Driving style identification method and system and electronic equipment

Abstract

The invention relates to the technical field of driving style recognition, in particular to a driving style recognition method and system and electronic equipment. The method comprises the following steps: dynamically receiving vehicle historical driving data from a plurality of terminals; performing working condition classification on the historical driving data of the vehicle, and obtaining historical driving data corresponding to a plurality of target working conditions; based on the historical driving data corresponding to each target working condition, obtaining a driving characteristic data set corresponding to each target working condition; based on a fuzzy C-means clustering algorithm, clustering processing is carried out on the driving characteristic data corresponding to each target working condition, and a first clustering center corresponding to each target working condition is obtained; and sending the first clustering center corresponding to each target working condition to the target terminal, so that the target terminal performs driving style recognition on the target vehicle based on the collected multiple groups of real-time driving data of the target vehicle and the first clustering center corresponding to each target working condition. According to the invention, the accuracy and practicability of driving style identification can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of driving style recognition, and particularly to a driving style recognition method, system, and electronic device. Background Art

[0002] In recent years, with the rapid development of intelligent driving vehicles, the research on driver characteristics has shown great potential in achieving human-machine collaborative driving and personalized customization. The existing technologies mainly identify driving styles through three methods: questionnaire surveys, rule-based methods, and intelligent algorithms. These methods aim to adjust the driving mode of the vehicle according to the driver's driving style to improve driving comfort.

[0003] However, these technologies have encountered several challenges in practical applications: Questionnaire surveys are impractical due to the huge workload and data error problems; Rule-based methods are difficult to comprehensively capture the complexity of driving styles; While intelligent algorithms, although excellent in recognition effects, are limited by the lack of sufficient labeled samples, resulting in heavy manual annotation work and being vulnerable to subjective factors. These limiting factors affect the accuracy and practicality of the existing technologies in driving style recognition. Summary of the Invention

[0004] Based on the above technical problems, the present invention provides a driving style recognition method, system, and electronic device.

[0005] According to one aspect of the present invention, there is provided a driving style recognition method applied to the cloud, including: dynamically receiving vehicle historical driving data from multiple terminals; classifying the vehicle historical driving data by working conditions to obtain historical driving data corresponding to several target working conditions respectively; obtaining a driving feature dataset corresponding to each target working condition based on the historical driving data corresponding to each target working condition; performing clustering processing on the driving feature dataset corresponding to each target working condition based on the fuzzy C-means clustering algorithm to obtain a first clustering center corresponding to each target working condition; sending the first clustering center corresponding to each target working condition to the target terminal, so that the target terminal performs driving style recognition on the target vehicle based on the collected multiple sets of real-time driving data of the target vehicle and the first clustering center corresponding to each target working condition.

[0006] According to the driving style recognition method of one aspect of the present invention, the vehicle historical driving data at least includes vehicle speed information of multiple vehicles at different time points within a unit collection time, and the target working conditions include low vehicle speed working conditions and high vehicle speed working conditions; classifying the vehicle historical driving data by working conditions to obtain historical driving data corresponding to several target working conditions respectively, including: obtaining first historical driving data corresponding to the low vehicle speed working condition and second historical driving data corresponding to the high vehicle speed working condition based on the vehicle historical driving data and the vehicle speed information.

[0007] A driving style recognition method according to an aspect of the present invention, based on historical driving data corresponding to each target working condition, obtains a driving feature data set corresponding to each target working condition, including: based on the first historical driving data, obtaining a first driving feature data set for low vehicle speed working conditions, and each set of first driving feature data in the first driving feature data set includes the average value of acceleration, the standard deviation of acceleration, the average value of positive acceleration change rate, the average value of negative acceleration change rate, the standard deviation of acceleration change rate, the average value of throttle opening, and the maximum value of throttle opening; based on the second historical driving data, obtaining a second driving feature data set for high vehicle speed working conditions, and each set of second driving feature data in the second driving feature data set includes the standard deviation of throttle opening, the average value of the change rate of positive throttle opening, the average value of the change rate of negative throttle opening, and the standard deviation of the change rate of throttle opening.

[0008] A driving style recognition method according to an aspect of the present invention, based on the fuzzy C-means clustering algorithm, performs clustering processing on the driving feature data set corresponding to each target working condition to obtain a first clustering center corresponding to each target working condition, including: performing normalization processing on the driving feature data set; setting the number of clusters to 2, dividing the driving style into aggressive and mild types, and initializing the membership matrix; calculating and updating the clustering center of each cluster until a preset iteration stop condition is met; outputting the first clustering center corresponding to each target working condition.

[0009] According to still another aspect of the present invention, there is provided a driving style recognition method applied to a target terminal, including: receiving first clustering centers corresponding to a plurality of target working conditions from the cloud; collecting multiple groups of real-time driving data of the target vehicle; based on the multiple groups of real-time driving data and the first clustering center corresponding to each target working condition, performing driving style recognition on the target vehicle; wherein, the first clustering center is obtained by the cloud based on dynamically receiving vehicle historical driving data from multiple terminals; classifying the vehicle historical driving data by working conditions to obtain historical driving data corresponding to a plurality of target working conditions respectively; based on the historical driving data corresponding to each target working condition, obtaining a driving feature data set corresponding to each target working condition; and performing clustering processing on the driving feature data corresponding to each target working condition based on the fuzzy C-means clustering algorithm.

[0010] In addition, a driving style recognition method according to still another aspect of the present invention, based on multiple groups of real-time driving data and the first clustering center corresponding to each target working condition, performs driving style recognition on the target vehicle, including: Judging whether the real-time working condition corresponding to the multiple groups of real-time driving data belongs to one of a plurality of target working conditions; When it is determined that the situation belongs to one of several target working conditions, based on multiple sets of real-time driving data and the corresponding first clustering center of each target working condition, the driving style of the target vehicle is identified.

[0011] In addition, according to another aspect of the driving style recognition method of the present invention, based on multiple sets of real-time driving data and the first clustering center of each target working condition, the driving style of the target vehicle is identified, including: based on each set of real-time driving data, obtaining the real-time driving feature data of the corresponding real-time working condition; determining the current driving style of the target vehicle based on at least each set of real-time driving feature data and the first clustering center of each target working condition.

[0012] In addition, according to another aspect of the driving style recognition method of the present invention, the method further includes: Based on the current driving style of the target vehicle, adjusting the throttle opening of the target vehicle.

[0013] In addition, according to another aspect of the driving style recognition method of the present invention, the method further includes: When it is recognized that the target vehicle is in the cruise mode, maintaining the throttle opening of the target vehicle.

[0014] According to another aspect of the present invention, there is provided a driving style recognition system, which is applied to the cloud and includes: A driving data acquisition module, configured to dynamically receive the vehicle historical driving data from multiple terminals; A driving feature acquisition module, configured to classify the working conditions of the vehicle historical driving data and obtain the historical driving data corresponding to several target working conditions respectively; A target working condition driving data acquisition module, configured to obtain the driving feature data set corresponding to each target working condition based on the historical driving data corresponding to each target working condition; A clustering center acquisition module, configured to perform clustering processing on the driving feature data corresponding to each target working condition based on the fuzzy C-means clustering algorithm, and obtain the first clustering center corresponding to each target working condition; A driving style recognition module, configured to send the first clustering center corresponding to each target working condition to the target terminal, so that the target terminal identifies the driving style of the target vehicle based on multiple sets of real-time driving data of the target vehicle collected and the first clustering center corresponding to each target working condition.

[0015] According to yet another aspect of the present invention, there is provided a driving style recognition system, which is applied to the target terminal and includes: The clustering center receiving module is used to receive the first clustering centers corresponding to a number of target working conditions from the cloud; wherein, the first clustering centers are obtained by the cloud based on dynamically receiving the historical driving data of vehicles from multiple terminals, classifying the working conditions of the historical driving data of vehicles to obtain the historical driving data corresponding to a number of target working conditions, obtaining the driving feature data set corresponding to each target working condition based on the historical driving data corresponding to each target working condition, and performing clustering processing on the driving feature data corresponding to each target working condition based on the fuzzy C-means clustering algorithm; The real-time driving data acquisition module is used to acquire multiple groups of real-time driving data of the target vehicle; The driving style recognition module is used to recognize the driving style of the target vehicle based on multiple groups of real-time driving data and the first clustering center corresponding to each target working condition.

[0016] According to another aspect of the present invention, there is provided an electronic device, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the driving style recognition method as described above.

[0017] For the driving style recognition method, system and electronic device provided by the present invention, the cloud dynamically receives the historical driving data of vehicles from multiple terminals, classifies the working conditions, obtains the driving feature data set of the target working conditions, uses the fuzzy C-means clustering algorithm to process the driving feature data set to dynamically obtain the first clustering center corresponding to each target working condition, and sends the first clustering center corresponding to each target working condition to the target terminal, so that the target terminal combines the real-time driving data of the target vehicle to recognize the driving style of the target vehicle. This method can recognize complex patterns of driving styles without relying on predefined rules, avoids the subjectivity of questionnaire surveys and the limitations of rule-based methods, reduces the dependence on labeled samples, reduces the workload of manual labeling, and realizes the efficient collaborative work between the cloud and the terminal, thereby significantly improving the accuracy and practicality of driving style recognition. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is one of the flow diagrams of the driving style recognition method provided by the present invention.

[0020] Figure 2 It is the second flow diagram of the driving style recognition method provided by the present invention.

[0021] Figure 3 It is the third flow schematic diagram of the driving style recognition method provided by the present invention.

[0022] Figure 4 It is one of the structural schematic diagrams of the driving style recognition system provided by the present invention.

[0023] Figure 5 It is the second structural schematic diagram of the driving style recognition system provided by the present invention. Specific Embodiments

[0024] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts fall within the scope of protection of the present invention.

[0025] In recent years, fuel saving, environmental protection, comfort and safety of vehicles in the commercial vehicle field have attracted wide attention. With the development of commercial vehicle technology, vehicles can adjust the driving mode according to the driver's driving style, and different driving modes correspond to different driving parameters, such as response parameters of throttle opening, braking, shock absorption, etc. Currently, the methods for distinguishing driving styles are mainly divided into three categories: questionnaire surveys, rule-based, and intelligent algorithm-based. Questionnaire surveys involve extremely large workloads and excessive data errors, and do not have application scenarios. The rule-based method mainly selects one or more characteristic parameters as indicators for evaluating driving styles, and this method cannot fully reflect driving styles. Identifying driving styles based on supervised learning algorithms can achieve good results, but in reality, labeled samples are scarce, the workload of manually adding labels is huge, and there is subjectivity.

[0026] Based on the above technical problems, the present invention provides a driving style recognition method. The following combines Figure 1 - Figure 2 to describe the driving style recognition method of the present invention.

[0027] Figure 1 It is one of the flow schematic diagrams of the driving style recognition method provided by the present invention, and this method can be executed by the cloud.

[0028] As Figure 1 shown, according to the driving style recognition method provided by the embodiments of the present invention, it specifically includes the following steps: Step 101, dynamically receive vehicle historical driving data from multiple terminals.

[0029] In one embodiment of the present invention, the cloud refers to a centralized data processing and storage system, which is usually located on a remote server and can be accessed through the Internet, and is responsible for collecting, storing and processing data from different vehicles. A terminal refers to a device connected to a vehicle, which can receive and process vehicle historical driving data from the vehicle, including but not limited to a vehicle-mounted terminal, a vehicle-mounted computer, a smart phone, a tablet computer or other dedicated vehicle-mounted devices. The vehicle historical driving data of multiple terminals covers the actual driving data of different vehicles (such as at least 100 vehicles in various regions across the country), which can cover various working conditions as much as possible, and can ensure the breadth and diversity of samples. The vehicle historical driving data contains driving data related to vehicle driving of multiple vehicles in the past period of time, such as: parameters such as vehicle speed, acceleration, braking force, steering angle, engine speed, throttle opening, as well as time, date and geographic location. Exemplarily, the vehicle is usually equipped with a data transmission device, such as a vehicle-mounted terminal. The vehicle can dynamically collect the vehicle's vehicle historical driving data through the vehicle-mounted data transmission software on the vehicle-mounted terminal, and upload it to the cloud and the vehicle's backend database.

[0030] Step 102: classify the vehicle's historical driving data according to working conditions, and obtain historical driving data corresponding to a number of target working conditions.

[0031] In one embodiment of the present invention, the operating condition may be various situations encountered in actual driving, such as driving in different environments and conditions such as urban congested roads, highways, rural roads, rainy and snowy weather. It may also be the driving state of the vehicle in various actual driving, such as: accelerating, high-speed driving, low-speed driving, and even-speed driving. The target operating condition refers to the operating condition related to driving style recognition. The cloud classifies the collected vehicle historical driving data according to different operating conditions. After completing the operating condition classification, several target operating conditions related to driving style recognition are determined from the classified operating conditions, and the corresponding historical driving data are extracted. These data will be used for subsequent driving style analysis.

[0032] Exemplarily, before classifying the working conditions, the historical driving data of the vehicle is screened to remove abnormal data and retain valid data.

[0033] Step 103: Based on the historical driving data corresponding to each target working condition, a driving characteristic data set corresponding to each target working condition is obtained.

[0034] In an embodiment of the present invention, for each target working condition, according to the corresponding historical driving data, a driving feature data set that can reflect the driving style is further extracted. The driving feature data set refers to a set composed of feature quantities that can reflect the driving style under the target working condition. The driving feature data set includes, but is not limited to, vehicle speed, acceleration, braking force, steering angle, average value of acceleration, standard deviation of acceleration, average value of positive acceleration change rate, average value of negative acceleration change rate, standard deviation of acceleration change rate, average value of throttle opening, maximum value of throttle opening, standard deviation of throttle opening, average value of change rate of positive throttle opening, average value of change rate of negative throttle opening, standard deviation of change rate of throttle opening, etc. By analyzing the driving feature data set, unique patterns and trends of different driving styles can be identified.

[0035] Step 104: Based on the fuzzy C - means clustering algorithm, perform clustering processing on the driving feature data set corresponding to each target working condition to obtain the first clustering center corresponding to each target working condition.

[0036] In an embodiment of the present invention, the fuzzy C - means clustering algorithm is applied to the driving feature data set of each target working condition. This algorithm can group data points into multiple clusters according to the similarity between data points and allows a data point to belong to multiple clusters with different membership degrees. Through this clustering processing, the first clustering center of each target working condition can be determined, that is, the point representing the typical features of the driving style under this working condition. The first clustering center can be regarded as the "prototype" of the driving style under this working condition.

[0037] Exemplarily, deploy a MATLAB Production Server in the cloud, and use the MATLAB Production Server to deploy and train a driving style identification model, which is obtained through the fuzzy C - means clustering algorithm. That is, the first clustering center corresponding to each target working condition is obtained by training the driving style identification model in the cloud. By dynamically receiving the vehicle historical driving data from multiple terminals and continuously updating the vehicle historical driving data, it is ensured that the training data of the model contains valid old data and the latest new data, so as to continuously train and optimize the membership degree and clustering center of the model.

[0038] As new data is continuously added, the cloud will periodically repeat steps 101 to 104 to update and optimize the first clustering center corresponding to each target working condition. This cyclic update mechanism ensures that the model can adapt to new driving data and changing driving patterns. By continuously integrating valid old data and the latest new data, the membership degree and clustering center of the model are continuously trained and optimized. This dynamic data update and model training process ensure the accuracy and reliability of the driving style recognition in step 105.

[0039] Step 105: Send the first cluster center corresponding to each target working condition to the target terminal, so that the target terminal can identify the driving style of the target vehicle at the terminal based on the collected multiple groups of real-time driving data of the target vehicle and the first cluster center corresponding to each target working condition.

[0040] In an embodiment of the present invention, the target terminal refers to a device installed on or connected to the target vehicle, which can receive and process real-time driving data from the target vehicle, including but not limited to in-vehicle devices, in-vehicle computers, smartphones, tablets or other dedicated in-vehicle devices. After completing the clustering process, the cloud will send the first cluster center corresponding to each target working condition to the target terminal. The target terminal collects multiple groups of real-time driving data of the target vehicle in real time, and these data may include parameters related to driving behavior such as vehicle speed, acceleration, and throttle opening. After receiving the first cluster center corresponding to each target working condition, the target terminal combines the real-time driving data with the first cluster center corresponding to each target working condition for analysis, and identifies the driving style of the target vehicle by calculating the similarity or membership degree.

[0041] Exemplarily, the cloud transmits the first cluster center of each target working condition obtained through the MATLAB production server to the target terminal (such as the vehicle controller). After receiving the first cluster center corresponding to each target working condition sent by the cloud, the target terminal uses the real-time driving data and the cluster center of the target vehicle collected in real time, and calculates the similarity or membership degree between the real-time driving data and the first cluster center corresponding to each target working condition through the established real-time driving style identification model (for example, using the fuzzy C-means algorithm), so as to identify the driving style of the target vehicle.

[0042] In summary, according to the technical solution provided by the embodiment of the present invention, the cloud dynamically receives the historical driving data of the vehicle from multiple terminals, classifies the working conditions, obtains the driving feature data set of the target working condition, uses the fuzzy C-means clustering algorithm to process the driving feature data set to dynamically obtain the first cluster center corresponding to each target working condition, and sends the first cluster center corresponding to each target working condition to the target terminal, so that the target terminal combines the real-time driving data of the target vehicle to identify the driving style of the target vehicle. This method can identify complex patterns of driving styles without relying on predefined rules, avoids the subjectivity of questionnaire surveys and the limitations of rule-based methods, reduces the dependence on labeled samples, reduces the workload of manual labeling, realizes the efficient collaborative work between the cloud and the terminal, and thus significantly improves the accuracy and practicality of driving style identification. Further, calculating the first cluster center of each target working condition in the cloud can save the storage space of the vehicle controller, improve the calculation efficiency and the accuracy of the training results.

[0043] Furthermore, the vehicle historical driving data at least includes the vehicle speed information of multiple vehicles at different time points within a unit collection time, and the target driving conditions include low vehicle speed conditions and high vehicle speed conditions.

[0044] Specifically, the vehicle historical driving data set at least covers the vehicle speed data of multiple vehicles at different time points recorded at a fixed time interval (unit collection time).

[0045] Exemplarily, taking 10s as a unit collection time and the sampling frequency as 0.1s, the cloud can receive the vehicle historical driving data of multiple terminals.

[0046] Furthermore, classify the vehicle historical driving data to obtain the historical driving data corresponding to several target driving conditions, including: Based on the vehicle historical driving data and vehicle speed information, obtain the first historical driving data corresponding to the low vehicle speed condition and the second historical driving data corresponding to the high vehicle speed condition.

[0047] Specifically, screen out the data that meets the low vehicle speed condition and high vehicle speed condition from the vehicle historical driving data. Set speed thresholds. For example, being lower than the low vehicle speed threshold is considered as the low vehicle speed condition, and being higher than the high vehicle speed threshold is considered as the high vehicle speed condition. Classify the vehicle historical driving data according to the above threshold classification and the vehicle speed data in the vehicle historical driving data. Classify the data in the vehicle historical driving data with a vehicle speed lower than the low vehicle speed threshold as the first historical driving data (low vehicle speed condition), and classify the data higher than the high vehicle speed threshold as the second historical driving data (high vehicle speed condition).

[0048] Exemplarily, classify the vehicle historical driving data according to the vehicle speed: When the vehicle speed is greater than 60 km / h, it is defined that the vehicle is in the high vehicle speed condition. When the vehicle speed is less than 20 km / h, it is defined that the vehicle is in the low vehicle speed condition. It can also be that when the vehicle speed is less than 20 km / h and the acceleration is greater than 0.05 m / s², it is defined that the vehicle is in the low vehicle speed condition.

[0049] In summary, according to the technical solution provided by the embodiments of the present invention, the first historical driving data (low vehicle speed condition) and the second historical driving data (high vehicle speed condition) corresponding to the two target driving conditions can be accurately extracted from the vehicle historical driving data, thereby refining the analysis of driving styles and improving the recognition accuracy.

[0050] Furthermore, based on the historical driving data corresponding to each target driving condition, obtain the driving feature data set corresponding to each target driving condition, including: Based on the first historical driving data, obtain the first driving feature dataset for low vehicle speed conditions. Each set of first driving feature data in the first driving feature dataset includes the average value of acceleration, the standard deviation of acceleration, the average value of positive acceleration change rate, the average value of negative acceleration change rate, the standard deviation of acceleration change rate, the average value of throttle opening, and the maximum value of throttle opening.

[0051] Specifically, under low vehicle speed conditions, the driving feature data that can reflect the driving style includes the average value of acceleration, the standard deviation of acceleration, the average value of positive acceleration change rate, the average value of negative acceleration change rate, the standard deviation of acceleration change rate, the average value of throttle opening, and the maximum value of throttle opening. The above driving feature data constitutes a set of first driving feature data. According to the first historical driving data, multiple sets of first driving feature data can be obtained through calculation, and these multiple sets of first driving feature data constitute the first driving feature dataset.

[0052] Exemplarily, according to the vehicle speed and throttle opening in the first historical driving data, the first driving feature dataset can be obtained through calculation.

[0053] Based on the second historical driving data, obtain the second driving feature dataset for high vehicle speed conditions. Each set of second driving feature data in the second driving feature dataset includes the standard deviation of throttle opening, the average value of positive throttle opening change rate, the average value of negative throttle opening change rate, and the standard deviation of throttle opening change rate.

[0054] Specifically, under high vehicle speed conditions, the driving feature data that can reflect the driving style includes the standard deviation of throttle opening, the average value of positive throttle opening change rate, the average value of negative throttle opening change rate, and the standard deviation of throttle opening change rate. The above driving feature data constitutes a set of second driving feature data. According to the second historical driving data, multiple sets of second driving feature data can be obtained through calculation, and these multiple sets of second driving feature data constitute the second driving feature dataset.

[0055] Exemplarily, according to the vehicle speed and throttle opening in the second historical driving data, the second driving feature dataset can be obtained through calculation.

[0056] Figure 2 It is the second schematic diagram of the process of the driving style recognition method provided by the present invention.

[0057] As Figure 2 shown, based on the fuzzy C-means clustering algorithm, perform clustering processing on the driving feature dataset corresponding to each target condition to obtain the first clustering center corresponding to each target condition, which specifically includes the following steps: Step 201: Perform normalization processing on the driving feature dataset.

[0058] In an embodiment of the present invention, there are multiple data samples in the driving characteristic dataset corresponding to each target working condition. Each data sample contains a set of driving characteristic data. Each set of driving characteristic data in the driving characteristic dataset corresponding to each target working condition is normalized to convert the data into a value between 0 and 1.

[0059] Exemplarily, the normalization formula is: Where, represents the maximum value of the i-th feature, represents the minimum value of the i-th feature, represents the k-th value of the i-th feature, is the converted value.

[0060] Step 202: Set the number of clustering clusters to 2, divide the driving styles into aggressive and gentle types, and initialize the membership matrix.

[0061] In an embodiment of the invention, using the fuzzy C-means algorithm, setting the number of clustering clusters to 2 means dividing the driving styles into aggressive and gentle types, and each cluster represents a driving style. Initialize the membership matrix. If there are n data samples in the driving characteristic dataset corresponding to each target working condition, and the number of clustering clusters is represented by c, then the membership matrix should be an n×c matrix.

[0062] Exemplarily, each element in the membership matrix represents the membership degree of sample j belonging to cluster center i.

[0063] At the beginning of the algorithm, each element in the membership matrix is initialized to a random value in the interval [0, 1] to represent the initial membership degree of sample j belonging to cluster center i. Subsequently, through normalization processing, it is ensured that the sum of the membership degrees of each row in the membership matrix is equal to 1, thus meeting the requirements of the fuzzy clustering algorithm.

[0064] Step 203: Calculate and update the cluster center of each cluster until the preset iteration stop condition is met.

[0065] Calculate and update the cluster center of each cluster, set a threshold, and stop the iteration when the change in the membership matrix is less than this threshold. Or set the maximum number of iterations, for example, 100 times, and stop the iteration when this number is reached.

[0066] Exemplarily, use the following formula to calculate each cluster center: The membership degree is calculated as follows: Among them, f is the sample blur degree, n is the number of samples, is the sample j belonging to the membership degree of the clustering center i ; is the clustering center i to the sample j distance, is the sample j to the clustering center u distance, and k is the number of clustering centers, that is, the number of clusters.

[0067] Step 204: Output the first clustering center corresponding to each target working condition.

[0068] In an embodiment of the present invention, when the iteration stops and the fuzzy C-means clustering algorithm is completed, the first clustering center corresponding to each target working condition is output.

[0069] Figure 4 is the third schematic flow chart of the driving style recognition method provided by the present invention.

[0070] As Figure 4 shown, according to the driving style recognition method provided by the embodiment of the present invention, it is executed by the target terminal, and the target terminal refers to a device installed on or connected to the target vehicle, and this device can receive and process real-time driving data from the vehicle, including but not limited to in-vehicle devices, in-vehicle computers, smart phones, tablet computers or other dedicated in-vehicle devices. The driving style recognition method specifically includes the following steps.

[0071] Step 401: Receive the first clustering centers corresponding to several target working conditions from the cloud.

[0072] Step 402: Collect multiple groups of real-time driving data of the target vehicle.

[0073] Step 404: Based on multiple groups of real-time driving data and the first clustering center corresponding to each target working condition, perform driving style recognition on the target vehicle; Among them, the first clustering center is based on the cloud dynamically receiving vehicle historical driving data from multiple terminals; classify the vehicle historical driving data by working condition to obtain the historical driving data corresponding to several target working conditions; Based on the historical driving data corresponding to each target working condition, obtain the driving feature data set corresponding to each target working condition; It is obtained by performing clustering processing on the driving feature data corresponding to each target working condition based on the fuzzy C-means clustering algorithm.

[0074] Specifically, the first clustering center refers to the cloud collecting and analyzing the historical driving data of multiple vehicles and using the fuzzy C-means clustering algorithm to process this data to determine the clustering center for each target working condition. These first clustering centers are calculated based on the features in the historical data using the fuzzy C-means clustering algorithm and represent the typical features of each driving style corresponding to each target working condition. The target terminal receives the first clustering centers corresponding to several target vehicle conditions from the cloud and analyzes them in combination with multiple groups of real-time driving data of the target vehicle collected in real time, thereby identifying the driving style of the target vehicle.

[0075] In summary, according to the technical solution provided by the embodiment of the present invention, the cloud dynamically receives the historical driving data from multiple terminals, performs working condition classification, obtains the driving feature data set of the target working condition, uses the fuzzy C-means clustering algorithm to process the driving feature data set to dynamically obtain the first clustering center corresponding to each target working condition, and sends the first clustering center corresponding to each target working condition to the target terminal. The target terminal combines multiple groups of real-time driving data of the target vehicle to identify the driving style of the target vehicle. This method can identify complex patterns of driving styles without relying on predefined rules, avoids the subjectivity of questionnaire surveys and the limitations of rule-based methods, reduces the dependence on labeled samples, reduces the workload of manual labeling, realizes the efficient collaborative work between the cloud and the terminal, and thus significantly improves the accuracy and practicality of driving style recognition. Further, calculating the first clustering center of each target working condition in the cloud can save the storage space of the vehicle controller, improve the calculation efficiency and the accuracy of the training results.

[0076] Further, based on multiple groups of real-time driving data and the first clustering center corresponding to each target working condition, the driving style of the target vehicle is identified, including: Judging whether the real-time working condition corresponding to multiple groups of real-time driving data belongs to one of several target working conditions; In the case of determining that it belongs to one of several target working conditions, based on multiple groups of real-time driving data and the first clustering center of each target working condition, the driving style of the target vehicle is identified.

[0077] Specifically, the real-time driving condition refers to the current driving condition of the target vehicle determined based on the real-time driving data at the current moment. The target terminal (such as the in-vehicle terminal of the vehicle) collects multiple sets of real-time driving data of the target vehicle in real time. These data may include vehicle speed, acceleration, throttle opening, braking force, steering angle, etc. By analyzing multiple sets of real-time driving data, the driving condition corresponding to each set of real-time driving data is determined. When it is determined whether the real-time driving condition corresponding to each set of real-time driving data of the target vehicle belongs to the low-speed driving condition or the high-speed driving condition in the target driving conditions, multiple sets of real-time driving data are combined with the first clustering center corresponding to each target driving condition to identify the driving style of the target vehicle.

[0078] Exemplarily, the target terminal collects multiple sets of real-time driving data of the target vehicle in real time. By analyzing each set of real-time driving data (such as vehicle speed), it can be determined whether the real-time driving condition corresponding to each set of real-time driving data belongs to the low-speed driving condition or the high-speed driving condition in the target driving conditions. When it is determined that the real-time driving condition corresponding to each set of real-time driving data belongs to the low-speed driving condition or the high-speed driving condition, the driving style of the target vehicle can be identified based on multiple sets of real-time driving data and the first clustering center of each target driving condition.

[0079] Further, identifying the driving style of the target vehicle based on multiple sets of real-time driving data and the first clustering center of each target driving condition includes: Based on each set of real-time driving data, obtain the real-time driving feature data of the corresponding real-time driving condition; Determine the current driving style of the target vehicle based on at least each set of real-time driving feature data and the first clustering center of each target driving condition.

[0080] Specifically, the target terminal collects multiple sets of real-time driving data of the target vehicle in real time. Each set of real-time driving data includes key parameters such as vehicle speed, acceleration, and throttle opening. The multiple sets of real-time driving data can be the driving data for a period of time or the driving data at different time points. Each set of real-time driving data corresponds to a real-time driving condition, and the real-time driving condition belongs to the target driving condition. According to the target driving condition corresponding to each real-time driving condition and the corresponding real-time driving data, the real-time driving feature data corresponding to each real-time driving condition is determined. Combine each set of real-time driving feature data with the first clustering center of each target driving condition to determine the current driving style of the target vehicle.

[0081] Exemplarily, obtain the corresponding real-time driving data for the most recent 10 real-time operating conditions (i.e., multiple sets of real-time driving data), and determine whether the real-time operating condition corresponding to each set of real-time driving data belongs to a low-speed operating condition or a high-speed operating condition. If it is determined to belong to a low-speed operating condition or a high-speed operating condition, then calculate the real-time driving characteristic data corresponding to this real-time operating condition based on the real-time driving data corresponding to each real-time operating condition. For example, when the real-time operating condition belongs to a low-speed operating condition, calculate the real-time driving characteristic data (including the average value of acceleration, the standard deviation of acceleration, the average value of the positive acceleration change rate, the average value of the negative acceleration change rate, the standard deviation of the acceleration change rate, the average value of the throttle opening, the maximum value of the throttle opening) based on the real-time driving data corresponding to this real-time operating condition. Similarly, when the real-time operating condition belongs to a high-speed operating condition, calculate the real-time driving characteristic data corresponding to this real-time operating condition based on the real-time driving characteristic data corresponding to this real-time operating condition (such as the standard deviation of the throttle opening, the average value of the change rate of the positive throttle opening, the average value of the change rate of the negative throttle opening, the standard deviation of the change rate of the throttle opening). Statistically analyze the real-time driving characteristic data sets corresponding to the low-speed operating conditions and the real-time driving characteristic data sets corresponding to the high-speed operating conditions among the 10 real-time operating conditions, and calculate the mean value of each characteristic value in the real-time driving characteristic data set corresponding to the low-speed operating condition, and calculate the mean value of each characteristic value in the real-time driving characteristic data set corresponding to the high-speed operating condition, to obtain a set of driving characteristic data corresponding to the low-speed operating condition after averaging, that is, the first sample; and a set of driving characteristic data corresponding to the high-speed operating condition after averaging, that is, the second sample. Then, perform normalization processing on the characteristic data in the first sample and the second sample respectively to eliminate the influence of dimensions. Calculate the Euclidean distance between the first sample and the first cluster center corresponding to the target operating condition (corresponding to the low-speed operating condition) obtained from the cloud to obtain the membership degree for each cluster, and calculate the Euclidean distance between the second sample and the first cluster center corresponding to the target operating condition (corresponding to the high-speed operating condition) obtained from the cloud to obtain the membership degree for each cluster. For the first sample and the second sample, respectively take the membership degree corresponding to the first-dimensional value (i.e., the aggressive type) as the membership degree of the first sample and the second sample, where the membership degree represents the driving style factor.

[0082] Calculate the comprehensive driving style factor (i.e., representing the current driving style of the target vehicle) according to the following formula: Where: facAct: The driving style factor at the current moment, representing the current driving style of the target vehicle; facPre: The driving style factor at the previous moment, when initially calculating, take the value of 0.5, and after calculating a new facAct each time, use its value as facPre for the next calculation; facHigh: The driving style factor corresponding to the high-speed operating condition based on the proportion of the operating condition time; facLow: Driving style factor corresponding to low-speed driving conditions based on the proportion of driving conditions.

[0083] facLow = LowModeSect * (num1 / (num1 + num2)) facHign = HighModeSect * (num2 / (num1 + num2)) HighModeSect: Membership degree under high vehicle speed conditions, taking the first-dimensional value in the two-dimensional membership degree (representing the radical degree), that is, the membership degree of the second sample; LowModeSec: Membership degree under low vehicle speed conditions, taking the first-dimensional value in the two-dimensional membership degree (representing the radical degree), that is, the membership degree of the first sample; num1: Cumulative value of low-speed segments in multiple groups of real-time driving data. Each time a low-speed driving condition is recognized, 1 is added; num2: Cumulative value of high-speed segments in multiple groups of real-time driving data. Each time a high-speed driving condition is recognized, 1 is added.

[0084] Finally, based on the comprehensive driving style factor, the system recognizes and outputs the current driving style of the target vehicle, realizing real-time monitoring and analysis of driving behavior. This method not only improves the recognition accuracy of driving style but also provides important information for personalized driving experience and driving assistance systems.

[0085] In summary, according to the technical solution provided by the embodiment of the present invention, through the above method, not only can the recognition accuracy of driving style be improved, but also important information can be provided for personalized driving experience and driving assistance systems, thereby realizing more intelligent and user-friendly vehicle control.

[0086] Further, the above driving style recognition method further includes: Adjust the throttle opening of the target vehicle based on the current driving style of the target vehicle.

[0087] Specifically, if the recognized style is aggressive, the system may adjust the throttle opening to provide a more sensitive acceleration response. If the recognized style is mild, the system may adjust the throttle opening to provide a smoother acceleration and better fuel economy.

[0088] According to the driving style recognition method of one aspect of the present invention, the method further includes: When it is recognized that the target vehicle is in cruise mode, keep the throttle opening of the target vehicle adjusted.

[0089] Specifically, the cruise mode usually refers to the vehicle automatically driving at a set speed without the driver continuously operating the throttle. This can be determined by the vehicle's speed sensor, throttle position sensor, and the status of the cruise control system. After identifying that the vehicle is in the cruise mode, the throttle opening will not be adjusted. This is because in the cruise mode, the throttle opening of the vehicle is already managed by the cruise control system to maintain the set vehicle speed. Not adjusting the throttle opening can avoid interfering with the driver's intention and maintain driving comfort and stability. In the cruise mode, the driver may prefer a stable driving state and does not want the system to make unnecessary interventions.

[0090] Figure 4 It is one of the structural schematic diagrams of the driving style recognition system provided by the present invention.

[0091] As Figure 4 shown, the driving style recognition system 400 provided by the present invention is applied to the cloud and includes: a driving data acquisition module 401, a target working condition driving data acquisition module 402, a driving feature acquisition module 403, a clustering center acquisition module 404, and an identification module 405.

[0092] Further, the driving data acquisition module 401 is used to dynamically receive the historical driving data of vehicles from multiple terminals in the cloud; Further, the target working condition driving data acquisition module 402, the driving feature acquisition module, is used to classify the working conditions of the historical driving data of the vehicle and obtain the historical driving data corresponding to several target working conditions; Further, the driving feature acquisition module 403 is used to obtain the driving feature data set corresponding to each target working condition based on the historical driving data corresponding to each target working condition; Further, the clustering center acquisition module 404 is used to perform clustering processing on the driving feature data set corresponding to each target working condition based on the fuzzy C-means clustering algorithm to obtain the first clustering center corresponding to each target working condition; Further, the driving style recognition module 405 is used to send the first clustering center corresponding to each target working condition to the target terminal so that the target terminal can perform driving style recognition on the target vehicle based on the collected multi-group real-time driving data of the target vehicle and the first clustering center corresponding to each target working condition.

[0093] Figure 5 It is another structural schematic diagram of the driving style recognition system provided by the present invention.

[0094] As Figure 5 shown, the driving style recognition system 500 provided by the present invention is applied to the target terminal and includes: a clustering center receiving module 501, a real-time driving data acquisition module 502, and a driving style recognition module 503.

[0095] Further, a clustering center receiving module 501 is configured to receive a plurality of first clustering centers corresponding to respective target driving conditions from the cloud; wherein, the first clustering centers are obtained by the cloud based on dynamically receiving vehicle historical driving data from multiple terminals; Classify the vehicle historical driving data by driving condition to obtain historical driving data corresponding to respective target driving conditions; Based on the historical driving data corresponding to each target driving condition, obtain a driving feature data set corresponding to each target driving condition; Obtained by performing clustering processing on the driving feature data set corresponding to each target driving condition based on the fuzzy C-means clustering algorithm; Further, a real-time driving data acquisition module 502 is configured to acquire multiple groups of real-time driving data of the target vehicle.

[0096] Further, a driving style recognition module 503 recognizes the driving style of the target vehicle based on multiple groups of real-time driving data of the target vehicle and the first clustering center corresponding to each target driving condition.

[0097] According to another aspect of the present invention, there is provided an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the driving style recognition method as described above is implemented.

[0098] As described above, the driving style recognition method and system of the present invention have been described in detail with reference to the accompanying drawings. The total historical driving data of different vehicles (i.e., a large amount of vehicle driving data) is transmitted to the cloud. For example, at least 100 pieces of actual driving data from various regions across the country are collected to cover various working conditions as much as possible. In the cloud, by calculating the total historical driving data, driving style features reflecting different driving styles corresponding to different working conditions can be obtained, that is, the first clustering centers corresponding to different target working conditions are calculated. Furthermore, the cloud sends the first clustering centers corresponding to different target working conditions to the terminal. At the terminal, the current driving style of the target vehicle is recognized, which can effectively save the storage space of the terminal (such as the vehicle controller) and improve the recognition accuracy. Specifically, the first clustering centers calculated by the cloud can more accurately represent the characteristics of different driving styles because they are obtained based on a large amount of data and complex algorithms. The terminal uses these accurate clustering centers for real-time recognition, which can reduce errors caused by local computing power limitations or insufficient data. And since the calculation of the clustering centers is completed in the cloud, the terminal device does not need to store a large amount of historical data or run complex algorithms, thus saving the storage space of the terminal. The terminal can directly use the clustering centers calculated by the cloud for real-time recognition of the driving style, which improves the recognition efficiency. The cloud is responsible for data processing and the calculation of the clustering centers, while the terminal is responsible for the collection of real-time data and style recognition. This division of labor and cooperation mode improves the performance of the entire system. Further, with the continuous addition of new data, the cloud can dynamically update the clustering centers to ensure that the terminal uses the latest information, thereby improving the recognition accuracy.

[0099] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0100] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A driving style recognition method, characterized in that: Applied in the cloud, including: Dynamically receive vehicle historical driving data from multiple terminals; Classifying the operating conditions of the historical driving data of the vehicle to obtain historical driving data corresponding to a plurality of target operating conditions; Based on the historical driving data corresponding to each target operating condition, obtaining a driving characteristic data set corresponding to each target operating condition; Based on the fuzzy C-means clustering algorithm, clustering the driving feature data set corresponding to each target working condition is performed to obtain a first cluster center corresponding to each target working condition; The first cluster center corresponding to each of the target working conditions is sent to a target terminal, so that the target terminal identifies the driving style of the target vehicle based on multiple sets of collected real-time driving data of the target vehicle and the first cluster center corresponding to each of the target working conditions.

2. The driving style recognition method according to claim 1, characterized in that: The vehicle historical driving data at least includes vehicle speed information of multiple vehicles at different time points within a unit collection time, and the target operating condition includes a low speed operating condition and a high speed operating condition; The operating condition classification of the historical driving data of the vehicle to obtain the historical driving data corresponding to a plurality of target operating conditions includes: Based on the vehicle historical driving data and the vehicle speed information, first historical driving data corresponding to the low vehicle speed condition and second historical driving data corresponding to the high vehicle speed condition are acquired.

3. The driving style recognition method according to claim 2, characterized in that: The step of acquiring a driving characteristic data set corresponding to each target operating condition based on the historical driving data corresponding to each target operating condition includes: Based on the first historical driving data, a first driving characteristic data set of the low vehicle speed condition is obtained, wherein each set of first driving characteristic data in the first driving characteristic data set includes an average value of acceleration, a standard deviation of acceleration, an average value of a positive acceleration change rate, an average value of a negative acceleration change rate, a standard deviation of an acceleration change rate, an average value of a throttle opening, and a maximum value of a throttle opening; Based on the second historical driving data, a second driving characteristic data set of the high-speed condition is obtained, wherein each group of second driving characteristic data in the second driving characteristic data set includes a throttle opening standard deviation, a positive throttle opening change rate mean, a negative throttle opening change rate mean, and a standard deviation of the throttle opening change rate.

4. The driving style recognition method according to claim 1, characterized in that: The method of clustering the driving characteristic data set corresponding to each target working condition based on the fuzzy C-means clustering algorithm to obtain a first cluster center corresponding to each target working condition includes: Processing the driving characteristic data set by normalization; Set the number of clusters to 2, divide the driving styles into aggressive and mild, and initialize the membership matrix; Calculate and update the cluster center of each cluster until the preset iteration stop condition is met; Output the first cluster center corresponding to each target operating condition.

5. A driving style recognition method, characterized in that: Applied to target terminals, including: Receiving first cluster centers corresponding to a plurality of target working conditions from the cloud; Collect multiple sets of real-time driving data of the target vehicle; Based on the multiple sets of real-time driving data and the first cluster center corresponding to each of the target operating conditions, performing driving style recognition on the target vehicle; Wherein, the first cluster center is formed by the cloud based on dynamic reception of historical vehicle driving data from multiple terminals; Classifying the operating conditions of the historical driving data of the vehicle to obtain historical driving data corresponding to a plurality of target operating conditions; Based on the historical driving data corresponding to each target operating condition, obtaining a driving characteristic data set corresponding to each target operating condition; Based on the fuzzy C-means clustering algorithm, the driving characteristic data corresponding to each target working condition is clustered and obtained.

6. The driving style control method according to claim 5, characterized in that: The performing driving style recognition on the target vehicle based on the multiple sets of real-time driving data and the first cluster center corresponding to each target working condition includes: Determining whether the real-time operating conditions corresponding to the multiple sets of real-time driving data belong to one of the plurality of target operating conditions; When it is determined that the target vehicle belongs to one of the several target operating conditions, the driving style of the target vehicle is identified based on the multiple groups of real-time driving data and the first cluster center corresponding to each of the target operating conditions.

7. The driving style control method according to claim 6, characterized in that: The performing driving style recognition on the target vehicle based on the multiple sets of real-time driving data and the first cluster center corresponding to each target working condition includes: Based on each set of the real-time driving data, acquiring the real-time driving characteristic data of the real-time working condition corresponding thereto; The current driving style of the target vehicle is determined based on at least each group of the real-time driving characteristic data and the first cluster center of each target operating condition.

8. A driving style recognition system, characterized in that: Applied in the cloud, including: A driving data acquisition module, used to dynamically receive historical vehicle driving data from multiple terminals; A driving feature acquisition module, used to classify the vehicle's historical driving data into working conditions, and acquire historical driving data corresponding to a plurality of target working conditions; A target working condition driving data acquisition module, used for acquiring a driving characteristic data set corresponding to each target working condition based on the historical driving data corresponding to each target working condition; A cluster center acquisition module, used for performing clustering processing on the driving characteristic data corresponding to each target working condition based on a fuzzy C-means clustering algorithm, and acquiring a first cluster center corresponding to each target working condition; A driving style recognition module is used to send the first cluster center corresponding to each of the target working conditions to a target terminal, so that the target terminal can recognize the driving style of the target vehicle based on multiple sets of real-time driving data collected from the target vehicle and the first cluster center corresponding to each of the target working conditions.

9. A driving style recognition system, characterized in that: Applied to target terminals, including: A cluster center receiving module, used to receive first cluster centers corresponding to a plurality of target operating conditions from the cloud; wherein the first cluster center is formed by the cloud based on dynamic reception of historical vehicle driving data from a plurality of terminals; Classifying the operating conditions of the historical driving data of the vehicle to obtain historical driving data corresponding to a plurality of target operating conditions; Based on the historical driving data corresponding to each target operating condition, obtaining a driving characteristic data set corresponding to each target operating condition; Based on the fuzzy C-means clustering algorithm, the driving characteristic data set corresponding to each target working condition is clustered to obtain; A real-time driving data collection module is used to collect multiple sets of real-time driving data of the target vehicle; The driving style recognition module recognizes the driving style of the target vehicle based on the multiple groups of real-time driving data and the first cluster center corresponding to each of the target working conditions.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the driving style recognition method according to any one of claims 1 to 7 is implemented.

Images