state space is defined as: selecting a set of state spaces whose dimensions are most similar to those of each state space under high-frequency operating data from the state spaces under low-frequency operating data. Based on the final load boundary, calculate the full life cycle damage of the target part.

[0006] In one possible implementation of the first aspect, the mileage percentage of each state space in the corresponding running trip is calculated as follows:

[0007]

[0008]

[0009] In the formula, For the first The percentage of mileage in each state space within the corresponding running trip; For the first The mileage of each state space in the corresponding running trip; The total distance traveled for each trip; For the first j The speed of each sampling point; This represents the number of sampling points assigned to this state space. The sampling frequency for low-frequency or high-frequency operating data; This represents the number of sampling points for the corresponding running round.

[0010] In one possible implementation of the first aspect, determining the load count results of the target part in each state space includes: If the target part is a gear-type part, the TN counting method is used to determine the load counting result; If the target part is a shaft-type part, the rainflow counting method is used to determine the load counting result.

[0011] In one possible implementation of the first aspect, calculating the unit mileage damage in each state space for each run based on the load count result and the mileage percentage specifically includes: Based on the load count results of each state space in the corresponding running trip, calculate the total damage of each state space in the corresponding running trip; Calculate the running mileage of each state space in the corresponding running trip based on the mileage ratio of each state space in the corresponding running trip. By using the total damage of each state space in the corresponding running trip and the running mileage of each state space in the corresponding running trip, the unit mileage damage of each state space in the corresponding running trip is calculated.

[0012] In one possible implementation of the first aspect, the distribution fitting is performed using a Weibull distribution, the functional form of which is:

[0013] The initial load boundary or final load boundary is taken as the 95th quantile of the Weibull distribution and calculated as follows:

[0014] In the formula, The damage value per unit mileage shall not exceed the damage value per unit mileage. The probability of; This is the mathematical expression for the load boundary, i.e., the 95th quantile of the Weibull distribution; is the scale parameter of the Weibull distribution in the state space; Let be the shape parameter of the Weibull distribution in the state space.

[0015] In one possible implementation of the first aspect, the correction of the initial load boundary using the deviation rate specifically includes:

[0016]

[0017] In the formula, i State space numbering is defined as the state space numbering in a run under high-frequency operating data exceeding a predetermined proportion of the state space numbering in a run under low-frequency operating data.k a state space number for which a quantity of the state space in the operation trip under the high-frequency operation data is not more than a predetermined proportion of a quantity of the state space in the operation trip under the low-frequency operation data; a bias rate of the mth state space; i an initial load boundary of the target part in the mth state space; a final load boundary of the target part in the mth state space; i an initial load boundary of the target part in the mth state space; a final load boundary of the target part in the mth state space; i an initial load boundary of the target part in the mth state space. In a possible implementation manner of the first aspect, the full life cycle damage of the target part is calculated according to the final load boundary, and a formula is as follows: k k

[0018]

[0019] In the formula, D is the full life cycle damage of the target part; m is the mth state space; is an average mileage proportion of the mth state space in all the first operation trips; is the final load boundary of the target part in the mth state space; and n is the state space number.

[0020] In a possible implementation manner of the first aspect, the driving style includes a conservative type, a stable type and an aggressive type. The operation trip is divided into a plurality of state spaces based on the driving style dimension, and specifically includes the following steps. extracting a driving speed and a motor torque of the target vehicle in each operation trip; calculating a plurality of characteristic parameters according to the driving speed and the motor torque; performing cluster analysis on the plurality of characteristic parameters to obtain cluster centers corresponding to the conservative type, the stable type and the aggressive type; calculating Euclidean distances between the plurality of characteristic parameters and the cluster centers; classifying the operation trip into the conservative type, the stable type or the aggressive type according to the sizes of the Euclidean distances.

[0021] ​​​​​​​​​​According to a second aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the vehicle life cycle damage determination method when executing the computer program.

[0022] According to a third aspect of the present application, a computer readable storage medium is provided, which stores a computer program, wherein the computer program is executable on a processor to implement the vehicle life cycle damage determination method.

[0023] Compared with the prior art, the present application has at least the following beneficial effects: The present application fully considers three key dimensions of load state, running area and driving style, and finely divides each running trip in low-frequency and high-frequency running data. Compared with the prior art which directly statistically fits and ignores differences such as load and single dimension division dimension roughness, the present application can more comprehensively and accurately reflect the diversified working conditions of different users in actual use of the vehicle, so that the calculated load boundary truly represents the maximum expected load or damage level under the actual use conditions of the user, and greatly improves the representativeness of the load boundary. In the unit mileage damage calculation, according to the load count results and mileage proportion of the target part in each state space, the unit mileage damage of each state space in each running trip is accurately calculated, and the actual running conditions of different state spaces are comprehensively considered, avoiding the problems that the direct application of high-frequency data is difficult to obtain data on a large scale due to high acquisition cost and narrow coverage, and the direct application of low-frequency data is difficult to capture high-frequency load fluctuations due to low sampling rate, resulting in damage calculation distortion; in the initial load boundary determination, distribution fitting is performed on low-frequency and high-frequency running data respectively to obtain the initial load boundary of the target part in each state space, and in the final load boundary determination step, according to the number proportion relationship of running trips under high-frequency and low-frequency running data, if the number of high-frequency running trips exceeds a predetermined proportion, the unit mileage damage of each state space under the two types of data is combined for distribution fitting, fully utilizing the characteristics of high-frequency data accurately capturing high-frequency load fluctuations and low-frequency data covering a wide range, to obtain more accurate final load boundary, and if it does not exceed, the deviation rate of similar state spaces is calculated, and the initial load boundary is corrected by using the deviation rate, avoiding the problem that the empirical formula estimation method excessively relies on historical data and lacks pertinence, ensuring that the final load boundary is more accurate and reliable; since the present application can accurately determine the load boundary and accurately calculate the life cycle damage, the obtained result can truly reflect the damage of vehicle parts under different working conditions.

[0024] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the specific embodiments of the present application, the drawings needed to be used in the specific embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0026] Figure 1 Flow chart of a vehicle full life cycle damage determination method of the present application.

[0027] Figure 2 Flow chart of original data pass division and data preprocessing.

[0028] Figure 3 Flow chart of multi-dimensional state recognition. Figure 4 Flow chart of state space division and flow chart of commercial vehicle component damage calculation.

[0029] Figure 5 Flow chart of load boundary correction.

[0030] Figure 6 Result chart of each state space unit mileage damage before and after correction.

[0031] Figure 7 Full life cycle damage calculation result. DETAILED DESCRIPTION

[0032] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the present application will be described clearly and completely in conjunction with the drawings. Obviously, the described embodiments are some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0033] As shown in Figure 1 The present application provides a vehicle full life cycle damage determination method, which specifically comprises the following steps: S1. Data acquisition step: acquiring low-frequency operation data and high-frequency operation data of a target vehicle; The low-frequency operation data and the high-frequency operation data are divided into several operation passes respectively, and each operation pass is divided according to the standard that the vehicle runs more than a predetermined time and runs more than a predetermined distance from running to stopping.

[0034] Specifically, firstly, low-frequency operating data of the target vehicles is obtained from a big data platform. Low-frequency data has a low sampling frequency but a wide coverage, reflecting the actual operating conditions of a large number of users. To ensure data representativeness, the impact of the data volume on load boundaries needs to be analyzed to ensure a sufficiently large sample size so that the load boundaries tend to stabilize.

[0035] For example, the criteria for classifying a trip are: the vehicle transitions from a running state to a parked state, with a parking time exceeding half an hour and a running distance exceeding 10 kilometers. The running time for each trip is... and running mileage The calculation formula is as follows:

[0036]

[0037] In the formula: The running time for each run, in seconds; The distance traveled for each trip, in km; This represents the number of sampling points for the corresponding running round; For the first j The speed of each sampling point; The sampling frequency for low-frequency or high-frequency operating data.

[0038] After partitioning, a dataset of running trips containing information such as load, vehicle speed, and motor torque is obtained.

[0039] To correct for low-frequency data distortion, high-frequency operating data from actual vehicle tests is collected. High-frequency data has a high sampling frequency (e.g., above 100Hz), enabling more accurate capture of load fluctuations. The pass / flight division standard for high-frequency data is consistent with that for low-frequency data.

[0040] S2. State space partitioning step: For each trip in the low-frequency and high-frequency operating data, the trip is divided into multiple state spaces based on three dimensions: load status, operating area, and driving style, and the mileage ratio of each state space in the corresponding trip is calculated.

[0041] Specifically, for each run, the state space is divided along three dimensions: Under load conditions, the trips are divided into full load (≥70%) or half load (<70%) with 70% of the rated load as the threshold.

[0042] Operating areas are divided based on vehicle speed and continuous operating mileage. If the average vehicle speed of a segment is higher than 70 km / h and the continuous operating mileage exceeds 20 km, it is classified as a high-speed area; otherwise, it is a non-high-speed area.

[0043] Driving style: 15 feature parameters (such as average speed, acceleration, torque change rate, etc.) are extracted based on vehicle speed and motor torque data, and the driving style is divided into conservative, smooth and aggressive types by K-Means clustering algorithm. Specifically: extract the driving speed and motor torque of the target vehicle in each running trip; According to the driving speed and motor torque, calculate several feature parameters; Cluster analysis is performed on several feature parameters to obtain the cluster centers corresponding to conservative, smooth and aggressive types; Calculate the Euclidean distance of several feature parameters and the cluster center; According to the size of the Euclidean distance, the running trip is classified into conservative, smooth or aggressive type.

[0044] Through the combination of the above three dimensions, 12 state spaces (such as non-high-speed full-load conservative, high-speed half-load aggressive, etc.) are formed. The mileage proportion of each state space in the corresponding running trip is calculated as follows:

[0045]

[0046]

[0047] wherein, is the mileage proportion of the i-th state space in the corresponding running trip; is the mileage of the i-th state space in the corresponding running trip; is the total mileage of the running trip; is the speed of the i-th sampling point; is the number of sampling points divided into this state space; is the sampling frequency of low-frequency running data or high-frequency running data; j is the number of sampling points of the corresponding running trip. S3. Unit mileage damage calculation step: for each running trip, determine the load count result of the target part in each state space, and calculate the unit mileage damage of each state space in each running trip according to the load count result and the mileage proportion. Specifically, for different types of commercial vehicle parts, the corresponding load counting method is used:

[0048] Gear type parts use TN counting method, based on speed-torque interval statistics loading frequency;

[0049] Specifically, for different types of commercial vehicle parts, the corresponding load counting method is used: Gear type parts use TN counting method, based on speed-torque interval statistics loading frequency; ​The rainflow counting method is used for the shaft parts to identify the load cycles and count the amplitude-mean matrix.

[0050] In an implementation, according to the load counting result and the mileage proportion, the unit mileage damage of each state space in each running trip is calculated, specifically including: According to the load counting result of each state space in the corresponding running trip, the total damage of each state space in the corresponding running trip is calculated; According to the mileage proportion of each state space in the corresponding running trip, the running mileage of each state space in the corresponding running trip is calculated; The total damage of each state space in the corresponding running trip and the running mileage of each state space in the corresponding running trip are used to calculate the unit mileage damage of each state space in the corresponding running trip.

[0051] S4. Initial load boundary determination step: For low-frequency running data, distribution fitting is performed based on the unit mileage damage to obtain the initial load boundary of the target part in each state space; S5. Final load boundary determination step: whether the number of each state space in the running trip under the high-frequency running data exceeds the predetermined proportion of the number of each state space in the running trip under the low-frequency running data is determined; for example, the predetermined proportion can be one-third, that is, whether the number of running trips under the high-frequency running data exceeds one-third of the number of running trips under the low-frequency running data is determined, and then whether the high-frequency data is sufficient to correct the low-frequency result is determined.

[0052] If yes, the unit mileage damage of each state space under the high-frequency running data and the low-frequency running data is combined and distribution fitting is performed to obtain the final load boundary of the target part in each state space; If no, the deviation rate of the similar state space is calculated, and the deviation rate is used to correct the initial load boundary to obtain the final load boundary of the target part in each state space; wherein the similar state space is defined as: from each state space under the low-frequency running data, a group of state spaces with the most similar dimensions to each state space under the high-frequency running data is selected.

[0053] In an implementation, the distribution fitting uses Weibull distribution for fitting, and the function form of Weibull distribution is:

[0054] The initial load boundary or the final load boundary takes the 95th percentile of Weibull distribution, which is calculated as:

[0055] In the formula, The damage value per unit mileage shall not exceed the damage value per unit mileage. The probability of; This is the mathematical expression for the load boundary, i.e., the 95th quantile of the Weibull distribution; is the scale parameter of the Weibull distribution in the state space; Let be the shape parameter of the Weibull distribution in the state space.

[0056] S6. Lifecycle damage calculation steps: Calculate the lifecycle damage of the target part based on the final load boundary.

[0057] In one feasible approach, the life-cycle damage of the target part is calculated based on the final load boundary using the following formula:

[0058]

[0059] In the formula, Damage throughout the entire lifecycle of the target component; For the first The average mileage percentage of each state space in all first runs; For the target part in the first The final load boundary of each state space; Number the state space.

[0060] To explain the present invention more clearly, a more detailed description is provided below.

[0061] Step 1: Obtain low-frequency data from the big data platform.

[0062] 1.1 Impact of Data Volume. First, the load boundary under different data volumes was analyzed. By gradually increasing the number of data samples, the changing trend of the load boundary was observed. For example, sample data of different orders of magnitude (10, 50, 100, 200, 500, 1000, 2000, 3000, and 5000 sets) were selected, and the corresponding load boundary values ​​were calculated. Analysis revealed that as the data volume increased, the load boundary gradually stabilized. When the sample size reached 1000 sets, further increases in sample size resulted in load boundary changes controlled within 3%, indicating that when the data volume reaches a certain scale, the load boundary has good representativeness. Based on the above research results, a large-scale sampling method was used on a big data platform to obtain a sufficiently large sample of low-frequency data. During the sampling process, it was ensured that the data covered different operating conditions, vehicle types, and other factors to guarantee the comprehensiveness and representativeness of the data.

[0063] 1.2 Determine the criteria for trip division. In a continuous operational data set along the time dimension, if there exists a segment where the vehicle speed changes from 0 km / h to 0 km / h, and the stopping time exceeds half an hour (i.e., the vehicle transitions from a stopped state to an operational state and back to a stopped state), and the distance traveled in the operational state exceeds ten kilometers, and the stopping time from the stopped state to the next operational state exceeds half an hour, then these segments are divided into a new trip. All acquired low-frequency data are traversed, and based on the vehicle's operational and stopping time information, the data is divided into multiple operational trips, forming dataset 1. The formulas for calculating operational time and operational mileage are as follows:

[0064]

[0065] In the formula: The running time for each run, in seconds; The distance traveled for each trip, in km; ; For the first j The speed of each sampling point; The sampling frequency for low-frequency or high-frequency operating data.

[0066] Ensure that dataset 1 contains rich information, such as load information, real-time vehicle speed, and motor torque, as key data to serve as the basis for subsequent state space partitioning and damage calculation.

[0067] Step 2, low-frequency data preprocessing is as follows: 2.1 When removing outliers, statistical methods (such as the 3σ criterion) are used to detect outliers in the collected low-frequency data. For each data feature (such as load, vehicle speed, motor torque, etc.), its mean and standard deviation are calculated. Data exceeding the mean ± 3 times the standard deviation are identified as outliers and removed from the dataset. That is, outliers are removed from the dataset. The range of data, where x represents the original data. This represents the mean of the data set. This indicates the variance of the data set.

[0068] 2.2 Idle Data Removal: Define the criteria for judging idling state: when the vehicle speed is below a certain threshold (e.g., 0.1 km / h) and the duration exceeds a certain duration (e.g., 10 seconds), it is judged as idling state. For each data point in the dataset, the data belonging to the idling state are judged and removed.

[0069] 2.3 Since the sampling frequency of the acquired low-frequency data is not uniform, it is necessary to resample the acquired data to the same frequency. A suitable resampling frequency (the uniform frequency selected in this embodiment is 1 Hz) is selected, and the data is resampled using a linear interpolation method to make all the data have a uniform time interval. Ensure that the resampled data can accurately reflect the characteristics of the original data and meet the requirements of subsequent calculations. The segment division and data processing flowchart is shown in Figure 2

[0070] Step 3, user state space division is as follows: 3.1 Load distribution, analyze the vehicle load information in the trip data, and determine the full load division standard. In this embodiment, according to the rated load of the vehicle, the load reaching 70% or more of the rated load is defined as full load, and less than 70% is defined as half load. The load information of each trip data is evaluated, and different trips are divided into full load or half load state. That is:

[0071] In the formula: is the load information of each trip, is the rated load of the vehicle.

[0072] 3.2 Running area distribution, analyze the vehicle speed information in the trip data, determine the division threshold of high-speed and non-high-speed running area, define the segment with vehicle speed higher than 70km / h and continuous running mileage more than 20km as high-speed running area, otherwise as non-high-speed running area. According to the above threshold, the data of each trip is divided into high-speed or non-high-speed running area. That is:

[0073] In the formula: and are the running mileage and running time of the i th segment of the j th trip.

[0074] 3.3 Driving style distribution, extract the vehicle speed and motor torque, calculate the relevant feature parameters of each trip, including "average speed", "maximum speed", "speed standard deviation", "average acceleration in acceleration section", "average deceleration in deceleration section", "maximum acceleration", "maximum deceleration", "acceleration absolute value standard deviation", "average motor torque", "maximum motor torque absolute value", "motor torque absolute value standard deviation", "high load proportion", "speed-torque correlation", "number of rapid accelerations per kilometer", "number of rapid decelerations per kilometer", a total of 15 characteristic parameters. The characteristic value calculation method is as follows: Average speed: ​

[0075] Maximum speed:

[0076] Speed standard deviation:

[0077] Average acceleration of acceleration section:

[0078] Average deceleration of deceleration section:

[0079] Maximum acceleration:

[0080] Maximum deceleration:

[0081] Acceleration absolute value standard deviation:

[0082] Average motor torque:

[0083] Maximum motor torque absolute value:

[0084] Motor torque absolute value standard deviation:

[0085] High load proportion:

[0086] Speed-torque correlation:

[0087] Number of rapid accelerations per kilometer:

[0088] Number of rapid decelerations per kilometer:

[0089] wherein, represents average speed, n is the number of sampling points corresponding to the running trip, is the speed of the j th sampling point in the corresponding trip; is the maximum speed, represents the number of sampling points in the acceleration state, represents the number of sampling points in the deceleration state, represents the acceleration of each sampling point, represents the average value of the acceleration absolute value, is the motor torque of each sampling point, is the average value of the motor torque absolute value, The number of sampling points representing the motor torque greater than 0.8 times the maximum motor torque, X is the speed, Y represents the absolute value of the torque, is the covariance of the speed and torque; , is the standard deviation of the speed and torque, respectively, , corr is close to 1 when the speed and the absolute value of the torque are strongly positively correlated (when the speed increases, the absolute value of the torque tends to increase), close to -1 when the speed and the absolute value of the torque are strongly negatively correlated (when the speed increases, the absolute value of the torque tends to decrease). Close to 0, there is no linear correlation. 、 represent the number of sampling points with acceleration greater than 1.5 m / s 2 and deceleration less than -1.5 m / s 2 , and L represents the driving distance of the trip.

[0090] After the calculation, the K-Means clustering algorithm is used for clustering analysis of the feature parameters. First, the number of clusters is determined to be divided into three categories (corresponding to conservative, stable and aggressive driving styles). Then a large amount of data is selected to train the K-Means algorithm to obtain three cluster centers that meet the requirements. Before clustering training, the calculated feature values of each trip are standardized to avoid the scale of each feature value being different, which causes the clustering effect to be unsatisfactory. The standardization formula is as follows:

[0091] Among them: d represents the serial number of the feature value, is the standardized feature value, is the feature value before standardization, is the mean of the feature value, is the standard deviation of the feature value.

[0092] According to the standardized feature values, three cluster centers are trained, and the training centers are divided into three types according to the size of the average speed, the number of average acceleration per unit distance and the torque change rate three-dimensional information, represented by G k ( k =1, 2, 3) indicates, representing the feature vectors of the three driving styles. Then the feature vectors of the three cluster centers and the mean and variance of each training set feature value are saved. Then the feature values of each trip data are standardized according to the mean and variance of the training set. Then the K-Means algorithm is run according to the three cluster centers, and each trip is divided into three categories according to the size of the Euclidean distance from each cluster center, i.e. the shortest Euclidean distance from each segment to the three cluster centers is divided into this category. According to the clustering result, different driving segments are divided into three driving styles: conservative, stable and aggressive. The formula for calculating the Euclidean distance is as follows:

[0093]

[0094] wherein, represents the number of selected characteristic parameters, in the embodiment is 15; e represents the number of each pass; ∈G k represents the characteristic element in the three cluster centers, k =1, 2, 3; represents the b th characteristic value of the e th pass; represents the Euclidean distance from the b th cluster center in the k th pass, represents the b th cluster center in the k th pass has the smallest Euclidean distance. The driving style classification flow chart is shown in Figure 3 .

[0095] In the driving style classification, only through the basic data of vehicle speed, motor speed and torque easily obtained during vehicle driving, combined with K-Means clustering of 15 characteristic parameters, the accurate classification of conservative type, stable type and aggressive type is realized. Compared with the traditional method which needs to collect multi-dimensional complex data such as steering angle, pedal opening degree and vehicle body posture (and online big data often lacks such information due to sensor limitations), the method significantly reduces the hardware dependence and information integrity requirements of data collection, ensures the accuracy of classification, and is more suitable for the scene of limited data collection conditions in commercial vehicle actual operation, enhances the engineering landing nature of the method and the feasibility of online big data analysis.

[0096] 3.4 By combining the three dimensions of load state (full load, half load), running area (high speed, non-high speed) and driving style (conservative type, stable type, aggressive type), each running pass of the user is divided into 12 state spaces, which are non-high speed full load conservative, non-high speed full load stable, non-high speed full load aggressive, non-high speed half load conservative, non-high speed half load stable, non-high speed half load aggressive, high speed full load conservative, high speed full load stable, high speed full load aggressive, high speed half load conservative, high speed half load stable, high speed half load aggressive. The mileage proportion of each state space in the corresponding pass is counted.

[0097]

[0098]

[0099]

[0100] In the formula, For the first The percentage of mileage in each state space within the corresponding running trip; For the first The mileage of each state space in the corresponding running trip; The total distance traveled for each trip; For the first j The speed of each sampling point; This represents the number of sampling points assigned to this state space. The sampling frequency for low-frequency or high-frequency operating data; This represents the number of sampling points for the corresponding running iterations. The state space partitioning flowchart is as follows: Figure 3 As shown.

[0101] Step 4, Load Counting and Damage Calculation, are as follows: 4.1 For each trip's data, load counting is performed based on information such as speed and torque. For gear parts in commercial vehicles, the TN counting method is used. TN counting is a time-based counting method that can accurately capture load changes in gears during operation. The specific principle is as follows: the preprocessed time-domain data is divided into several small intervals (e.g., 128×128 levels) according to the target speed-torque interval. The load frequency between the i-th intervals is... Calculated by integrating rotational speed over time:

[0102]

[0103] In the formula: For the first i OK j The frequency of loading between the speed-torque zones of the column; s This represents the number of torque segments within this small interval; For the first i OK j The time intervals corresponding to the smaller intervals in the column; The rotational speed at time t (unit: r / s); N This represents the final TN counting matrix; v This represents the number of speed and torque levels.

[0104] For shaft components, the rainflow counting method is employed. Rainflow counting is a commonly used fatigue load counting method that can effectively count the cyclic loads borne by shaft components. The technical principle is as follows: it converts the continuous load time history into a peak-valley sequence. ,in For the firsti The cycle recognition rule is: take three consecutive peak-valley values from the load sequence in turn 、 、 If the following conditions are met:

[0105] A complete cycle is identified, and its amplitude A and mean M are respectively: The corresponding points of the cycle are removed from the sequence, and the remaining data is processed. If there are remaining points after the load sequence is processed, they are combined into a half cycle, which is usually counted as 0.5 complete cycle in fatigue calculation.

[0106] Finally, the cycle count matrix RFM(A, M) is generated, indicating the number of cycles with amplitude A and mean M, and the rainflow count matrix is obtained:

[0107] Any rainflow matrix element r ij represents the rainflow cycle frequency from load level S i to S j ; RFM represents the rainflow count matrix; w represents the mean and amplitude division order. Through the above two different counting methods, the corresponding count matrix is obtained, which reflects the number of times different load levels appear.

[0108] 4.2 According to the obtained count matrix and the running mileage of the vehicle in each state space, the unit mileage damage result of each state space in each trip is calculated using the pseudo-damage theory. The pseudo-damage theory is a method of converting load cycles into damage, by calculating the corresponding damage value of each load cycle in the count matrix, and then accumulating all damage values and dividing by the total mileage to obtain the unit mileage damage result.

[0109]

[0110]

[0111] In the formula: d is the total damage, is the number of cycles under a certain load, represents the maximum number of cycles that the part can cycle under that load, represents the unit mileage damage. The damage calculation flowchart of commercial vehicle parts is shown in Figure 4 .

[0112] Step 5, the Weibull fitting of low-frequency data is as follows: 5.1 Extract the unit mileage damage calculation results of each state space for each trip, and integrate the unit mileage damage results of each state space to form the damage dataset of that state space.

[0113] 5.2 The Weibull distribution is used to fit the damage dataset for each state space. The Weibull distribution is a commonly used reliability distribution that can well describe the fatigue life distribution of materials.

[0114] The Weibull distribution parameters (such as shape parameters) for each state space are obtained by fitting. k and scale parameters This process determines the load boundary (D95) that meets the requirements of the state space. D95 refers to the damage value that 95% of the samples in the Weibull distribution do not exceed; it represents the maximum expected damage in that state space. The core of the two-parameter Weibull distribution is the cumulative distribution function (CDF). The CFD definition formula and the D95 calculation formula are as follows:

[0115]

[0116] in:, The damage value per unit mileage shall not exceed the damage value per unit mileage. The probability of; This is the mathematical expression for the load boundary, i.e., the 95th quantile of the Weibull distribution; is the scale parameter of the Weibull distribution in the state space; Let be the shape parameter of the Weibull distribution in the state space.

[0117] Right now, For random variables The probability of; >0 represents shape parameters (which determine the distribution pattern, such as whether it is symmetrical or skewed); >0 represents the scale parameter (which determines the "expansion" of the distribution); Step 6, High-Frequency Data Acquisition: (Details follow) 6.1 Load cycle counting and fatigue damage were studied at different sampling rates. Data at different sampling rates (e.g., 10Hz, 20Hz, 50Hz) were obtained by resampling the original high-frequency data. Load cycle counting and fatigue damage calculations were performed on the data at different sampling rates. Comparison of the results showed that the low-sampling-rate data had poor consistency with the original high-frequency data in terms of load cycle counting and fatigue damage. Based on the above research results, it was determined that high-frequency data from actual vehicle tests (dataset 2) needed to be collected to correct the results of the low-frequency data.

[0118] 6.2 Based on the state space partitioning results, develop a real vehicle road collection scheme from the aspects of operating area, load, etc. Determine the high-frequency data collection amount of each state space, and try to ensure that high-frequency data can cover every state space. For example, for each state space, a certain number of high-frequency data samples are collected to ensure the accuracy of the correction.

[0119] Step 7, the high-frequency data processing is as follows: Repeat steps 2 to 4 for sufficient high-frequency data, including outlier rejection, idle data rejection, and resampling (here, resampling is no longer sampled to 1 Hz, but high-frequency data of different frequencies are resampled to high-frequency data of the same frequency, such as 100, 200, 512 Hz, etc. Different frequencies of high-frequency data are resampled to 100 Hz. ) and other preprocessing operations, as well as state space partitioning, load counting, and damage calculation. Finally, the unit mileage damage results of high-frequency data in each state space are obtained.

[0120] Step 8, using high-frequency data results to correct load boundary is as follows: 8.1 For state spaces with sufficient high-frequency data, combine the low-frequency data and high-frequency data damage results of the state space. Perform Weibull distribution fitting on the combined data set to obtain the corrected load boundary of the state space.

[0121] 8.2 For state spaces not covered by high-frequency data or insufficient high-frequency data to correct the results, use the deviation rate of similar state spaces for correction. Similar state spaces are defined as: two state spaces differ as little as possible in load, driving style, and operating area. If two state spaces are the same in load and driving style, they are considered similar. Calculate the load boundary deviation rate of similar state spaces before and after correction, and then apply the deviation rate to the load boundary of the uncovered state space to obtain the correction result. The deviation rate calculation formula and damage correction formula are as follows:

[0122]

[0123] In the formula, i is the state space number whose state space quantity in the high-frequency running data exceeds the predetermined proportion of the state space quantity in the low-frequency running data; k is the state space number whose state space quantity in the high-frequency running data does not exceed the predetermined proportion of the state space quantity in the low-frequency running data; is the deviation rate of the i th state space; is the deviation rate of the target part in thei the final load boundary of the target part in the first state space; the initial load boundary of the target part in the first state space; i the final load boundary of the target part in the first state space; the initial load boundary of the target part in the first state space; k the final load boundary of the target part in the first state space; the initial load boundary of the target part in the first state space. k the initial load boundary of the target part in the first state space.

[0124] The load boundary correction flow chart and the unit mileage damage result chart of each state space before and after correction are shown in Figs. Figure 5 and Figure 6 .

[0125] Step 9, the full life cycle damage is calculated according to the correction result, which is specifically as follows: 9.1 Because the low-frequency data have a large number of characteristics, the proportion of each state space is statistically representative, so the average proportion of each state space is calculated according to the proportion of each state space of the low-frequency data calculated in step three. For example, the mileage proportion of each state space in multiple trips is averaged to obtain the average proportion of the state space in the full life cycle.

[0126] 9.2 Knowing that the design mileage of a commercial vehicle is 2 million kilometers, the full life cycle mileage is multiplied by the average proportion of each state space and the unit mileage damage.

[0127] The calculation results of each state space are added to obtain the full life cycle damage. The full life cycle damage reflects the total damage of the vehicle in the entire design service life and is an important indicator for evaluating the reliability and durability of the vehicle. The calculation formula is as follows:

[0128]

[0129] In the formula, is the full life cycle damage of the target part; is the average mileage proportion of the first state space in all first running trips; is the final load boundary of the target part in the first state space; is the final load boundary of the target part in the first state space; is the final load boundary of the target part in the first state space. is the state space number.

[0130] The target part involved in the present application has a comparison chart of unit mileage damage before and after correction in each state space and a full life cycle damage result distribution chart, which are shown in Figs. Figure 6 and Figure 7The significant difference of damage results under different state spaces shows that the method can effectively represent the significant influence of multi-dimensional influencing factors such as vehicle operation area, load, and driving style on the damage of parts, and further verifies the effectiveness and feasibility of the application in the damage calculation and load boundary correction link.

[0131] The application guarantees the accuracy of load boundary calculation in all aspects from data acquisition to result correction through a multi-dimensional and systematic processing flow. In the data acquisition stage, not only the influence of data quantity on the representativeness of load boundary is studied in depth, but also typical low-frequency data of a large enough sample are obtained by means of a big data platform, so that the data are more universal and representative. In the state space division, three dimensions of load, operation area and driving style are innovatively divided in detail, and 12 state spaces are obtained. This multi-dimensional division method can more comprehensively and carefully reflect the actual operation conditions of the vehicle, avoiding the calculation errors caused by the rough division of working conditions in the traditional method. In addition, different counting methods are selected for different types of commercial vehicle parts in the calculation process. TN counting is used for gear parts, and rainflow counting is used for shaft parts. This accurate and adaptive counting method further improves the accuracy of load counting, so that the subsequent damage calculation and load boundary determination are more accurate. Through these measures, the application effectively solves the problem of insufficient accuracy of load boundary calculation in the prior art.

[0132] The application fully considers various complex working conditions in the vehicle operation process and realizes comprehensive coverage of different scenes. In terms of data acquisition, not only low-frequency data are obtained, but also high-frequency data of real vehicle testing are collected for correction in view of the poor consistency of low-frequency data in load cycle counting and fatigue damage, so as to ensure that the data can reflect the real load of the vehicle under different operating conditions. In the state space division, three key dimensions of load, operation area and driving style are classified, covering the operating conditions of the vehicle under different load conditions, different running speed intervals and different driving habits. For example, the load state includes different conditions such as full load and half load, the operation area is divided into high speed and non-high speed, and the driving style has conservative, smooth and aggressive types. This multi-dimensional and fine-grained division method enables the application to accurately capture the load characteristics of the vehicle under various complex working conditions, providing comprehensive and detailed basic data for subsequent load boundary calculation and full life cycle damage assessment. In contrast, the traditional method can only consider a single or a few working conditions, and it is difficult to comprehensively reflect the complex situation in the actual operation of the vehicle. The application has obvious advantages in this regard.

[0133] The present application greatly improves the reliability of the whole life cycle damage assessment. First, by accurate load boundary calculation, the load boundary (D95) of each state space is determined, providing accurate basic data for damage assessment. Second, when calculating the whole life cycle damage, the proportion and unit mileage damage of each state space are fully considered, and the method of multiplying the whole life cycle mileage by the average proportion and unit mileage damage of each state space is adopted to add the calculation results of each state space to obtain the total damage. This method can comprehensively and accurately reflect the damage accumulation of the vehicle in the whole life cycle. In addition, through the correction of low-frequency data by high-frequency data, the accuracy of damage assessment is further improved. High-frequency data can more accurately reflect the load fluctuation and damage of the vehicle in actual operation, and through the correction of low-frequency data, the whole life cycle damage assessment result is more reliable. Compared with the traditional method, the whole life cycle damage assessment result of the present application can provide stronger support for product design optimization, maintenance plan formulation, etc., help enterprises improve product quality and reliability, and reduce operating costs.

[0134] In another embodiment of the present application, a computer device is provided, which includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in the computer storage medium to implement a corresponding method flow or a corresponding function; the processor in the embodiment of the present application can be used for the operation of a whole life cycle damage determination method of a vehicle.

[0135] In still another embodiment of the present application, the present application also provides a storage medium, specifically a computer readable storage medium (Memory), which is a memory device in a computer device, used for storing programs and data. It can be understood that the computer readable storage medium here can include the built-in storage medium in the computer device, and of course can also include the extended storage medium supported by the computer device. The computer readable storage medium provides a storage space, which stores the operating system of the terminal. Moreover, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium here can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory. One or more instructions stored in the computer readable storage medium can be loaded and executed by the processor to realize the corresponding steps of the vehicle full life cycle damage determination method in the above embodiment.

[0136] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can 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-ROMs, optical storage devices, etc.) containing computer usable program code.

[0137] The present application is described with reference to the flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device implemented in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0138] These computer program instructions can also be stored in a computer readable memory that can direct the computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction apparatus, which realizes the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocksFigure 1 the function specified in the one or more blocks.

[0139] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide a process for implementing the flowchart Figure 1 the flowchart or flowcharts and / or a block Figure 1 the steps of the function specified in the one or more blocks.

[0140] The present application also provides a computer program product, which is used for executing any one of the vehicle full life cycle damage determination methods described above. Since the computer program product provided by the present application belongs to the same inventive concept as the vehicle full life cycle damage determination method described above, the computer program product provided by the present application has all the advantages of the vehicle full life cycle damage determination method described above, and therefore the beneficial effects of the computer program product provided by the present application will not be described one by one here.

[0141] In the present application, the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the skilled person in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0142] Finally, it should be noted that the above-described embodiments are merely specific embodiments of the present application, which are used to illustrate the technical solutions of the present application, rather than limit the same. The protection scope of the present application is not limited thereto, and although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can make modifications or easily think of changes to the technical solutions recorded in the foregoing embodiments within the technical range disclosed by the present application, or make equivalent replacements to some of the technical features; and these modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application.

Claims

1. A method for determining vehicle lifecycle damage, characterized in that, include: The low-frequency and high-frequency operating data of the target vehicle are acquired; wherein the low-frequency and high-frequency operating data are each divided into several operating trips, and each operating trip is divided according to the standard that the vehicle's running time from start to stop exceeds a predetermined time and the running distance exceeds a predetermined distance; For each trip in the low-frequency and high-frequency operation data, the trip is divided into multiple state spaces based on three dimensions: load status, operating area, and driving style, and the mileage percentage of each state space in the corresponding trip is calculated. For each run, determine the load count results of the target part in each state space, and calculate the unit mileage damage in each state space in each run based on the load count results and the mileage ratio. For low-frequency operating data, the distribution is fitted based on the damage per unit mileage to obtain the initial load boundary of the target part in each state space; For high-frequency operating data, determine whether the number of state spaces in each run under high-frequency operating data exceeds a predetermined proportion of the number of state spaces in each run under low-frequency operating data; if so, merge the unit mileage damage of each state space under high-frequency and low-frequency operating data and perform distribution fitting to obtain the final load boundary of the target part in each state space; if not, calculate the deviation rate of similar state spaces, and use the deviation rate to correct the initial load boundary to obtain the final load boundary of the target part in each state space; wherein, the similar state space is defined as: selecting a set of state spaces whose dimensions are most similar to those of each state space under high-frequency operating data from the state spaces under low-frequency operating data. Based on the final load boundary, calculate the full life cycle damage of the target part.

2. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The mileage percentage of each state space in the corresponding running trip is calculated as follows: In the formula, For the first The percentage of mileage in each state space within the corresponding running trip; For the first The mileage of each state space in the corresponding running trip; The total distance traveled for each trip; For the first j The speed of each sampling point; This represents the number of sampling points assigned to this state space. The sampling frequency for low-frequency or high-frequency operating data; This represents the number of sampling points for the corresponding running round.

3. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The determination of the load count results of the target part in each state space includes: If the target part is a gear-type part, the TN counting method is used to determine the load counting result; If the target part is a shaft-type part, the rainflow counting method is used to determine the load counting result.

4. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The step of calculating the unit mileage damage in each state space for each running trip based on the load count results and the mileage percentage specifically includes: Based on the load count results of each state space in the corresponding running trip, calculate the total damage of each state space in the corresponding running trip; Calculate the running mileage of each state space in the corresponding running trip based on the mileage ratio of each state space in the corresponding running trip. By using the total damage of each state space in the corresponding running trip and the running mileage of each state space in the corresponding running trip, the unit mileage damage of each state space in the corresponding running trip is calculated.

5. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The distribution fitting uses the Weibull distribution, whose functional form is: The initial load boundary or final load boundary is taken as the 95th quantile of the Weibull distribution and calculated as follows: In the formula, The damage value per unit mileage shall not exceed the damage value per unit mileage. The probability of; This is the mathematical expression for the load boundary, i.e., the 95th quantile of the Weibull distribution; is the scale parameter of the Weibull distribution in the state space; Let be the shape parameter of the Weibull distribution in the state space.

6. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The method of correcting the initial load boundary using the deviation rate specifically involves: In the formula, i State space numbering is defined as the state space numbering in a run under high-frequency operating data exceeding a predetermined proportion of the state space numbering in a run under low-frequency operating data. k State space numbering is a predetermined proportion of the number of state spaces in a run under high-frequency operating data, which does not exceed the number of state spaces in a run under low-frequency operating data. For the first i Deviation rate of each state space; For the target part in the first i The final load boundary of each state space; For the target part in the first i Initial load boundaries of each state space; The target part is in k The final load boundary of each state space; The target part is in k The initial load boundary of each state space.

7. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, Based on the final load boundary, the life-cycle damage of the target part is calculated using the following formula: In the formula, Damage throughout the entire lifecycle of the target component; For the first The average mileage percentage of each state space in all first runs; For the target part in the first The final load boundary of each state space; Number the state space.

8. The method for determining vehicle lifecycle damage according to claim 1, characterized in that, The driving styles include conservative, steady, and aggressive. The division of a running trip into multiple state spaces based on driving style dimension specifically includes: Extract the target vehicle's speed and motor torque for each trip; Based on the driving speed and motor torque, calculate several characteristic parameters; Cluster analysis was performed on several of the aforementioned feature parameters to obtain the cluster centers corresponding to conservative, stationary, and radical types; Calculate the Euclidean distance between several of the aforementioned feature parameters and the cluster centers; Based on the magnitude of the Euclidean distance, the number of runs can be categorized into conservative, steady, or aggressive types.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a method for determining vehicle lifecycle damage as described in any one of claims 1 to 5.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a method for determining vehicle lifecycle damage as described in any one of claims 1 to 5.

Images