A data processing method and system for typical driving cycle construction

Abstract

The present application relates to the technical field of driving cycle, and more particularly to a data processing method and system for constructing a typical driving cycle. The method comprises determining a sampling time state based on actual driving cycle data to obtain a Markov state transition matrix; combining the Markov state transition matrix with a Monte Carlo algorithm to obtain a suboptimal driving cycle set; determining a reachable domain, and obtaining a candidate driving cycle set within the reachable domain by using the Monte Carlo algorithm; resampling to obtain a candidate subsegment of the candidate driving cycle set; constructing a better driving cycle set, and selecting an iterative subsegment of the better driving cycle set; obtaining a candidate bridge path set by using the Monte Carlo algorithm to randomly sample within the reachable domain of the iterative subsegment; replacing the iterative subsegment with the candidate bridge path set, and obtaining an optimal driving cycle set after repeating the iteration. The present application can simultaneously achieve a wide range of search, fast generation and good feature retention.

Description

Technical Field

[0001] This invention relates to the field of driving cycle technology, and in particular to a data processing method and system for constructing a typical driving cycle. Background Technology

[0002] A typical driving cycle is a standardized and repeatable speed-time curve used on a laboratory bench to simulate a vehicle driving in the real world; a typical driving cycle can reflect the driving patterns of a vehicle under different operating conditions.

[0003] Constructing typical driving cycles is crucial for accurately assessing vehicle energy consumption and emission characteristics, and optimizing energy-saving control strategies. However, existing methods are generally hampered by the inherent contradiction between exploration breadth, computational efficiency, and disturbance suppression, making it difficult to simultaneously achieve large-scale search, rapid generation, and preservation of excellent features.

[0004] To address the aforementioned issues, there is an urgent need for a data processing method capable of generating high-quality typical driving cycles, thereby providing more reliable and accurate data support for vehicle energy consumption assessment. Summary of the Invention

[0005] In view of the above problems, this disclosure provides a data processing method and system for constructing typical driving cycles that overcomes or at least partially solves the above problems. The purpose is to resolve the inherent contradiction between the breadth of exploration, computational efficiency and disturbance suppression in existing methods, and can simultaneously achieve large-scale search, rapid generation and preservation of excellent features.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A first aspect of the present invention provides a data processing method for constructing a typical driving cycle, comprising the following steps:

[0008] S1. Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix;

[0009] S2. Based on the Markov state transition matrix and Monte Carlo algorithm, the set of suboptimal driving cycles is obtained;

[0010] S3. Based on the suboptimal driving cycle set and the total step size, determine the reachable region, and obtain the candidate driving cycle set within the reachable region using the Monte Carlo algorithm;

[0011] S4. Based on the candidate driving cycle set, resample within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set;

[0012] S5. Construct a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and select the iterative sub-segments of the better driving cycle set;

[0013] Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm;

[0014] S6. Replace the iterative sub-segment with the candidate bridging path set, and repeat the iteration to obtain the preferred driving loop set.

[0015] Furthermore,

[0016] Based on actual driving cycle data, the state at each sampling time is determined, and the Markov state transition matrix is ​​obtained, including the following steps:

[0017] S11. Determine the sampling time state: Divide the actual driving cycle data into several sampling time states, denoted as... ; The set of all sampled states is represented as ;

[0018] S12, Statistical Transition Count: Count the number of transitions between two adjacent sampling times. Transferred to The number of times it appears is denoted as ;

[0019] S13. Obtain the Markov state transition matrix: For each state in the set at the sampling time... Transferred to The Markov transition matrix is:

[0020] ;

[0021] By constructing the Markov state transition matrix, we can obtain the current state. Under the condition that, the state transitions to the next time step. The probability; where the actual driving cycle data includes speed and acceleration.

[0022] Furthermore,

[0023] Based on the Markov state transition matrix and the Monte Carlo algorithm, the set of suboptimal driving cycles is obtained, including the following steps:

[0024] S21. Generate the initial driving cycle set: Based on the Markov state transition matrix, generate several initial driving cycle sets using the Monte Carlo algorithm. ;

[0025] S22. Calculate the characteristic deviation: Calculate the i-th dimension statistical characteristic of each driving cycle in the initial driving cycle set. Compared with actual driving cycle reference characteristics deviation and the sum of deviations ;

[0026] in,

[0027] ;

[0028] S23. Obtain the set of suboptimal driving cycles: Sort the initial set of driving cycles by size d to obtain the set of suboptimal driving cycles, denoted as: .

[0029] The Monte Carlo algorithm simulates the driving cycle through random sampling. Using the established Markov state transition matrix, it randomly samples according to the Markov state transition probabilities to randomly sample a completely new driving speed-time curve.

[0030] Furthermore,

[0031] Based on the set of suboptimal driving cycles and the total step size, the reachable region is determined, including:

[0032] S31. Set the initial state: Based on the suboptimal driving cycle set, set the starting state. and endpoint state Set the total step size ;

[0033] S32. Constructing the divergence domain Determine the starting state ,exist Reaching the final state within steps The set of states at all sampling times;

[0034] Recorded as: ;

[0035] S33. Constructing the convergence region : Determine the state from the endpoint ,exist Reaching the starting state within one step The set of states at all sampling times;

[0036] Recorded as: ;

[0037] S34. Construct reachable domains Divergence domain With convergence region The intersection of these two domains is denoted as the reachable region. ;

[0038] Based on its bidirectional connectivity, it can determine the reachable domain under fixed step size constraints in milliseconds, significantly reducing the high computational time caused by blindly searching the entire state space and ensuring a sufficient search for potential optimal solutions. At the same time, based on the suboptimal driving cycle set and the total step size, a massive number of candidate segments can be obtained in the reachable domain.

[0039] Furthermore,

[0040] Based on the candidate driving cycle set, resampling is performed within the set of states at the sampling time to obtain candidate segments of the candidate driving cycle set, including:

[0041] Construct a state lookup table based on the state at the sampling time. When the order of the candidate driving cycle set remains unchanged, several resample sets are divided, and a temporary state lookup table is obtained based on each resample set. Based on the candidate driving cycle set and the temporary state lookup table The driving cycle characteristics were obtained. ; Calculate the generated driving cycle features and its reference features Total deviation ;like If the sum of deviations d is less than the sum of deviations d of the candidate driving cycle set, then... Updated to This serves as a new state lookup table; then, candidate paths, or candidate segments, are generated within the reachable domain using the Monte Carlo algorithm.

[0042] By exploring the reachable domain more effectively , This represents the velocity-acceleration values; thus, without changing the Markov chain structure of the driving cycle, it enables fine calibration of the statistical characteristics of the driving cycle, avoiding falling into local optima, and achieving the goal of high-precision fitting of target features.

[0043] Furthermore,

[0044] The state lookup table Assign a specific physical quantity to the state at each discrete sampling time. , This indicates the velocity and acceleration values ​​corresponding to the state at that sampling moment;

[0045] State lookup table A subset of the velocity is resampled to obtain a resampled set, and each resampled set contains a new velocity. and acceleration Combined into a temporary state lookup table .

[0046] Furthermore,

[0047] Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm, including:

[0048] S51. Constructing a set of better driving cycles: Merge candidate segments into the corresponding candidate driving cycle sets to obtain a set of better driving cycles, denoted as: ,in Indicates the first In the next iteration, M is the number of driving loops contained in the set;

[0049] S52. Constructing an iterative segment set: In each driving cycle Random selection Non-overlapping iterative subsegments , recorded as Set the start-up state and terminal state They are respectively recorded as:

[0050] ;

[0051] ;

[0052] in, Indicates the first The starting index of each iterative sub-segment. To set the step size;

[0053] S53. Construct candidate bridging paths: when exist Step length can reach To obtain the locally reachable region In the locally reachable region Monte Carlo random sampling is performed to construct multiple candidate bridging paths, denoted as . ,in This represents the number of sub-segments sampled.

[0054] Furthermore,

[0055] The iterative segments are replaced with the candidate bridging path set, and the iterations are repeated to obtain the preferred driving loop set, including:

[0056] S61, Local Replacement: Replace the iterative sub-section Replace with Then, the iterative candidate driving loop is obtained. , recorded as ;

[0057] S62. Calculate the bias: Calculate the iterative candidate driving cycle. Statistical characteristics of each driving cycle Compared with actual driving cycle reference characteristics The sum of the deviations ,

[0058] ;

[0059] S63. Constructing an optimal driving cycle set: driving cycle according to Rearrange the driving cycles from smallest to largest, and select the first w driving cycles to form a set; repeat S3-S6, and after T iterations, obtain the optimal set of driving cycles.

[0060] The local reconstruction iterative method utilizes a massive number of generated feasible bridging segments to replace local regions of existing suboptimal driving cycles. While preserving the majority of the superior features, it iterates multiple times to narrow the gap with the target baseline features. This achieves compatibility between low perturbation and a large local exploration range while minimizing disruption to the core features of existing superior driving cycles.

[0061] A second aspect of the present invention provides a data processing system for constructing a typical driving cycle, comprising:

[0062] Data processing module: Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix;

[0063] Based on the Markov state transition matrix and the Monte Carlo algorithm, a set of suboptimal driving cycles is obtained;

[0064] Resampling module: Based on the suboptimal driving cycle set and the total step size, the reachable region is determined, and the candidate driving cycle set within the reachable region is obtained through the Monte Carlo algorithm;

[0065] Based on the candidate driving cycle set, resampling is performed within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set;

[0066] The local reconstruction iteration module constructs a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and selects the iterative sub-segments of the better driving cycle set;

[0067] Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm;

[0068] The iterative segments are replaced with the set of candidate bridging paths, and the optimal driving loop set is obtained by repeating the iteration.

[0069] A third aspect of the present invention provides a computer-readable storage medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the data processing method for constructing a typical driving cycle as described in the first aspect.

[0070] The technical solution proposed in this application can bring the following beneficial effects:

[0071] 1. This invention can determine the reachable domain under fixed step size constraints within milliseconds, significantly reducing the high computational time caused by blindly searching the entire state space. Through resampling-based Monte Carlo sampling, efficient exploration can be achieved, avoiding getting trapped in local suboptimal solutions, thus balancing search breadth and computational efficiency.

[0072] 2. This invention efficiently reconstructs the driving loop structure on a large scale within a small range. Compared with the traditional optimization method of inefficient small-scale reconstruction on a large scale, it significantly improves both time efficiency and accuracy. Based on the initially generated suboptimal driving loop, it uses a large number of candidate segments in the reachable domain to reconstruct local segments, which can retain the excellent characteristics of the existing driving loop to the maximum extent and reduce the damage to the overall driving loop, thereby gradually approaching the target statistical features in multiple iterations.

[0073] The above description is merely an overview of the technical solution disclosed herein. In order to better understand the technical means of this disclosure and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this disclosure more apparent and understandable, specific embodiments of this disclosure are described below. Attached Figure Description

[0074] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this disclosure. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0075] Figure 1 This is a flowchart illustrating the steps of a data processing method for constructing a typical driving cycle, as provided in the embodiments of this specification.

[0076] Figure 2 This is a Markov state transition matrix diagram provided in the embodiments of this specification;

[0077] Figure 3 This is a diagram showing the reachable domain configuration provided in the embodiments of this specification;

[0078] Figure 4 This is a diagram illustrating the iterative process of 100 experiments provided in the embodiments of this specification;

[0079] Figure 5 This is a distribution range diagram of the sum of the experimental iteration deviations provided in the embodiments of this specification;

[0080] Figure 6 This is a comparison chart of the driving cycle construction time and accuracy provided in the embodiments of this specification;

[0081] Figure 7 This is a schematic diagram of the structure of a data processing system for constructing a typical driving cycle, provided in the embodiments of this specification. Detailed Implementation

[0082] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. The technical solutions provided by various embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0083] This application provides a data processing method and system for constructing a typical driving cycle.

[0084] A first aspect of the present invention provides a data processing method for constructing a typical driving cycle, comprising the following steps:

[0085] S1. Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix;

[0086] First, the actual driving cycle data is obtained by filtering the raw data. The raw data in this embodiment is based on 1,053,548 high-precision vehicle terminal records from a city's new energy vehicle big data platform. The raw data includes vehicle operating status, charging status, vehicle speed, cumulative driving range, and power battery state of charge (SOC), reflecting the vehicle's driving cycle in a typical city.

[0087] The raw data is first processed by systematically combining data such as vehicle operating status, speed, and remaining range to perform collaborative filtering on data records during the charging process. Only when the vehicle is stationary (operating status marked as 1) and the charging status is 3 (charging) or 4 (fully charged) will the system automatically identify and remove invalid operating data generated during that period. This step effectively ensures the consistency between charging data and the actual operating status of the vehicle, eliminating abnormal records caused by system errors or environmental interference.

[0088] After initial cleaning, the remaining valid operational data is further divided into trip segments. To ensure that the constructed driving cycle data has typical urban operational representativeness, all data is screened to remove invalid data with idling time exceeding 3 minutes due to parking or loading / unloading operations, while retaining the dynamic characteristics of the vehicle under normal driving conditions.

[0089] In addition, 16 statistical characteristics were obtained from the analysis, and a multi-dimensional characteristic description system for real-world driving data was constructed to quantitatively describe the data.

[0090] The 16 statistical characteristics include:

[0091] average speed The calculation formula is: ;

[0092] Average driving speed The calculation formula is: ;

[0093] velocity variance The calculation formula is: ;

[0094] average acceleration The calculation formula is: ;

[0095] Average deceleration The calculation formula is: ;

[0096] Acceleration time percentage The calculation formula is: ;

[0097] Deceleration time percentage The calculation formula is: ;

[0098] Percentage of constant speed time The calculation formula is: ;

[0099] Percentage of free time The calculation formula is: ;

[0100] Percentage of time spent in the 0~10km / h speed range The calculation formula is: ;

[0101] Percentage of time spent in the 10~20km / h speed range The calculation formula is:

[0102] ;

[0103] Percentage of time spent in the 20-30km / h speed range The calculation formula is:

[0104] ;

[0105] Percentage of time spent in the 30-40km / h speed range The calculation formula is:

[0106] ;

[0107] Percentage of time spent in the 40-50km / h speed range The calculation formula is:

[0108] ;

[0109] Percentage of time spent in the 50-60km / h speed range The calculation formula is:

[0110] ;

[0111] Percentage of time spent in the 60-70km / h speed range The calculation formula is:

[0112] ;

[0113] Finally, abnormal segments are identified by setting preset feature parameter thresholds to achieve intelligent screening of data segments. At the same time, in order to eliminate the impact of vehicle idling speed fluctuations in the range of 0-3km / h on data accuracy, data points with measured speed values ​​below 3km / h are set to zero.

[0114] After the above preprocessing, the final effective data volume is 923,164 records. Compared with the original dataset, the data quality has been significantly improved, ensuring the accuracy and representativeness of the data.

[0115] After processing the raw data, actual driving cycle data is obtained. The velocity-acceleration state is spatially discretized in order to perform Markov state transition matrix calculation.

[0116] After obtaining the actual driving cycle data, the specific data includes:

[0117] S11. Determine the sampling time state: Discretize the velocity and acceleration data of the actual driving cycle data, and divide them into several sampling time states according to the discrete grid, denoted as... ; The set of all sampled states is represented as ;

[0118] In this embodiment, the speed range of actual driving cycle data According to speed resolution The speed range is divided into several discrete speed intervals;

[0119] The nodes in the discrete velocity range ,in and These represent the upper and lower limits of the discrete velocity range; where, Set to 1 ,make = ;

[0120] Similarly, the acceleration range of actual driving cycle data According to acceleration resolution The acceleration range is divided into several discrete acceleration intervals, and the nodes of the discrete acceleration intervals are obtained as follows: ,in and The upper and lower limits of acceleration for a discrete acceleration interval; acceleration resolution. Set to 0.3 ,make = ;

[0121] If the state at a certain sampling time meets the state determination condition, then the state at that sampling time is denoted as... The state determination condition is: ;

[0122] This forms a grid with velocity as rows and acceleration as columns, where the state at a single sampling moment within the grid corresponds to the grid coordinates; thus, all possible grid combinations... Construct a finite set of states containing n elements That is, the velocity-acceleration within the grid corresponding to the state at a single sampling moment;

[0123] Number the grids obtained above to obtain a state lookup table. State lookup table Represented as: the set of positive integers from 1 to N, a state lookup table. Each positive integer in the table corresponds to a range of velocity and acceleration, for example, in a state lookup table. When the value is 1, the corresponding speed range is The acceleration range is ; and so on, state lookup table Each value corresponds to a grid interval.

[0124] S12, Statistical Transition Count: Count the number of transitions between two adjacent sampling times. Transferred to The number of times it appears is denoted as ;

[0125] ;in This is an indicator function; if the condition inside the parentheses is met, it is counted as 1, otherwise it is counted as 0. This represents the total sampling time.

[0126] S13. Obtain the Markov state transition matrix: For each state in the set at the sampling time... Transferred to The Markov transition matrix is:

[0127] ;

[0128] like Figure 2 As shown, Figure 2 middle This indicates the state at time t, the sampling time. This represents the state at sampling time t+1; TransitionProbability is the transition probability; by constructing the Markov state transition matrix, we can obtain the state at the current time. Under the condition that, the state transitions to the next time step. The probability of.

[0129] S2. Based on the Markov state transition matrix and Monte Carlo algorithm, the set of suboptimal driving cycles is obtained;

[0130] Based on the Markov state transition matrix, a new driving cycle is generated using the Monte Carlo algorithm.

[0131] According to the Monte Carlo algorithm, let the current state be... Examine the Markov state transition matrix. This corresponding line lists all possible next states and their probabilities.

[0132] Then the next moment will transition to the state of :

[0133] ;

[0134] A uniformly distributed random number r can be generated in the interval [0,1].

[0135] Then iterate through the states. The corresponding probability rows are incremented, and the probabilities are accumulated until the sum exceeds the random number r. At this point, the corresponding state is the selected state. ;in Indicates from Random numbers obtained by uniformly distributed sampling within an interval. This represents the cumulative sum of transition probabilities. In other words, it involves finding a random number within the cumulative probability distribution. The next state is obtained based on probability sampling from the position of the interval into which the object falls. = ;

[0136] Repeat the Monte Carlo random sampling process hundreds or thousands of times to generate a long-term velocity-acceleration sequence. Cut the obtained velocity-acceleration sequence into segments that match the actual driving cycle length (e.g., 1800 seconds). Filter and splice these segments to form the initial driving cycle set.

[0137] Specifically, the following steps are included:

[0138] S21. Generate the initial driving cycle set: Based on the Markov state transition matrix, generate several initial driving cycle sets using the Monte Carlo algorithm. ;

[0139] S22. Calculate the characteristic deviation: Calculate the i-th dimension statistical characteristic of each driving cycle in the initial driving cycle set. Compared with actual driving cycle reference characteristics deviation and the sum of deviations ;

[0140] in,

[0141] ;

[0142] S23. Obtain the set of suboptimal driving cycles: Sort the initial set of driving cycles by size d to obtain the set of suboptimal driving cycles, denoted as: .

[0143] Assuming the generated driving cycle time is 1800s, generate 10,000 initial driving cycle sets, and then calculate the sum of deviations for each cycle. According to the sum of deviations The driving cycles are rearranged in ascending order to obtain the suboptimal driving cycle set; however, the deviations of each driving cycle in the suboptimal driving cycle set are still relatively large and cannot meet the requirements.

[0144] S3. Based on the suboptimal driving cycle set and the total step size, determine the reachable region, and obtain the candidate driving cycle set within the reachable region using the Monte Carlo algorithm;

[0145] Based on the set of suboptimal driving cycles obtained in S2, an reachability domain is constructed for the driving cycles in the set of suboptimal driving cycles.

[0146] Based on the set of suboptimal driving cycles and the total step size, the reachable region is determined, specifically including:

[0147] S31. Set the initial state: Based on the suboptimal driving cycle set, set the starting state. and endpoint state Set the total step size ;

[0148] S32. Constructing the divergence domain Determine the starting state ,exist Reaching the final state within steps The set of states at all sampling times;

[0149] Recorded as: ;

[0150] S33. Constructing the convergence region : Determine the state from the endpoint ,exist Reaching the starting state within one step The set of states at all sampling times;

[0151] Recorded as: ;

[0152] S34. Construct reachable domains Divergence domain With convergence region The intersection of these two domains is denoted as the reachable region. ;

[0153] If it is necessary to just right Step from Transferred to ,So intermediate states within one step It must be in both the divergence region and the convergence region, denoted as Reachable domain;

[0154] in, ; It is the convergence region The order was reversed and Corresponding, reachable domain The sampling time state must be related to and Bidirectional connectivity, based on this characteristic, is possible within the reachable domain. This method enables rapid Monte Carlo search and constructs feasible state paths with a specified step size.

[0155] Based on its bidirectional connectivity, it can determine the reachable domain under fixed step size constraints in milliseconds, significantly reducing the high computational time caused by blindly searching the entire state space and ensuring a sufficient search for potential optimal solutions. At the same time, based on the suboptimal driving cycle set and the total step size, a massive number of candidate segments can be obtained in the reachable domain.

[0156] S4. Based on the candidate driving cycle set, resample within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set;

[0157] like Figure 3 As shown, in a given and Given the total step size L, the reachability domain and several Monte Carlo sampling paths are constructed; in the figure, the divergence domain is the region representing the divergence from... Starting point of divergence It contains the first All steps from The reachable states. In the diagram, the convergence domain (or scalar region) represents the region that can be backtracked to the th... When walking, you will eventually be able to reach... region of convergence The intersection of these two regions, i.e., the overlapping area of ​​reachable domains in the diagram, constitutes the reachable domain at each step. A set of states that has bidirectional connectivity is called a reachable region. The Monte Carlo Sampling curve in the figure represents the candidate segments generated by the Monte Carlo sampling method.

[0158] To further reduce the deviation between the statistical characteristics of the candidate driving cycle set and the statistical characteristics of the real driving data, the state lookup table is resampled and optimized.

[0159] Construct a state lookup table based on the state at the sampling time. When the order of the candidate driving cycle set remains unchanged, several resample sets are divided, and a temporary state lookup table is obtained based on each resample set. Based on the candidate driving cycle set and the temporary state lookup table The driving cycle characteristics were obtained. ; Calculate the generated driving cycle features and its reference features Total deviation ;like If the sum of deviations d is less than the sum of deviations d of the candidate driving cycle set, then... Updated to This serves as a new state lookup table; then, the Monte Carlo algorithm is used to generate candidate paths within the reachable domain, which are the candidate segments.

[0160] The specific steps are as follows:

[0161] S41, Random Selection a subset The sampling time state, and the corresponding velocity and acceleration values ​​range as follows: and Among them, the state lookup table This means assigning a specific physical quantity to the state at each discrete sampling time. That is, the velocity and acceleration values ​​corresponding to the state at that sampling moment;

[0162] S42. For each Resample and generate new velocities within the allowed VA (velocity-acceleration) cells. and acceleration , represented as , ,in Indicates a uniform distribution; Indicates speed Evenly distributed in Within the range; Indicates acceleration Evenly distributed in ( Within the range;

[0163] S43, Speed and acceleration Replace the original speed and acceleration ,but ( , ) was updated to , The middle does not belong to a subset The entries remain unchanged, resulting in a temporary state lookup table. ;

[0164] S44. Based on the candidate driving cycle set and the temporary state lookup table The driving cycle characteristics were obtained. The generated driving cycle characteristics are calculated according to the calculation formula in S22. and its reference features Total deviation ;

[0165] S45, if Then Updated to As a new state lookup table, it is represented as ;in, This indicates a system based on a candidate driving cycle set and a state lookup table. The total deviation is calculated; conversely, if Then maintain the state lookup table. ;

[0166] Through the resampling optimization of the state lookup table described above, a better approach can be explored within the physical limits allowed by the state at each sampling time. This allows for precise calibration of the statistical characteristics of the driving cycle without altering its Markov chain structure, avoiding local optima and achieving high-precision fitting of the target features.

[0167] S5. Construct a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and select the iterative sub-segments of the better driving cycle set;

[0168] Replace the corresponding local segments in the candidate driving cycle set with the candidate segments optimized in S4 to obtain a better driving cycle set.

[0169] Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm;

[0170] Since the transition between states at the sampling time conforms to the Markov state transition rule, the bridging path refers to the intermediate state path that can transition from the start state to the end state through a specific number of sampling steps; because it connects the two states, it is called the bridging path.

[0171] Specifically, it includes:

[0172] S51. Constructing a set of better driving cycles: Merge candidate segments into the corresponding candidate driving cycle sets to obtain a set of better driving cycles, denoted as: ,in Indicates the first In the next iteration, M is the number of driving loops contained in the set;

[0173] S52. Constructing an iterative segment set: In each driving cycle Random selection Non-overlapping iterative subsegments , recorded as Set the start-up state and terminal state They are respectively recorded as:

[0174] ;

[0175] ;

[0176] in, Indicates the first The starting index of each iterative sub-segment. To set the step size;

[0177] S53. Construct candidate bridging paths: when exist Step length can reach To obtain the locally reachable region In the locally reachable region Monte Carlo random sampling is performed to construct multiple candidate bridging paths, denoted as . ,in This represents the number of segments sampled.

[0178] S6. Replace the iterative sub-segment with the candidate bridging path set, and repeat the iteration to obtain the preferred driving loop set.

[0179] Specifically, the following steps are included:

[0180] S61, Local Replacement: Replace the iterative sub-section Replace with Then, the iterative candidate driving loop is obtained. , recorded as ;

[0181] S62. Calculate the bias: Calculate the iterative candidate driving cycle. Statistical characteristics of each driving cycle Compared with actual driving cycle reference characteristics The sum of the deviations ,

[0182] ;

[0183] S63. Constructing an optimal driving cycle set: driving cycle according to Rearrange the driving cycles from smallest to largest, and select the first w driving cycles to form a set; repeat S3-S6, and after T iterations, obtain the optimal set of driving cycles.

[0184] For example, w is 10-100; after obtaining the set of the first w driving cycles, the gap between the target baseline feature and the target feature is reduced by continuous iteration, for example, T is 1000 times;

[0185] like Figure 4 As shown, the iterative process of 100 experiments is illustrated. In the figure, "Iteration" refers to the number of iterations; "Number" refers to the number of experiments; and "Sum of deviation" refers to the sum of deviations.

[0186] like Figure 5 As shown, the distribution range of the sum of deviations during these 100 experimental iterations is illustrated, with the red line representing the average value. In the figure, Iteration refers to the number of iterations; Min / MaxRange refers to the minimum / maximum boundary values, representing the boundary values ​​of the minimum and maximum deviations; Mean Deviation refers to the average deviation; and Sum of deviation refers to the sum of deviations.

[0187] pass Figure 4 and Figure 5 It can be seen that the deviation is high in the early stage of iteration, exceeding 150%. With continuous local optimization, the deviation decreases rapidly; after about 1000 iterations, the rate of decrease of the sum of deviations of the 16 indicators slows down significantly, and gradually converges to 1.2%~1.9% through the resampling step. This shows that the local reconstruction iterative algorithm can effectively search the feasible space of the driving cycle and gradually approach the target statistical features.

[0188] The local reconstruction iterative method utilizes the generated massive number of feasible bridging segments to replace local regions of existing suboptimal driving cycles. While preserving most of the superior features, it iterates multiple times to narrow the gap with the target baseline features. This achieves compatibility between low perturbation and high local exploration range while minimizing damage to the core features of existing superior driving cycles.

[0189] This invention can determine the reachable domain under fixed step size constraints within milliseconds, significantly reducing the high computational time caused by blindly searching the entire state space. Through resampling-based Monte Carlo sampling, efficient exploration can be achieved, avoiding getting trapped in local suboptimal solutions, thus balancing search breadth and computational efficiency.

[0190] As shown in Table 1, the above-mentioned method of the present invention, ResamplingReachableDomain (hereinafter referred to as MC-RRD), is compared with the widely used Markov chain method (hereinafter referred to as MC method) and Markov chain-genetic algorithm method (hereinafter referred to as MC-GA method) in terms of driving cycle construction efficiency.

[0191] The main bottleneck of the MC method is that it requires frequent random sampling in the entire probability transition matrix. As the number of samples Q increases, the time consumption also increases linearly. When the number of samples reaches 100,000, the time consumption is as high as 4692.359s, and the efficiency of candidate driving loop construction is very low.

[0192] The MC-GA method significantly improves the time consumption compared to the MC method, but it still needs to repeatedly judge the transferability of the state in the crossover and mutation stages for each driving cycle, making it difficult to generate a large number of candidate driving cycles more quickly. Moreover, since its candidate driving cycles are mainly generated through the crossover process, its time consumption increases linearly with the increase of the number of samples Q due to the efficiency of the crossover process, resulting in low computational efficiency.

[0193] Due to the sublinear increase in reconstruction time of the MC-RRD sub-segment, by setting the step size Δ and the number of samples Q, when Δ is within the normal range of 10~25s (the reconstruction ratio is about 5%~15%), it can generate 10~100 times more candidate driving cycles than the MC-GA method in the same amount of time. Therefore, the MC-RRD method has a huge advantage over other optimization methods in terms of refined optimization.

[0194] When Δ=10s and Q=1000, the candidate driving cycle construction efficiency of MC-RRD is 966 times that of MC and 16.7 times that of MC-GA. Since the reachability domain of MC-RRD optimizes the relationship between the number of candidate driving cycles and the time consumption from a linear relationship to a sublinear relationship, it can construct a large number of candidate driving cycles for optimization exploration in a short time, thus achieving a compatibility between time efficiency and exploration capability.

[0195] Table 1 is a comparison of the time consumption for constructing candidate driving cycles for each method;

[0196]

[0197] like Figure 6 As shown, a comparative analysis of the accuracy of three typical driving cycle construction methods, MC, MC-GA and MC-RRD, is presented. Figure 6 The horizontal axis represents time, and the vertical axis represents the sum of deviations of 16 statistical indicators.

[0198] Figure 6 In this context, MC represents the sum of deviations curve obtained through the MC method;

[0199] GA represents the sum of biases curve obtained by the MC-GA method. Converged at 2692s indicates that the biases converge at 2692s, where the sum of biases is 26.65%.

[0200] RRD represents the sum of biases curve obtained by the MC-RRD method. Converged at 375s in the curve indicates that the biases converge at 375s, where the sum of biases is 1.22%.

[0201] The Markov Chain (MC) method failed to reduce the bias to a low level over time, indicating that relying solely on random sampling of Markov chains to generate driving loops lacks effective constraints on the target statistical characteristics, resulting in large index bias and difficulty in convergence. The sum of the biases of the 16 final statistical indicators of the MC method was 164.88%, and the average bias was 10.305%.

[0202] Through continuous iteration, MC-GA reduced the sum of biases to 30% at around 1,100s; it converged at approximately 2,700s: the sum of biases for the 16 indicators was 26.65%, and the average bias of each indicator was 1.66%.

[0203] The sum of biases of the MC-RRD method decreased to 30% at about 65s and converged to 1.22% at 375s, with the final average bias of each index being 0.076%.

[0204] The MC-RRD method of this invention not only significantly shortens the convergence time, but also has a much higher accuracy than MC and MC-GA. This shows that the approach of efficient large-scale reconstruction in a small range is significantly improved in terms of both time efficiency and accuracy compared to the approach of inefficient small-scale reconstruction in a large range. Based on the idea of ​​refined driving cycle construction, this study achieves compatibility of the three "impossible trinity" of perturbation controllability, time efficiency, and exploration range through local resampling reachable domain.

[0205] A second aspect of the present invention provides a data processing system 700 for constructing a typical driving cycle, comprising:

[0206] Data processing module 701: Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix;

[0207] Based on the Markov state transition matrix and the Monte Carlo algorithm, a set of suboptimal driving cycles is obtained;

[0208] Resampling module 702: Based on the suboptimal driving cycle set and the total step size, it determines the reachable region and obtains the candidate driving cycle set within the reachable region through the Monte Carlo algorithm;

[0209] Based on the candidate driving cycle set, resampling is performed within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set;

[0210] The local reconstruction iteration module 703 constructs a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and selects the iterative sub-segments of the better driving cycle set;

[0211] Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm;

[0212] The iterative segments are replaced with the set of candidate bridging paths, and the optimal driving loop set is obtained by repeating the iteration.

[0213] A third aspect of the present invention provides a computer-readable storage medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the data processing method for constructing a typical driving cycle as described in the first aspect.

[0214] This embodiment can divide the method into functional modules based on the above method example. For example, each function can be assigned to a separate module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0215] When each functional module is divided according to its corresponding function, the vehicle may include: a data processing module, a resampling module, a local reconstruction iteration module, etc. It should be noted that all relevant content of each step involved in the above method embodiments can be referenced from the functional description of the corresponding functional module, and will not be repeated here.

[0216] This embodiment also provides a computer-readable storage medium (including but not limited to disk storage, CD-ROM, optical storage, etc.) storing computer program code. When the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement the data processing method and system for constructing a typical driving cycle provided in the above embodiment.

[0217] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to implement the data processing method and system for constructing a typical driving cycle provided in the above embodiment. The beneficial effects of the above embodiments can be found in the corresponding methods described above, and will not be repeated here.

[0218] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0219] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules or units is merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. In the description of this disclosure, it should be understood that if terms such as "upper," "lower," "front," "rear," "left," and "right" are used to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, they are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the indicated position or element must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this disclosure.

[0220] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0221] The above are merely embodiments of this disclosure and are not intended to limit the scope of this disclosure. Various modifications and variations can be made to this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of the claims of this disclosure.

Claims

1. A data processing method for constructing a typical driving cycle, characterized in that, Includes the following steps: S1. Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix; S2. Based on the Markov state transition matrix and Monte Carlo algorithm, the set of suboptimal driving cycles is obtained; S3. Based on the suboptimal driving cycle set and the total step size, determine the reachable region, and obtain the candidate driving cycle set within the reachable region using the Monte Carlo algorithm; S4. Based on the candidate driving cycle set, resample within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set; S5. Construct a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and select the iterative sub-segments of the better driving cycle set; Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm; S6. Replace the iterative sub-segment with the candidate bridging path set, and repeat the iteration to obtain the preferred driving loop set.

2. The data processing method for constructing a typical driving cycle according to claim 1, characterized in that, The process of determining the state at each sampling time based on actual driving cycle data and obtaining the Markov state transition matrix includes the following steps: S11. Determine the sampling time state: Divide the actual driving cycle data into several sampling time states, denoted as... ; The set of all sampled states is represented as ; S12, Statistical Transition Count: Count the number of transitions between two adjacent sampling times. Transferred to The number of times it appears is denoted as ; S13. Obtain the Markov state transition matrix: For each state in the set at the sampling time... Transferred to The Markov transition matrix is: 。 3. The data processing method for constructing a typical driving cycle according to claim 1, characterized in that, The process of obtaining the suboptimal driving cycle set based on the Markov state transition matrix and Monte Carlo algorithm includes the following steps: S21. Generate the initial driving cycle set: Based on the Markov state transition matrix, generate several initial driving cycle sets using the Monte Carlo algorithm. ; S22. Calculate the characteristic deviation: Calculate the i-th dimension statistical characteristic of each driving cycle in the initial driving cycle set. Compared with actual driving cycle reference characteristics deviation and the sum of deviations ; in, ； S23. Obtain the set of suboptimal driving cycles: Sort the initial set of driving cycles by size d to obtain the set of suboptimal driving cycles, denoted as: .

4. The data processing method for constructing a typical driving cycle according to claim 1, characterized in that, The determination of the reachable domain based on the suboptimal driving cycle set and the total step size includes: S31. Set the initial state: Based on the suboptimal driving cycle set, set the starting state. and endpoint state Set the total step size ; S32. Constructing the divergence domain Determine the starting state ,exist Reaching the final state within steps The set of states at all sampling times; Recorded as: ; S33. Constructing the convergence region : Determine the state from the endpoint ,exist Reaching the starting state within one step The set of states at all sampling times; Recorded as: ; S34. Construct reachable domains Divergence domain With convergence region The intersection of these two domains is denoted as the reachable region. .

5. The data processing method for constructing a typical driving cycle according to claim 1, characterized in that, The step of resampling within the set of states at the sampling time based on the candidate driving cycle set to obtain candidate sub-segments of the candidate driving cycle set includes: Construct a state lookup table based on the state at the sampling time. When the order of the candidate driving cycle set remains unchanged, several resample sets are divided, and a temporary state lookup table is obtained based on each resample set. Based on the candidate driving cycle set and the temporary state lookup table The driving cycle characteristics were obtained. ; Calculate the generated driving cycle features and its reference features Total deviation ;like If the sum of deviations d is less than the sum of deviations d of the candidate driving cycle set, then... Updated to This serves as a new state lookup table; then, candidate paths, or candidate segments, are generated within the reachable domain using the Monte Carlo algorithm.

6. The data processing method for constructing a typical driving cycle according to claim 5, characterized in that, The state lookup table Assign a specific physical quantity to the state at each discrete sampling time. , This represents the velocity and acceleration values ​​corresponding to the state at that sampling moment; [referring to the state lookup table]. A subset of the velocity is resampled to obtain a resampled set, and each resampled set contains a new velocity. and acceleration Combined into a temporary state lookup table .

7. The data processing method for constructing a typical driving cycle according to claim 1, characterized in that, Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm, including: S51. Constructing a set of better driving cycles: Merge candidate segments into the corresponding candidate driving cycle sets to obtain a set of better driving cycles, denoted as: ,in Indicates the first In the next iteration, M is the number of driving loops contained in the set; S52. Constructing an iterative segment set: In each driving cycle Random selection Non-overlapping iterative subsegments , recorded as Set the start-up state and terminal state They are respectively recorded as: ； ； in, Indicates the first The starting index of each iterative sub-segment. To set the step size; S53. Construct candidate bridging paths: when exist Step length can reach To obtain the locally reachable region In the locally reachable region Monte Carlo random sampling is performed to construct multiple candidate bridging paths, denoted as . ,in This represents the number of segments sampled.

8. The data processing method for constructing a typical driving cycle according to claim 7, characterized in that, The step of replacing the iterative sub-segments with the candidate bridging path set and repeating the iteration to obtain the preferred driving loop set includes: S61, Local Replacement: Replace the iterative sub-section Replace with Then, the iterative candidate driving loop is obtained. , recorded as ; S62. Calculate the bias: Calculate the iterative candidate driving cycle. Statistical characteristics of each driving cycle Compared with actual driving cycle reference characteristics The sum of the deviations , ； S63. Constructing an optimal driving cycle set: driving cycle according to Rearrange the driving cycles from smallest to largest, and select the first w driving cycles to form a set; repeat S3-S6, and after T iterations, obtain the optimal set of driving cycles.

9. A data processing system for constructing a typical driving cycle, characterized in that... The system includes: Data processing module: Based on actual driving cycle data, determine the state at the sampling time and obtain the Markov state transition matrix; Based on the Markov state transition matrix and the Monte Carlo algorithm, a set of suboptimal driving cycles is obtained; Resampling module: Based on the suboptimal driving cycle set and the total step size, the reachable region is determined, and the candidate driving cycle set within the reachable region is obtained through the Monte Carlo algorithm; Based on the candidate driving cycle set, resampling is performed within the set of states at the sampling time to obtain candidate sub-segments of the candidate driving cycle set; The local reconstruction iteration module constructs a better driving cycle set based on the candidate driving cycle set and candidate sub-segments, and selects the iterative sub-segments of the better driving cycle set; Within the reachability domain of the iterative sub-segment, a set of candidate bridging paths is obtained by random sampling using the Monte Carlo algorithm; The iterative segments are replaced with the set of candidate bridging paths, and the optimal driving loop set is obtained by repeating the iteration.

10. A computer-readable storage medium, characterized in that, It stores instructions that, when executed by one or more processors, cause the processors to perform the data processing method for constructing a typical driving cycle as described in any one of claims 1-8.

Images