Method, system and equipment for constructing speed-slope driving condition of tractor

Abstract

The invention relates to the technical field of tractor working conditions, in particular to a tractor speed-slope driving working condition construction method, system and device, and the method comprises the steps: collecting the real driving data of a tractor in a target scene through a vehicle-mounted terminal device, and obtaining the collected original data in batches through a cloud platform; preprocessing the original data; based on the Markov property of the operation condition, discretizing the processed operation data per second of the vehicle into state variables including the vehicle speed, the acceleration and the road gradient, counting each state transition probability matrix, and generating a plurality of condition sequences by using a Markov chain Monte Carlo method; establishing working condition evaluation indexes containing working condition characteristic parameters, and constructing a working condition characteristic objective function; and performing iterative optimization on the generated working condition sequence by adopting an elitist adaptive genetic algorithm based on the working condition characteristic objective function until the comprehensive deviation value reaches a preset target value, and obtaining an optimal working condition sequence. And the accuracy of the speed-slope working condition can be effectively evaluated.

Description

Technical Field

[0001] This application relates to the field of tractor operating condition technology, specifically to a method, system and equipment for constructing tractor speed-gradient driving conditions. Background Technology

[0002] Driving conditions simulate real-world driving scenarios and can be used to test and evaluate vehicle performance, such as fuel economy, power, emissions, and the driving range of new energy vehicles. Driving conditions directly influence fuel consumption, emissions testing methods, and the development of corresponding standards. Fuel economy and emissions performance vary depending on the driving conditions. For automakers, developing driving conditions provides a reference for the technical evaluation and optimization of new models. Therefore, developing corresponding driving conditions based on specific market segments is of greater research significance.

[0003] Previous studies have typically used established driving cycles, such as C-WTVC and HD-UDDS, for vehicle performance evaluation and energy consumption simulation. These traditional driving cycles are developed based on local driving characteristics and are suitable for those regions. However, significant differences exist in vehicle driving characteristics across different regions and vehicle types, making standard driving cycles insufficient to meet the needs of diverse areas. Therefore, to ensure the representativeness of driving cycles, it is necessary to select appropriate driving cycle conditions, evaluation criteria, and implementation methods based on usage, vehicle type, and road conditions. Summary of the Invention

[0004] In view of the significant differences in vehicle driving characteristics between different regions and different vehicle models, and the difficulty in meeting the needs of different regions by standard driving conditions, this invention provides a method, system and equipment for constructing tractor speed-gradient driving conditions.

[0005] In a first aspect, the present invention provides a method for constructing a tractor speed-gradient driving condition, comprising: S1. Use on-board terminal equipment to collect real driving data of the tractor in the target scenario, and obtain the collected raw data in batches through the cloud platform; S2. Visualize and statistically analyze the raw data, use rule-based outlier identification to remove outlier data, and use filtering algorithms to reduce noise in the data. S3. Based on the Markov property of the operating conditions, the processed vehicle operating data per second is discretized into state variables including vehicle speed, acceleration and road slope. The state transition probability matrix is ​​calculated, and several operating condition sequences are generated using the Markov chain Monte Carlo method. S4. Establish working condition evaluation indicators that include working condition characteristic parameters, and construct working condition characteristic objective functions; S5. Based on the objective function of working condition characteristics, the generated working condition sequence is iteratively optimized using an elite adaptive genetic algorithm until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal working condition sequence.

[0006] As a further limitation of the technical solution of the present invention, S1 specifically includes: S11. Collect data on the tractor in the target scenario; the collected data includes time, vehicle speed, gradient, load, altitude, engine speed and torque parameters. S12. Filter vehicle operation data for specific areas or routes through the cloud platform.

[0007] As a further limitation of the technical solution of the present invention, S2 specifically includes: S21. Visualize the raw data to detect outliers in sampling interval, vehicle speed distribution, acceleration distribution, slope distribution, and load distribution. S22. The 3σ anomaly detection algorithm is used to identify anomalies in the acceleration data and to filter anomalies in the vehicle speed data. Interpolation methods are used to smooth out missing values ​​and acceleration data that exceed the threshold. S23. Use a second-order Butterworth low-pass filter with a cutoff frequency of the set value to perform noise reduction processing on the vehicle speed data.

[0008] As a further limitation of the technical solution of the present invention, S3 specifically includes: S31. Discretize the vehicle speed, acceleration, and road slope into state variables, and divide the state intervals; S32. Calculate the state transition frequency in the time series and generate the state transition probability matrix; S33. Use the Markov chain Monte Carlo method to simulate and generate a working condition sequence. If the last state of the simulation is not the idling state, select the parking segment that meets the conditions and splice it into the current state to form a complete working condition sequence from zero to zero.

[0009] As a further limitation of the technical solution of the present invention, S33 specifically includes: S331. Use the Markov chain Monte Carlo method to simulate and generate the working condition sequence; S332. After the MCMC simulation is completed, check the final state of the generated operating condition sequence; S333. If the final state is not the idling state, analyze the final state and obtain the vehicle speed, acceleration and gradient of the final state. S334. Select segments that meet the set conditions from the segment library as candidate segments; S335. Calculate the evaluation index of the candidate segments and make it dimensionless; S336. Calculate the sum of squares of the average index error between each candidate segment and the overall data, and select the segment with the smallest error as the parking segment. S337. Patch the selected parking segment to the end of the current sequence to ensure that the spliced ​​sequence meets the requirements of the state transition probability matrix; S338. After splicing, a complete cyclic operating condition sequence from zero speed to zero speed is generated; the set conditions include: The difference between the initial speed and the final speed of the current driving condition sequence should be less than the first speed threshold; the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than the first acceleration threshold; the difference between the gradient and the final gradient of the current driving condition sequence should be less than the first percentage; and the control segment duration should be less than the first time threshold.

[0010] As a further limitation of the technical solution of the present invention, S4 specifically includes: S41. Calculate speed-related and slope-related characteristic parameters, including average speed, acceleration time ratio, deceleration time ratio, slope standard deviation, average uphill slope, and average downhill slope. S42. Calculate the weights of each feature parameter using the feature contribution rate, and construct the objective function for the working condition features.

[0011] As a further limitation of the technical solution of the present invention, S42 specifically includes: S421. Standardize the calculated speed and slope characteristic parameters; S422. Train the model using the random forest algorithm and calculate the number of times the feature splits in the decision tree; S423. Based on the calculation results, derive the importance score for each feature. .

[0012] S424. Calculate the weight of each feature parameter based on the feature importance score; where, feature... weights This indicates the relative importance of the feature in the objective function; In the formula, The sum of the importance scores for all features, where n is the total number of feature parameters; S425. Using the calculated weights, construct a working condition characteristic objective function, wherein the objective function is used to measure the deviation between the constructed working condition sequence and the actual working condition sequence; objective function

[0013] In the formula, For characteristic parameters The weights are denoted by n, where n is the total number of feature parameters. These are the constructed operating condition sequence and the actual operating condition sequence, respectively. To obtain from the operating condition sequence The characteristic parameters calculated in The value, To obtain from the operating condition sequence The characteristic parameters calculated in The value of .

[0014] As a further limitation of the technical solution of the present invention, S5 specifically includes: S51. Select a set number of working condition sequences as the initial population, and calculate the objective function value for each working condition sequence; S52. Calculate the fitness value of the working condition sequence based on the objective function value, select the top N excellent individuals to establish an elite population, and establish a normal population for the remaining individuals, prioritizing the optimization of elite individuals; S53. Establish a crossover operator, select an individual from the elite population and an individual from the ordinary population, perform the crossover operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. S54. Establish a mutation operator, randomly select an individual from the elite population to perform mutation operation, generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. S55. A new generation of population is formed through crossover and mutation. Individuals with fitness below the set value are removed. The top N elite individuals are iteratively optimized until the objective function value of the working condition sequence meets the preset target value, thus obtaining the optimal working condition sequence.

[0015] Secondly, the present invention also provides a tractor speed-gradient driving condition construction system, comprising: Data acquisition module: Used to collect real driving data of the tractor in the target scenario using the vehicle terminal equipment and upload it to the cloud platform, and obtain the collected raw data in batches through the cloud platform; Data processing module: used to perform visual inspection and statistical feature analysis on raw data, use rule-based outlier identification method to remove outlier data, and use filtering algorithm to reduce noise in the data; The working condition generation module is used to discretize the processed vehicle running data per second into state variables containing vehicle speed, acceleration and road slope based on the Markov property of the operating conditions, calculate the state transition probability matrix, and generate several working condition sequences using the Markov chain Monte Carlo method. Operating condition evaluation module: used to establish operating condition evaluation indicators that include operating condition characteristic parameters and construct operating condition characteristic objective functions; Operating condition optimization module: Based on the operating condition characteristic objective function, it uses an elite adaptive genetic algorithm to iteratively optimize the generated operating condition sequence until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal operating condition sequence.

[0016] As a further limitation of the technical solution of the present invention, the data acquisition module is specifically used to collect data on the tractor in the target scenario; the collected data includes time, vehicle speed, gradient, load, altitude, engine speed and torque parameters; and vehicle operation data of specific areas or routes are filtered through the cloud platform.

[0017] As a further limitation of the technical solution of the present invention, the data processing module includes a visualization inspection unit, an anomaly detection unit, and a noise reduction processing unit; The visualization inspection unit is used to perform visual inspection of the raw data and detect outliers in sampling interval, vehicle speed distribution, acceleration distribution, slope distribution and load distribution. The anomaly detection unit is used to identify anomalies in acceleration data using the 3σ anomaly detection algorithm, filter anomalies in vehicle speed data, and use interpolation methods to smooth out missing values ​​and acceleration data that exceed the threshold. The noise reduction processing unit is used to perform noise reduction processing on vehicle speed data using a second-order Butterworth low-pass filter with a cutoff frequency set to a set value.

[0018] As a further limitation of the technical solution of the present invention, the working condition generation module includes a state variable processing unit, a transition probability calculation unit, and a working condition sequence simulation unit. The state variable processing unit is used to discretize vehicle speed, acceleration, and road slope into state variables and divide the state intervals. The transition probability calculation unit is used to calculate the transition frequency of each state in the time series and generate the state transition probability matrix. The operating condition sequence simulation unit is used to simulate and generate operating condition sequences using the Markov chain Monte Carlo method. If the last state of the simulation is not the idling state, a parking segment that meets the conditions is selected and spliced ​​into the current state to form a complete operating condition sequence from zero to zero.

[0019] As a further limitation of the technical solution of this invention, the working condition sequence simulation unit is specifically used to generate a working condition sequence using the Markov chain Monte Carlo method; after the MCMC simulation is completed, the final state of the generated working condition sequence is checked; if the final state is not the idling state, the final state is parsed to obtain the vehicle speed, acceleration, and gradient of the final state; segments that meet the set conditions are selected from the segment library as candidate segments; the evaluation index of the candidate segments is calculated and dimensionless; the sum of squares of the average index error between each candidate segment and the overall data is calculated, and the segment with the smallest error is selected as the parking segment; the selected parking segment is spliced ​​to the end of the current sequence to ensure that the spliced ​​sequence meets the requirements of the state transition probability matrix; after splicing, a complete cyclic working condition sequence from zero speed to zero speed is generated; wherein the set conditions include: The difference between the initial speed and the final speed of the current driving condition sequence should be less than the first speed threshold; the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than the first acceleration threshold; the difference between the gradient and the final gradient of the current driving condition sequence should be less than the first percentage; and the control segment duration should be less than the first time threshold.

[0020] As a further limitation of the technical solution of the present invention, the working condition evaluation module includes a feature parameter calculation unit and an objective function construction unit; The feature parameter calculation unit is used to calculate speed-type and slope-type feature parameters, including average speed, acceleration time ratio, deceleration time ratio, slope standard deviation, average uphill slope, and average downhill slope. The objective function construction unit is used to calculate the weights of each feature parameter using the feature contribution and construct the working condition feature objective function.

[0021] As a further limitation of the technical solution of the present invention, the objective function construction unit includes a calculation processing submodule, an importance score calculation submodule, a weight calculation submodule, and an objective function construction submodule; The calculation and processing submodule is used to standardize the calculated speed and slope characteristic parameters; The importance score calculation submodule is used to train the model using the random forest algorithm, calculate the number of times a feature splits in the decision tree, and derive the importance score for each feature based on the calculation results. .

[0022] The weight calculation submodule is used to calculate the weight of each feature parameter based on the feature importance score; where, feature weights This indicates the relative importance of the feature in the objective function; In the formula, The sum of the importance scores for all features, where n is the total number of feature parameters; The objective function construction submodule is used to construct an objective function for working condition characteristics using the calculated weights. The objective function is used to measure the deviation between the constructed working condition sequence and the actual working condition sequence. objective function

[0023] In the formula, For characteristic parameters The weights are denoted by n, where n is the total number of feature parameters. These are the constructed operating condition sequence and the actual operating condition sequence, respectively. To obtain from the operating condition sequence The characteristic parameters calculated in The value, To obtain from the operating condition sequence The characteristic parameters calculated in The value of .

[0024] As a further limitation of the technical solution of this invention, the working condition optimization module is specifically used to select a set number of working condition sequences as the initial population and calculate the objective function value of each working condition sequence; calculate the fitness value of the working condition sequence based on the objective function value, select the top N excellent individuals to establish an elite population, and the remaining individuals to establish a normal population, prioritizing the optimization of elite individuals; establish a crossover operator, select one individual from the elite population and one individual from the normal population, perform a crossover operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, it is retained; establish a mutation operator, randomly select one individual from the elite population to perform a mutation operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, it is retained; form a new generation population through crossover and mutation, delete individuals with fitness lower than the set value, iteratively optimize the top N elite individuals until the objective function value of the working condition sequence meets the preset target value, and obtain the optimal working condition sequence.

[0025] Thirdly, the present invention also provides an electronic device, the electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; the memory storing computer program instructions executable by the at least one processor, the computer program instructions being executed by the at least one processor to enable the at least one processor to execute the tractor speed-gradient driving condition construction method as described in the first aspect.

[0026] As can be seen from the above technical solutions, this application has the following advantages: The rule-based anomaly detection and filtering algorithm proposed in this application can quickly identify and remove outliers in the dataset, improving the data quality of subsequent processing. This method has high universality. Since there are few short-stroke samples of the tractor unit, it is difficult to construct driving conditions that reflect actual operation using the traditional short-stroke method. This invention, based on the Markov property of the driving conditions, uses the Markov chain Monte Carlo algorithm to deduce the driving conditions, effectively avoiding the characteristics of excessively long kinematic segments and few driving segments of the tractor unit. To address the randomness and low efficiency of the Markov chain Monte Carlo algorithm, an elite adaptive genetic algorithm is used to specifically optimize the driving condition sequence. A speed-gradient driving condition evaluation index is established. By using weight vectors to calculate the absolute comprehensive deviation between the constructed driving conditions and the original data, the accuracy of the speed-gradient driving condition can be effectively evaluated. Attached Figure Description

[0027] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic flowchart illustrating the method provided in an embodiment of the present invention.

[0029] Figure 2 This is a statistical feature diagram of acceleration.

[0030] Figure 3 This is for anomaly detection based on the 3σ principle.

[0031] Figure 4 This refers to valid slope data identified based on an identification algorithm.

[0032] Figure 5 This is the speed-gradient curve.

[0033] Figure 6 This describes the process of constructing working conditions based on the Markov chain Monte Carlo algorithm.

[0034] Figure 7 This describes the working condition optimization process based on genetic algorithms.

[0035] Figure 8 To set the road section operating condition curve.

[0036] Figure 9 This is a system block diagram provided for an embodiment of the present invention. Detailed Implementation

[0037] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0038] like Figure 1 As shown, this embodiment of the invention provides a method for constructing a tractor speed-gradient driving condition, including: S1. Use on-board terminal equipment to collect real driving data of the tractor in the target scenario, and obtain the collected raw data in batches through the cloud platform; It should be noted that S1 specifically includes: S11. Collect data on the tractor in the target scenario; the collected data includes time, vehicle speed, gradient, load, altitude, engine speed and torque parameters. S12. Filter vehicle operation data for specific areas or routes through the cloud platform.

[0039] Depending on the operating scenario, tractor units can be categorized into express delivery, industrial transportation, coal transportation, container transportation, hazardous materials transportation, and heavy haulage transportation. Data collection can be conducted for a specific sub-scenario. The cloud platform can not only filter vehicles in a certain area but also set the starting and ending points of routes to accurately match vehicles running on those routes. Intelligent driving systems and equipment are used to collect driving data and upload it to the cloud. The cloud then performs batch analysis and processing of the operating data, which includes parameters such as time, vehicle speed, gradient, load, altitude, engine speed, and torque.

[0040] S2. Visualize and statistically analyze the raw data, use rule-based outlier identification to remove outlier data, and use filtering algorithms to reduce noise in the data. It should be noted that this step specifically includes: S21. Visualize the raw data to detect outliers in sampling interval, vehicle speed distribution, acceleration distribution, slope distribution, and load distribution. S22. The 3σ anomaly detection algorithm is used to identify anomalies in the acceleration data and to filter anomalies in the vehicle speed data. Interpolation methods are used to smooth out missing values ​​and acceleration data that exceed the threshold. S23. A second-order Butterworth low-pass filter with a cutoff frequency of 0.5Hz is used to reduce noise in the vehicle speed data.

[0041] S3. Based on the Markov property of operating conditions, the processed vehicle operating data per second is discretized into state variables including vehicle speed, acceleration, and road gradient. The state transition probability matrices are calculated, and several operating condition sequences are generated using the Markov chain Monte Carlo method; specifically including: S31. Discretize the vehicle speed, acceleration, and road slope into state variables, and divide the state intervals; The second-level data is discretized into state variables, including vehicle speed, acceleration, and road gradient. The driving states are divided into intervals based on the range of vehicle speed, acceleration, and gradient. The speed step size is 0.1 m / s, resulting in m speed states. The acceleration step size is 0.05 m / s. 2 A total of n acceleration states were obtained, with a slope step size of 0.1%, resulting in f slope states. Each state consists of vehicle speed, acceleration, and slope. The data is labeled with the state labels. , , , … .

[0042] S32. Calculate the state transition frequency in the time series and generate the state transition probability matrix; Based on the Markov property of the operating conditions, the state transition frequencies of each time series are calculated, and then the state transition probability matrix TPM is calculated. Each element in TPM represents the probability that the vehicle transitions from one driving state to another (including remaining in the same state). The expression for the state transition matrix TPM is:

[0043] Due to vehicles entering tunnels or signal interference, there are some missing values ​​in the vehicle speed and gradient data. For segments with long missing time periods, these need to be skipped when calculating the time-series state transition probability matrix.

[0044] As the current state, this value is the row number of the state transition probability matrix. The sum of probabilities for that row is calculated:

[0045] Using uniform distribution, according to The accumulated probability is used to segment the interval (0,1], and the length of the segment corresponds to the value of the state transition probability. A random number k is generated in the interval (0,1], and the value of k is related to... Each cumulative probability element in the data is compared, and an equation is used to determine which interval the random number k falls into. The smallest element greater than the random data k is found, thereby determining the next driving state.

[0046]

[0047] S33. Use the Markov chain Monte Carlo method to simulate and generate a sequence of operating conditions. If the last state in the simulation is not the idling state, select a stopping segment that meets the conditions and splice it to the current state to form a complete sequence of operating conditions from zero to zero. Specifically, S33 includes: S331. Use the Markov chain Monte Carlo method to simulate and generate the working condition sequence; S332. After the MCMC simulation is completed, check the final state of the generated operating condition sequence; S333. If the final state is not the idling state, analyze the final state and obtain the vehicle speed, acceleration and gradient of the final state. S334. Select segments that meet the set conditions from the segment library as candidate segments; S335. Calculate the evaluation index of the candidate segments and make it dimensionless; S336. Calculate the sum of squares of the average index error between each candidate segment and the overall data, and select the segment with the smallest error as the parking segment. S337. Patch the selected parking segment to the end of the current sequence to ensure that the spliced ​​sequence meets the requirements of the state transition probability matrix; S338. After splicing, a complete cyclic operating condition sequence from zero speed to zero speed is generated; the set conditions include: The difference between the initial speed and the final speed of the current driving condition sequence should be less than the first speed threshold; the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than the first acceleration threshold; the difference between the gradient and the final gradient of the current driving condition sequence should be less than the first percentage; and the control segment duration should be less than the first time threshold.

[0048] Based on the above, the specific method for deducing driving conditions using the Markov chain Monte Carlo method is as follows: Set initial state Initially, the vehicle speed, acceleration, and gradient are all set to 0, representing the vehicle starting from a flat road surface. A Monte Carlo method is used to simulate and generate a sequence of driving conditions, repeatedly checking the new state. After each update of the driving state, it is necessary to check whether the length of the driving condition sequence meets the expected value. Because the Markov chain Monte Carlo algorithm simulation has a randomness, the last state of the simulation... It may not be in an idling state. In this case, the final state needs to be determined. If the final state is not an idling state, a parking segment needs to be selected from the processed driving condition database and spliced ​​together. After the state, a complete sequence of working conditions from zero to zero is formed.

[0049] The selection of parking segments is based on the principle of minimizing the overall characteristic deviation. First, the final state is considered. The analysis is performed based on the principle of state partitioning, and the results are obtained. The speed, acceleration, and gradient of the point are measured. Then, a driving segment is selected from the processed driving segment library. The difference between the initial speed and the final speed of the current driving condition sequence should be less than 1 km / h, and the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than 0.1 m / s². 2 The difference between the gradient and the gradient at the end of the current driving condition sequence should be less than 0.5%, and the duration of the control segment should be less than 30 seconds. If multiple segments meet the above conditions, calculate the average speed, average gradient, speed standard deviation, and gradient standard deviation for each candidate segment, and dimensionlessize these four evaluation indicators. Calculate the sum of squares of the errors between the four evaluation indicators of each candidate segment and the average index of the overall data using the least squares method. Select the segment with the smallest error as the parking segment and convert it into a state sequence, then splice it into the current driving condition sequence.

[0050] S4. Establish working condition evaluation indicators that include working condition characteristic parameters, and construct working condition characteristic objective functions; This step specifically includes: S41. Calculate speed-related and gradient-related characteristic parameters; speed-related characteristic parameters: average speed, acceleration time ratio, deceleration time ratio, constant speed time ratio, idling time ratio, driving speed standard deviation, average acceleration of acceleration segment, average deceleration of deceleration segment, acceleration standard deviation, deceleration standard deviation, and the percentage of the top five speed intervals. The percentage of the top five speed intervals refers to dividing the speed intervals at 5km / h intervals, calculating the proportion of each speed interval in the total driving condition duration, and statistically analyzing the top five speed intervals with the largest percentages as driving condition evaluation indicators.

[0051] Slope-related characteristic parameters: slope standard deviation, average uphill slope, average downhill slope, uphill ratio, and downhill ratio. A total of 20 characteristic parameters were obtained.

[0052] S42. Calculate the weights of each feature parameter using the feature contribution rate, and construct the objective function for the working condition features.

[0053] It should be noted that S42 specifically includes: S421. Standardize the calculated speed and slope characteristic parameters; S422. Train the model using the random forest algorithm and calculate the number of times the feature splits in the decision tree; S423. Based on the calculation results, derive the importance score for each feature. .

[0054] S424. Calculate the weight of each feature parameter based on the feature importance score; where, feature... weights This indicates the relative importance of the feature in the objective function; In the formula, The sum of the importance scores for all features, where n is the total number of feature parameters; S425. Using the calculated weights, construct a working condition characteristic objective function, wherein the objective function is used to measure the deviation between the constructed working condition sequence and the actual working condition sequence; objective function

[0055] In the formula, For characteristic parameters The weights are denoted by n, where n is the total number of feature parameters. These are the constructed operating condition sequence and the actual operating condition sequence, respectively. To obtain from the operating condition sequence The characteristic parameters calculated in The value, To obtain from the operating condition sequence The characteristic parameters calculated in The value of .

[0056] S5. Based on the objective function of working condition characteristics, the generated working condition sequence is iteratively optimized using an elite adaptive genetic algorithm until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal working condition sequence.

[0057] It should be noted that S5 specifically includes: S51. Select a set number of working condition sequences as the initial population, and calculate the objective function value for each working condition sequence; S52. Calculate the fitness value of the working condition sequence based on the objective function value, select the top N excellent individuals to establish an elite population, and establish a normal population for the remaining individuals, prioritizing the optimization of elite individuals; S53. Establish a crossover operator, select an individual from the elite population and an individual from the ordinary population, perform the crossover operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. In this step, a crossover operator is established. One individual is selected from the elite population, and another individual is selected from the ordinary population. The two parent individuals are then crossovered. A positive integer L is randomly generated as the crossover length between the two parent individuals. The crossover position is then determined. The crossover process ensures that the state after crossover always satisfies the state transition probability matrix. The fitness value of the offspring is calculated; if the offspring's fitness value is less than that of its parent, it is discarded.

[0058] S54. Establish a mutation operator, randomly select an individual from the elite population to perform mutation operation, generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. Establish a mutation operator, randomly select a parent individual from the elite population to perform the mutation operation, randomly select a position on the parent's working condition sequence to deduce a new driving condition, until the working condition duration is met and the final state is idling, calculate the fitness value of the offspring, and discard the offspring if the fitness value of the offspring is less than that of the parent.

[0059] S55. A new generation of population is formed through crossover and mutation. Individuals with fitness below the set value are removed. The top N elite individuals are iteratively optimized until the objective function value of the working condition sequence meets the preset target value, thus obtaining the optimal working condition sequence.

[0060] A new generation of individuals is formed through crossover and mutation. Within the new population, individuals with high similarity exist, and those with low fitness need to be removed. The process iteratively selects the top N operating condition sequences as elite individuals, and continues to optimize these elite individuals through crossover and mutation until the comprehensive index of the absolute deviation between the characteristic parameters of the operating condition sequence and the original data meets the target value.

[0061] The following is an example of a 4×2 tractor-trailer delivery vehicle operating in a Wuhan-Anqing section, including the following steps: Detailed scenario data collection and processing: Vehicles operating between Wuhan and Anqing are selected through a data cloud analysis platform. Relying on onboard terminal equipment and the vehicle's operating system, CAN bus information is continuously collected. Real-time vehicle operation data, measured in seconds, is uploaded to the cloud platform via wireless communication. The remote monitoring data platform center uses intelligent technology to analyze and organize massive amounts of data in real time, enabling remote monitoring, management, and dispatching of vehicles. Data on all 4×2 tractors equipped with a specific engine operating on the Wuhan-Anqing section is extracted. Based on the operating conditions, the full load mass is set to (42t, 52t), and full load operation data is filtered to design a full load operation curve. Due to communication issues, the original data suffers from duplicate sampling and data gaps. Duplicate sampling accounts for 4.12% of the total data, and missing data accounts for 1.21%.

[0062] The acceleration statistics of the database conform to a normal distribution, such as... Figure 2 As shown in the figure. Therefore, outliers in acceleration data can be identified using the 3σ anomaly detection method, and subsequently, outliers in speed data can be identified. The distribution characteristics of acceleration change significantly with speed. Therefore, an acceleration threshold is set according to the vehicle's power performance to identify outliers in the low-speed range. For the high-speed range, an anomaly filtering method based on the 3σ principle, dividing multiple speed intervals, is designed. The anomaly detection results are shown in the figure. Figure 3 As shown, blue represents normal data, while red represents abnormal data in different speed ranges.

[0063] Because AMT slope sensor sampling is low-frequency, the effective values ​​of low-frequency AMT slope sensor measurements can be identified using altitude and speed data, such as... Figure 4 The red circles shown represent valid slope data identified by the recognition algorithm.

[0064] Due to signal interference, some data transmitted by the sensors is lost or resampled. Resampled data is directly deleted. For missing data, if the missing time is short, interpolation can be used to fill in the missing values. If the data loss time is too long, the missing segment is recorded and skipped when calculating the state transition probability matrix. Missing values ​​do not affect the constructed operating condition results. Due to the complexity of the tractor's driving conditions, the sampled transmission signal may experience disturbances, resulting in spikes or sharp points in the collected driving data. This data needs to be processed. A Butterworth filter can be used for digital filtering. This filter can achieve zero phase shift of the original signal. Through analysis of the vehicle speed sampling frequency and power spectral density, a cutoff frequency of 0.5Hz and a second-order filter are determined for the vehicle speed filter.

[0065] State division and transition probability matrix calculation: States are divided based on vehicle speed, acceleration, and gradient range. The speed step size is 0.1 m / s, resulting in 315 speed states. The acceleration step size is 0.05 m / s. 2 A total of 45 acceleration states were obtained, with a slope step size of 0.1%, resulting in 170 slope states. These included vehicle speeds between 0 and 0.1 m / s and accelerations between -0.05 and 0.05 m / s². 2 A gradient between -0.1% and 0.1% is considered an idling state. Each state consists of vehicle speed, acceleration, and gradient. The processed data is labeled with states, resulting in 163,802 valid states. The transition frequency of each state is summarized based on the temporal state transitions, such as... represent Transferred to For missing labeled segments, they need to be skipped when summarizing the state transition probability matrix. The transition frequencies of all states are then summed and calculated. TPM .

[0066] Markov Chain Working Condition Construction: A Record In the idling state, the Markov chain Monte Carlo (MCMC) simulation principle is used to find the next state. First, it is necessary to calculate... The cumulative probability of this row in TPM :

[0067] Generate a random number k in the interval (0,1], where k is the same as... Each cumulative probability element in the data is compared to determine which interval the random number k falls into, and the smallest element greater than the random data k is found to determine the next driving state. This step is repeated until the final state of the generated driving condition sequence is idling and the length meets the specified duration of 1800 seconds. If the final state is not idling, a parking segment needs to be spliced ​​from the original data segment library to ensure that the difference between the initial speed of the driving segment and the final speed of the current driving condition sequence is less than 1 km / h, and the difference between the acceleration and the final acceleration of the current driving condition sequence is less than 0.1 m / s². 2 The slope difference between the current working condition sequence and the end slope is less than 0.5%, and the control segment duration is less than 30 seconds. The working condition construction flowchart is as follows: Figure 5 As shown, a total of 300 working conditions are generated. The working condition sequence can be represented as follows:

[0068] Design working condition evaluation indicators: Corresponding working condition evaluation indicators are calculated based on the generated working condition results. Since the speed range of semi-trailer tractors is relatively fixed, the percentage of the top five values ​​in each speed range is added. This application adds a design for slope, including slope standard deviation, average uphill slope, average downhill slope, uphill percentage, and downhill percentage. When calculating the deviations between the constructed working conditions and the various indicators in the database, the working condition sequence needs to be parsed into speed, acceleration, and slope interval values ​​according to the state division rules. The median value of each interval is taken as the parsed value for that point, thus obtaining the three-dimensional numerical values ​​of the working condition sequence.

[0069] The elite adaptive genetic algorithm was used to specifically optimize the working condition sequences: First, the comprehensive deviation value of the initial 300 working condition sequences was calculated, among which the smallest comprehensive deviation was 12.8%. The speed-gradient working condition curve is shown below. Figure 6 As shown. The comprehensive bias is used as the fitness function of the genetic algorithm. The five best working conditions are selected as elite individuals, and the remaining 295 are ordinary individuals. One elite individual is selected each time a crossover operation is performed. And an ordinary individual At the same time and Cross-concatenation of work condition segments of a certain length is selected, and the concatenation points still satisfy the state transition probability matrix. Two new work condition sequences are generated after concatenation. The fitness function value of the newly generated work condition sequence is calculated. If the fitness value of the newly generated work condition sequence is greater than that of the parent individual, the individual is retained. For the mutation operator, a work condition sequence is randomly selected from the elite individuals. After a certain work condition point, the Monta Carlo algorithm is used to simulate and generate a new work condition sequence. Similarly, if the fitness value of the newly generated work condition sequence is greater than that of the parent individual, the mutated individual is retained. A new generation of population is formed through crossover and mutation, and the elite population is directly replicated to the new generation. Individuals with high similarity exist in the new population; individuals with low fitness need to be removed. The crossover rate of the genetic algorithm is set to 0.8, and the mutation rate to 0.01. This process is iterated until a satisfactory work condition sequence is generated. In this example, the optimal comprehensive error of the initially generated 300 work condition sequences is 8.6%. By using the genetic algorithm to accelerate the convergence of work condition errors, after 200 iterations, the final optimal work condition sequence error is 2.2%. The work condition optimization process based on the genetic algorithm is as follows: Figure 7 As shown, the optimized operating condition curve is as follows: Figure 8 The characteristic values ​​of the operating conditions are shown in Table 1. It can be seen that the optimized operating conditions and the elite individuals share many similar operating condition data, such as... Figure 8 Around 1350s, some velocity gradient data and Figure 6 Similar around the 1500s.

[0070] Table 1: Operating Condition Characteristic Values ​​Table

[0071] To verify whether this operating condition can reflect real driving data, a drum test was used to verify the accuracy of the driving condition. The drum test showed that the driving condition constructed by this method has good following performance, and the drum fuel consumption differs from the actual fuel consumption by 2.33%, which can be used as the driving condition for 6×4 tractor vehicles on this route.

[0072] like Figure 9 As shown, this embodiment of the invention also provides a tractor speed-gradient driving condition construction system, including: Data acquisition module: Used to collect real driving data of the tractor in the target scenario using the vehicle terminal equipment and upload it to the cloud platform, and obtain the collected raw data in batches through the cloud platform; Data processing module: used to perform visual inspection and statistical feature analysis on raw data, use rule-based outlier identification method to remove outlier data, and use filtering algorithm to reduce noise in the data; The working condition generation module is used to discretize the processed vehicle running data per second into state variables containing vehicle speed, acceleration and road slope based on the Markov property of the operating conditions, calculate the state transition probability matrix, and generate several working condition sequences using the Markov chain Monte Carlo method. Operating condition evaluation module: used to establish operating condition evaluation indicators that include operating condition characteristic parameters and construct operating condition characteristic objective functions; Operating condition optimization module: Based on the operating condition characteristic objective function, it uses an elite adaptive genetic algorithm to iteratively optimize the generated operating condition sequence until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal operating condition sequence.

[0073] In some embodiments, the data acquisition module is specifically used to collect data on the tractor in the target scenario; the collected data includes time, vehicle speed, gradient, load, altitude, engine speed and torque parameters; and vehicle operation data for specific areas or routes are filtered through a cloud platform.

[0074] In some embodiments, the data processing module includes a visualization inspection unit, an anomaly detection unit, and a noise reduction processing unit; The visualization inspection unit is used to perform visual inspection of the raw data and detect outliers in sampling interval, vehicle speed distribution, acceleration distribution, slope distribution and load distribution. The anomaly detection unit is used to identify anomalies in acceleration data using the 3σ anomaly detection algorithm, filter anomalies in vehicle speed data, and use interpolation methods to smooth out missing values ​​and acceleration data that exceed the threshold. The noise reduction processing unit is used to perform noise reduction processing on vehicle speed data using a second-order Butterworth low-pass filter with a cutoff frequency set to a set value.

[0075] In some embodiments, the working condition generation module includes a state variable processing unit, a transition probability calculation unit, and a working condition sequence simulation unit; The state variable processing unit is used to discretize vehicle speed, acceleration, and road slope into state variables and divide the state intervals. The transition probability calculation unit is used to calculate the transition frequency of each state in the time series and generate the state transition probability matrix. The operating condition sequence simulation unit is used to simulate and generate operating condition sequences using the Markov chain Monte Carlo method. If the last state of the simulation is not the idling state, a parking segment that meets the conditions is selected and spliced ​​into the current state to form a complete operating condition sequence from zero to zero.

[0076] In some embodiments, the working condition sequence simulation unit is specifically used to generate a working condition sequence using the Markov chain Monte Carlo method; after the MCMC simulation is completed, the final state of the generated working condition sequence is checked; if the final state is not an idling state, the final state is parsed to obtain the vehicle speed, acceleration, and gradient of the final state; segments that meet the set conditions are selected from the segment library as candidate segments; the evaluation index of the candidate segments is calculated and dimensionlessized; the sum of squares of the average index error between each candidate segment and the overall data is calculated, and the segment with the smallest error is selected as the parking segment; the selected parking segment is spliced ​​to the end of the current sequence to ensure that the spliced ​​sequence meets the requirements of the state transition probability matrix; after splicing, a complete cyclic working condition sequence from zero speed to zero speed is generated; wherein the set conditions include: The difference between the initial speed and the final speed of the current driving condition sequence should be less than the first speed threshold; the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than the first acceleration threshold; the difference between the gradient and the final gradient of the current driving condition sequence should be less than the first percentage; and the control segment duration should be less than the first time threshold.

[0077] In some embodiments, the working condition evaluation module includes a feature parameter calculation unit and an objective function construction unit; The feature parameter calculation unit is used to calculate speed-type and slope-type feature parameters, including average speed, acceleration time ratio, deceleration time ratio, slope standard deviation, average uphill slope, and average downhill slope. The objective function construction unit is used to calculate the weights of each feature parameter using the feature contribution and construct the working condition feature objective function.

[0078] In some embodiments, the objective function construction unit includes a calculation processing submodule, an importance score calculation submodule, a weight calculation submodule, and an objective function construction submodule; The calculation and processing submodule is used to standardize the calculated speed and slope characteristic parameters; The importance score calculation submodule is used to train the model using the random forest algorithm, calculate the number of times a feature splits in the decision tree, and derive the importance score for each feature based on the calculation results. .

[0079] The weight calculation submodule is used to calculate the weight of each feature parameter based on the feature importance score; where, feature weights This indicates the relative importance of the feature in the objective function; In the formula, The sum of the importance scores for all features, where n is the total number of feature parameters; The objective function construction submodule is used to construct an objective function for working condition characteristics using the calculated weights. The objective function is used to measure the deviation between the constructed working condition sequence and the actual working condition sequence. objective function

[0080] In the formula, For characteristic parameters The weights are denoted by n, where n is the total number of feature parameters. These are the constructed operating condition sequence and the actual operating condition sequence, respectively. To obtain from the operating condition sequence The characteristic parameters calculated in The value, To obtain from the operating condition sequence The characteristic parameters calculated in The value of .

[0081] In some embodiments, the working condition optimization module is specifically used to select a set number of working condition sequences as an initial population and calculate the objective function value of each working condition sequence; calculate the fitness value of the working condition sequence based on the objective function value, select the top N excellent individuals to establish an elite population, and the remaining individuals to establish a normal population, prioritizing the optimization of elite individuals; establish a crossover operator, select an individual from the elite population and an individual from the normal population, perform a crossover operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, it is retained; establish a mutation operator, randomly select an individual from the elite population to perform a mutation operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, it is retained; form a new generation population through crossover and mutation, delete individuals with fitness values ​​lower than a set value, iteratively optimize the top N elite individuals until the objective function value of the working condition sequence meets the preset target value, and obtain the optimal working condition sequence.

[0082] This invention also provides an electronic device, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus. The communication bus can be used for information transmission between the electronic device and sensors. The processor can call logical instructions in memory to execute the following methods: S1. Use the on-board terminal device to collect real driving data of the tractor in the target scenario, and obtain the collected raw data in batches through the cloud platform; S2. Perform visualization inspection and statistical feature analysis on the raw data, use a rule-based outlier identification method to remove abnormal data, and use a filtering algorithm to denoise the data; S3. Based on the Markov property of the operating conditions, discretize the processed vehicle running data per second into state variables containing vehicle speed, acceleration, and road slope, calculate the probability matrix of each state transition, and use the Markov chain Monte Carlo method to generate several operating condition sequences; S4. Establish an operating condition evaluation index containing operating condition feature parameters and construct an operating condition feature objective function; S5. Based on the operating condition feature objective function, use an elite adaptive genetic algorithm to iteratively optimize the generated operating condition sequences until the comprehensive deviation value reaches the preset target value, and obtain the optimal operating condition sequence.

[0083] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0084] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a tractor speed-gradient driving condition, characterized in that, include: S1. Use on-board terminal equipment to collect real driving data of the tractor in the target scenario, and obtain the collected raw data in batches through the cloud platform; S2. Visualize and statistically analyze the raw data, use rule-based outlier identification to remove outlier data, and use filtering algorithms to reduce noise in the data. S3. Based on the Markov property of the operating conditions, the processed vehicle operating data per second is discretized into state variables including vehicle speed, acceleration and road slope. The state transition probability matrix is ​​calculated, and several operating condition sequences are generated using the Markov chain Monte Carlo method. S4. Establish working condition evaluation indicators that include working condition characteristic parameters, and construct working condition characteristic objective functions; S5. Based on the objective function of working condition characteristics, the generated working condition sequence is iteratively optimized using an elite adaptive genetic algorithm until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal working condition sequence.

2. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 1, characterized in that, S1 specifically includes: S11. Collect data on the tractor in the target scenario; the collected data includes time, vehicle speed, gradient, load, altitude, engine speed and torque parameters. S12. Filter vehicle operation data for specific areas or routes through the cloud platform.

3. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 1, characterized in that, S2 specifically includes: S21. Visualize the raw data to detect outliers in sampling interval, vehicle speed distribution, acceleration distribution, slope distribution, and load distribution; S22. The 3σ anomaly detection algorithm is used to identify anomalies in the acceleration data and to filter anomalies in the vehicle speed data. Interpolation methods are used to smooth out missing values ​​and acceleration data that exceed the threshold. S23. Use a second-order Butterworth low-pass filter with a cutoff frequency of the set value to perform noise reduction processing on the vehicle speed data.

4. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 1, characterized in that, S3 specifically includes: S31. Discretize the vehicle speed, acceleration, and road slope into state variables, and divide the state intervals. S32. Calculate the state transition frequency in the time series and generate the state transition probability matrix; S33. Use the Markov chain Monte Carlo method to simulate and generate a working condition sequence. If the last state of the simulation is not the idling state, select the parking segment that meets the conditions and splice it into the current state to form a complete working condition sequence from zero to zero.

5. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 4, characterized in that, S33 specifically includes: S331. Use the Markov chain Monte Carlo method to simulate and generate the working condition sequence; S332. After the MCMC simulation is completed, check the final state of the generated operating condition sequence; S333. If the final state is not the idling state, parse the final state to obtain the vehicle speed, acceleration and gradient of the final state; S334. Select segments that meet the set conditions from the segment library as candidate segments; S335. Calculate the evaluation index of the candidate segments and make it dimensionless; S336. Calculate the sum of squares of the average index error between each candidate segment and the overall data, and select the segment with the smallest error as the parking segment. S337. Patch the selected parking segment to the end of the current sequence to ensure that the spliced ​​sequence meets the requirements of the state transition probability matrix; S338. After splicing, a complete cyclic operating condition sequence from zero speed to zero speed is generated; the set conditions include: The difference between the initial speed and the final speed of the current driving condition sequence should be less than the first speed threshold; the difference between the acceleration and the final acceleration of the current driving condition sequence should be less than the first acceleration threshold; the difference between the gradient and the final gradient of the current driving condition sequence should be less than the first percentage; and the control segment duration should be less than the first time threshold.

6. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 1, characterized in that, S4 specifically includes: S41. Calculate speed-related and slope-related characteristic parameters, including average speed, acceleration time ratio, deceleration time ratio, slope standard deviation, average uphill slope, and average downhill slope. S42. Calculate the weights of each feature parameter using the feature contribution rate, and construct the objective function for the working condition features.

7. The method for constructing the speed-gradient driving condition of a tractor vehicle according to claim 6, characterized in that, S42 specifically includes: S421. Standardize the calculated speed and slope characteristic parameters; S422. Train the model using the random forest algorithm and calculate the number of times the feature splits in the decision tree; S423. Based on the calculation results, derive the importance score for each feature. ; S424. Calculate the weight of each feature parameter based on the feature importance score; where, feature... weights This indicates the relative importance of the feature in the objective function; In the formula, The sum of the importance scores for all features, where n is the total number of feature parameters; S425. Using the calculated weights, construct a working condition characteristic objective function, wherein the objective function is used to measure the deviation between the constructed working condition sequence and the actual working condition sequence; objective function In the formula, For characteristic parameters The weights are denoted by n, where n is the total number of feature parameters. These are the constructed operating condition sequence and the actual operating condition sequence, respectively. To obtain from the operating condition sequence The characteristic parameters calculated in The value, To obtain from the operating condition sequence The characteristic parameters calculated in The value of .

8. The method for constructing the speed-gradient driving condition of a tractor according to claim 1, characterized in that, S5 specifically includes: S51. Select a set number of working condition sequences as the initial population, and calculate the objective function value for each working condition sequence; S52. Calculate the fitness value of the working condition sequence based on the objective function value, select the top N excellent individuals to establish an elite population, and establish a normal population for the remaining individuals, prioritizing the optimization of elite individuals; S53. Establish a crossover operator, select an individual from the elite population and an individual from the ordinary population, perform the crossover operation to generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. S54. Establish a mutation operator, randomly select an individual from the elite population to perform mutation operation, generate a new working condition sequence, and calculate the objective function value of the new sequence. If the objective function value of the new sequence is better than that of the parent generation, then retain it. S55. A new generation of population is formed through crossover and mutation. Individuals with fitness below the set value are removed. The top N elite individuals are iteratively optimized until the objective function value of the working condition sequence meets the preset target value, thus obtaining the optimal working condition sequence.

9. A tractor speed-gradient driving condition construction system, characterized in that, include: Data acquisition module: Used to collect real driving data of the tractor in the target scenario using the vehicle terminal equipment and upload it to the cloud platform, and obtain the collected raw data in batches through the cloud platform; Data processing module: used to perform visual inspection and statistical feature analysis on raw data, use rule-based outlier identification method to remove outlier data, and use filtering algorithm to reduce noise in the data; The working condition generation module is used to discretize the processed vehicle running data per second into state variables containing vehicle speed, acceleration and road slope based on the Markov property of the operating conditions, calculate the state transition probability matrix, and generate several working condition sequences using the Markov chain Monte Carlo method. Operating condition evaluation module: used to establish operating condition evaluation indicators that include operating condition characteristic parameters and construct operating condition characteristic objective functions; Operating condition optimization module: Based on the operating condition characteristic objective function, it uses an elite adaptive genetic algorithm to iteratively optimize the generated operating condition sequence until the comprehensive deviation value reaches the preset target value, thus obtaining the optimal operating condition sequence.

10. An electronic device, characterized in that, The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores computer program instructions executable by the at least one processor, the computer program instructions being executed by the at least one processor to enable the at least one processor to perform the tractor speed-gradient driving condition construction method as described in any one of claims 1 to 8.

Images