Vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis

Abstract

The invention discloses a vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis, and the method comprises the steps: 1), evaluating the vehicle computing power sharing priority based on the historical driving behavior data of a vehicle-end user and the current state of a vehicle; 2) the user side sends a computing power request to the cloud side; 3) based on the vehicle computing power sharing priority, the cloud side dispatches one or more vehicle side computing power with the highest priority, responds to the computing power request, and uploads a response result to the cloud side; and 4) the cloud side summarizes the response results and transmits the response results to the user side. According to the method, the driving style data is subjected to clustering analysis, the vehicle starting computing power sharing time is dynamically determined, the method does not depend on a charging pile, and the idle computing power of the vehicle is effectively utilized.

Description

Technical Field

[0001] The present invention relates to the field of edge computing and vehicle-mounted computing power sharing, and specifically to a vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis. Background Art

[0002] With the rapid development of Internet of Vehicles technology, the processing power of on-board computing devices has been significantly enhanced. Compared with cloud servers that run 24 / 7, on-board computing resources are usually only used for intelligent driving assistance systems when the vehicle is running. During the period when the vehicle is stopped, these computing resources are often not fully utilized, resulting in a huge waste of resources. If these idle on-board computing powers can be effectively mobilized and utilized, resources can be saved and significant benefits can be brought to society and the economy. The existing vehicle computing power sharing method obtains its status information when the vehicle is connected to the charging pile, and uses the idle state of the vehicle during the low electricity cost period to execute the preset computing tasks issued by the cloud. This method releases the computing power resources of the vehicle when charging, and performs collaborative computing of multiple vehicles when the electricity cost is low, so as to achieve efficient computing task processing.

[0003] The existing technology in the field of vehicle computing power sharing mainly relies on the availability of charging piles and the fluctuation of electricity costs to schedule computing power resources, which limits its application capabilities in environments without charging piles. Limited by the dependence on charging infrastructure, the existing methods cannot effectively cope with the computing power requirements of vehicles when there is no support from charging piles, especially during peak hours when computing power shortages need to be avoided. In order to address these limitations, the present invention proposes an innovative vehicle computing power sharing scheduling method and device, which can optimize the allocation and utilization of vehicle computing power resources through intelligent prediction and scheduling algorithms when the vehicle is connected to or away from the charging pile, ensure the continuity and efficiency of computing power services, and maintain service quality even when power resources are limited. Therefore, the development of a vehicle-mounted computing power sharing scheduling method, system and device based on user driving behavior analysis is of great significance for improving the utilization rate of vehicle-mounted computing power and providing users with more flexible and reliable computing power sharing services. Summary of the invention

[0004] The purpose of the present invention is to provide a vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis, comprising the following steps:

[0005] 1) Build a vehicle-side computing power sharing system, including the cloud and vehicle side; the cloud side stores the vehicle-side user's historical driving behavior data and obtains the vehicle's current status in real time;

[0006] 2) Evaluate the priority of vehicle computing power sharing based on the vehicle-side user's historical driving behavior data and the vehicle's current status;

[0007] 3) The client sends a computing request to the cloud;

[0008] 4) Based on the vehicle computing power sharing priority, the cloud dispatches one or more vehicle-side computing powers with the highest priority, responds to the computing power request, and uploads the response result to the cloud;

[0009] 5) The cloud summarizes the response results and transmits them to the user end.

[0010] Further, in step 1), the current state of the vehicle includes the current use state of the vehicle λ 1 , the remaining power of the vehicle a, the state of the vehicle connected to the power supply λ 2 ; 2 =1 means the vehicle is connected to the power supply, λ 2 =0 means the vehicle is not connected to a power source.

[0011] Furthermore, in step 2), after the user sends a computing power request to the cloud, the cloud determines the response service type corresponding to the computing power request; the response service types include continuous response service and one-time response service.

[0012] If the response service type is continuous response service, the cloud allocates computing power with higher priority.

[0013] Continuous response services require contextual connections, so the cloud will allocate higher priority computing power to them to ensure the continuity of their services.

[0014] Furthermore, in step 3), the rules for evaluating the vehicle computing power sharing priority include:

[0015] The more remaining power a vehicle has or the more power it has connected to, the higher the priority of computing power startup.

[0016] The greater the probability that the vehicle will not be used in the next m time slots, the greater the priority of computing power startup;

[0017] The more conservative the owner's driving style, the higher the priority of computing power startup.

[0018] Further, in step 3), the step of evaluating the vehicle computing power sharing priority includes:

[0019] 3.1) Perform cluster analysis on the user's historical driving behavior data to obtain multiple driving style cluster centers;

[0020] 3.2) Based on the driving style cluster center sequence number, the vehicle-side driving style level s∈|K| is assigned; the higher the vehicle-side driving style level s, the greater the vehicle computing power sharing priority;

[0021] 3.3) Based on the driving style time series data, calculate the probability p = n that the owner will use the car in the future m time slots m / n;n m The number of times the owner has used the car in history nm , n is the total number of days;

[0022] 3.4) Calculate the vehicle computing power sharing priority S = sa (2-λ 1 )(λ 2 +1)(1-p); The larger the S value, the higher the vehicle's computing power startup priority.

[0023] Furthermore, in step 3.1), the method for clustering the user's historical driving behavior data includes a K-Means method.

[0024] Further, in step 3.1), the step of performing cluster analysis on the user's historical driving behavior data includes:

[0025] 3.1.1) Collect driving behavior data, denoted as x i ={x i1 ,x i2 ,...,x i5}; x i1 ,x i2 ,...,x i5 They represent the vehicle speed, accelerator pedal opening, following vehicle distance, acceleration, and braking frequency respectively;

[0026] 3.1.2) Use the elbow rule to determine the value of K and calculate the total inner sum of squares under different values ​​of K Where N is the total number of data points, K is the number of clusters, μ j is the centroid of the jth cluster, δ(y i ,μ j ) is an indicator function, if x i Belongs to cluster i is 1, otherwise it is 0, ||x i -μ j || is the data point x i With the center of mass μ j The Euclidean distance between

[0027] 3.1.3) Set the initial center of mass Initialize the number of iterations t = 1;

[0028] 3.1.4) For each data point x i Assign to the nearest centroid y i =argmin||x i -μ j || 2 ;

[0029] 3.1.5) Recalculate the centroid Among them, C j is the set of all data points in the jth cluster, |Cj ∣ is the number of data points in cluster j;

[0030] 3.1.6) Judgment Or whether t is greater than tmax. If not, set t = t + 1 and update the centroid. And return to step 3.1.4), if yes, then end the iteration and select the K value with the slowest SSE decrease rate as the optimal number of clusters; ∈ is the error threshold; tmax is the maximum number of iterations;

[0031] 3.1.7) Based on the cluster number K, cluster the user's historical driving behavior data.

[0032] Furthermore, in step 3.1.3), the initial centroid is set by randomly selecting K data points as the initial centroid, or by using the K-Means algorithm to initialize the centroid. D(x) is the cost function for selecting data x as the centroid.

[0033] Furthermore, in step 4), the cloud periodically schedules one or more vehicle-side computing powers with the highest priority, and the scheduling period is m time slots.

[0034] Furthermore, in step 4), after the cloud dispatches one or more vehicle-side computing powers with the highest priority, the vehicle-side revenue is settled periodically.

[0035] The technical effect of the present invention is unquestionable. The present invention proposes a method and device for sharing vehicle-side idle computing power that integrates vehicle status and user behavior analysis. Through an intelligent scheduling algorithm, the allocation and use of vehicle-side computing power are optimized to ensure the efficiency and continuity of computing power services, and to make full use of the vehicle's idle computing resources without interfering with the owner's normal use of the vehicle.

[0036] The present invention dynamically determines the time when the vehicle starts sharing computing power by clustering and analyzing driving style data, does not rely on charging piles, and effectively utilizes the idle computing power of the vehicle.

[0037] Under the vehicle-cloud-user system architecture, users can rent computing power according to their needs without having to worry about the computing power provider. Car owners can rent computing power according to their usage habits.

[0038] The cloud makes computing power scheduling decisions every m time slots, which can effectively sense changes in vehicle status in order to achieve load balancing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is the system model diagram;

[0040] Figure 2 For business process diagram;

[0041] Figure 3This is the flow chart of the computing power sharing algorithm. DETAILED DESCRIPTION

[0042] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0043] Embodiment 1:

[0044] See also Figures 1 to 3 , a vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis, comprising the following steps:

[0045] 1) Build a vehicle-side computing power sharing system, including the cloud and vehicle side; the cloud side stores the vehicle-side user's historical driving behavior data and obtains the vehicle's current status in real time;

[0046] 2) Evaluate the priority of vehicle computing power sharing based on the vehicle-side user's historical driving behavior data and the vehicle's current status;

[0047] 3) The client sends a computing request to the cloud;

[0048] 4) Based on the vehicle computing power sharing priority, the cloud dispatches one or more vehicle-side computing powers with the highest priority, responds to the computing power request, and uploads the response result to the cloud;

[0049] 5) The cloud summarizes the response results and transmits them to the user end.

[0050] In step 1), the current state of the vehicle includes the current usage state of the vehicle λ 1 , the remaining power of the vehicle a, the state of the vehicle connected to the power supply λ 2 ; 2 =1 means the vehicle is connected to the power supply, λ 2 =0 means the vehicle is not connected to a power source.

[0051] If the response service type is continuous response service, the cloud allocates computing power with higher priority.

[0052] Continuous response services require contextual connections, so the cloud will allocate higher priority computing power to them to ensure the continuity of their services.

[0053] In step 2), after the user sends a computing power request to the cloud, the cloud determines the response service type corresponding to the computing power request; the response service types include continuous response service and one-time response service.

[0054] In step 3), the rules for evaluating vehicle computing power sharing priorities include:

[0055] The more remaining power a vehicle has or the more power it has connected to, the higher the priority of computing power startup.

[0056] The greater the probability that the vehicle will not be used in the next m time slots, the greater the priority of computing power startup;

[0057] The more conservative the owner's driving style, the higher the priority of computing power startup.

[0058] In step 3), the steps of evaluating the vehicle computing power sharing priority include:

[0059] 3.1) Perform cluster analysis on the user's historical driving behavior data to obtain multiple driving style cluster centers;

[0060] 3.2) Based on the driving style cluster center sequence number, the vehicle-side driving style level s∈|K| is assigned; the higher the vehicle-side driving style level s, the greater the vehicle computing power sharing priority;

[0061] 3.3) Based on the driving style time series data, calculate the probability p = n that the owner will use the car in the future m time slots m / n;n m The number of times the owner has used the car in history n m , n is the total number of days;

[0062] 3.4) Calculate the vehicle computing power sharing priority S = sa (2-λ 1 )(λ 2 +1)(1-p); The larger the S value, the higher the vehicle's computing power startup priority.

[0063] In step 3.1), the method for clustering the user's historical driving behavior data includes the K-Means method.

[0064] In step 3.1), the step of clustering the user's historical driving behavior data includes:

[0065] 3.1.1) Collect driving behavior data, denoted as x i ={x i1 ,x i2 ,...,x i5}; x i1 ,x i2 ,...,x i5 They represent the vehicle speed, accelerator pedal opening, following vehicle distance, acceleration, and braking frequency respectively;

[0066] 3.1.2) Use the elbow rule to determine the value of K and calculate the total inner sum of squares under different values ​​of K Where N is the total number of data points, K is the number of clusters, μ j is the centroid of the jth cluster, δ(y i ,μi ) is an indicator function, if x i Belongs to cluster i is 1, otherwise it is 0, ||x i -μ j || is the data point x i With the center of mass μ j The Euclidean distance between

[0067] 3.1.3) Set the initial center of mass Initialize the number of iterations t = 1;

[0068] 3.1.4) For each data point x i Assign to the nearest centroid y i =argmin||x i -μ j || 2 ;

[0069] 3.1.5) Recalculate the centroid Among them, C j is the set of all data points in the jth cluster, |C j ∣ is the number of data points in cluster j;

[0070] 3.1.6) Judgment Or whether t is greater than tmax. If not, set t = t + 1 and update the centroid. And return to step 3.1.4), if yes, then end the iteration and select the K value with the slowest SSE decrease rate as the optimal number of clusters; ∈ is the error threshold; tmax is the maximum number of iterations;

[0071] 3.1.7) Based on the cluster number K, cluster the user's historical driving behavior data.

[0072] In step 3.1.3), the initial centroid is set by randomly selecting K data points as the initial centroid, or using the K-Means algorithm to initialize the centroid. D(x) is the cost function for selecting data x as the centroid.

[0073] In step 4), the cloud periodically schedules one or more vehicle-side computing forces with the highest priority, and the scheduling period is m time slots.

[0074] In step 4), after the cloud dispatches one or more vehicle-side computing power with the highest priority, the vehicle-side revenue is settled periodically.

[0075] Embodiment 2:

[0076] A vehicle computing power sharing scheduling method based on user driving behavior analysis includes the following steps:

[0077] 1) Build a vehicle-side computing power sharing system, including the cloud and vehicle side; the cloud side stores the vehicle-side user's historical driving behavior data and obtains the vehicle's current status in real time;

[0078] 2) Evaluate the priority of vehicle computing power sharing based on the vehicle-side user's historical driving behavior data and the vehicle's current status;

[0079] 3) The client sends a computing request to the cloud;

[0080] 4) Based on the vehicle computing power sharing priority, the cloud dispatches one or more vehicle-side computing powers with the highest priority, responds to the computing power request, and uploads the response result to the cloud;

[0081] 5) The cloud summarizes the response results and transmits them to the user end.

[0082] Embodiment 3:

[0083] A vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis, the technical content is the same as that of Example 2, further, in step 1), the current state of the vehicle includes the current use state of the vehicle λ 1 , the remaining power of the vehicle a, the state of the vehicle connected to the power supply λ 2 ; 2 =1 means the vehicle is connected to the power supply, λ 2 =0 means the vehicle is not connected to a power source.

[0084] Embodiment 4:

[0085] A method for sharing and scheduling vehicle computing power based on user driving behavior analysis, the technical content is the same as any one of embodiments 2-3, further, in step 2), after the user terminal sends a computing power request to the cloud, the cloud determines the response service type corresponding to the computing power request; the response service type includes continuous response service and one-time response service. If the response service type is continuous response service, the cloud allocates computing power with a higher priority.

[0086] Continuous response services require contextual connections, so the cloud will allocate higher priority computing power to them to ensure the continuity of their services.

[0087] Embodiment 5:

[0088] A method for scheduling vehicle computing power sharing based on user driving behavior analysis, the technical content of which is the same as any one of Embodiments 2-4, further, in step 3), the rules for evaluating the vehicle computing power sharing priority include:

[0089] The more remaining power a vehicle has or the more power it has connected to, the higher the priority of computing power startup.

[0090] The greater the probability that the vehicle will not be used in the next m time slots, the greater the priority of computing power startup;

[0091] The more conservative the owner's driving style, the higher the priority of computing power startup.

[0092] Embodiment 6:

[0093] A vehicle computing power sharing scheduling method based on user driving behavior analysis, the technical content is the same as any one of embodiments 2-5, further, in step 3), the step of evaluating the vehicle computing power sharing priority includes:

[0094] 3.1) Perform cluster analysis on the user's historical driving behavior data to obtain multiple driving style cluster centers;

[0095] 3.2) Based on the driving style cluster center sequence number, the vehicle-side driving style level s∈|K| is assigned; the higher the vehicle-side driving style level s, the greater the vehicle computing power sharing priority;

[0096] 3.3) Based on the driving style time series data, calculate the probability p = n that the owner will use the car in the future m time slots m / n;n m The number of times the owner has used the car in history n m , n is the total number of days;

[0097] 3.4) Calculate the vehicle computing power sharing priority S = sa (2-λ 1 )(λ 2 +1)(1-p); The larger the S value, the higher the vehicle's computing power startup priority.

[0098] Embodiment 7:

[0099] A vehicle computing power sharing and scheduling method based on user driving behavior analysis, the technical content of which is the same as any one of Examples 2-6. Furthermore, in step 3.1), the method for clustering analysis of user historical driving behavior data includes a K-Means method.

[0100] Embodiment 8:

[0101] A vehicle computing power sharing scheduling method based on user driving behavior analysis, the technical content is the same as any one of embodiments 2-7, further, in step 3.1), the step of clustering analysis of user historical driving behavior data includes:

[0102] 3.1.1) Collect driving behavior data, denoted as x i ={x i1 ,x i2 ,...,x i5}; x i1 ,x i2 ,...,x i5They represent the vehicle speed, accelerator pedal opening, following vehicle distance, acceleration, and braking frequency respectively;

[0103] 3.1.2) Use the elbow rule to determine the value of K and calculate the total inner sum of squares under different values ​​of K Where N is the total number of data points, K is the number of clusters, μ j is the centroid of the jth cluster, δ(y i ,μ j ) is an indicator function, if x i Belongs to cluster i is 1, otherwise it is 0, ||x i -μ j || is the data point x i With the center of mass μ j The Euclidean distance between

[0104] 3.1.3) Set the initial center of mass Initialize the number of iterations t = 1;

[0105] 3.1.4) For each data point x i Assign to the nearest centroid y i =argmin||x i -μ j || 2 ;

[0106] 3.1.5) Recalculate the centroid Among them, C j is the set of all data points in the jth cluster, |C j ∣ is the number of data points in cluster j;

[0107] 3.1.6) Judgment Or whether t is greater than tmax. If not, set t = t + 1 and update the centroid. And return to step 3.1.4), if yes, then end the iteration and select the K value with the slowest SSE decrease rate as the optimal number of clusters; ∈ is the error threshold; tmax is the maximum number of iterations;

[0108] 3.1.7) Based on the cluster number K, cluster the user's historical driving behavior data.

[0109] Embodiment 9:

[0110] A method for sharing and scheduling vehicle computing power based on user driving behavior analysis, the technical content of which is the same as any one of Embodiments 2-8. Further, in step 3.1.3), the method for setting the initial centroid is: randomly selecting K data points as the initial centroid, or using the K-Means algorithm to initialize the centroid D(x) is the cost function for selecting data x as the centroid.

[0111] Embodiment 10:

[0112] A method for sharing and scheduling vehicle-mounted computing power based on user driving behavior analysis, the technical content of which is the same as any one of Examples 2-9. Furthermore, in step 4), the cloud periodically schedules one or more vehicle-side computing powers with the highest priority, and the scheduling period is m time slots.

[0113] Embodiment 11:

[0114] A method for sharing and scheduling vehicle computing power based on user driving behavior analysis, the technical content of which is the same as any one of Examples 2-10. Furthermore, in step 4), after the cloud dispatches one or more vehicle-side computing powers with the highest priority, the vehicle-side income is settled periodically.

[0115] Embodiment 11:

[0116] A vehicle computing power sharing scheduling method based on user driving behavior analysis, the steps are as follows:

[0117] 1) Model the vehicle-side computing power sharing system. The car owner authorizes the vehicle computing power sharing. The cloud obtains the user's historical driving behavior data and the current status of the vehicle: the current usage status of the vehicle λ 1 , the remaining battery power of the vehicle a, whether the vehicle is connected to the power supply λ 2 .

[0118] 2) The user sends a computing request to the cloud, and the cloud evaluates whether it needs continuous service, such as AI dialogue, or a task that does not care about the context, such as a one-time AI conversation service. The cloud will maintain the consistency of user sessions for key AI model applications to ensure service continuity.

[0119] 3) The cloud dynamically determines the time for sharing vehicle computing power. The cloud performs computing power scheduling every m time slots. The cloud evaluates the priority of vehicle computing power sharing. We evaluate the remaining computing power of the vehicle based on the vehicle status and driving style. Specifically, the more remaining power the vehicle has or the more power it is connected to, the greater the priority of starting the computing power; the greater the probability that the vehicle will not be used in the next m time slots, the greater the priority of starting the computing power; the more conservative the owner's driving style, which means that he will not frequently perform tasks that require a lot of computing power (such as high-speed driving or complex driving operations), the greater the priority of starting the computing power. The specific steps for calculating the computing power startup priority are as follows:

[0120] 3.1) Use the K-Means algorithm to cluster the driving behavior and define the K value as the number of driving style classification levels. The specific steps are as follows:

[0121] 3.1.1) Collect driving behavior data, represented by xi =(vehicle speed, accelerator pedal opening, following vehicle distance, acceleration, and frequency of braking), and determine it as a characteristic indicator of driving style.

[0122] 3.1.2) Use the “elbow rule” to determine the value of K and calculate the total internal sum of squares (SSE) under different K values: Where N is the total number of data points, K is the number of clusters, μ j is the centroid of the jth cluster, δ(y i ,μ j ) is an indicator function, if x i Belongs to cluster i is 1, otherwise it is 0, ||x i -μ j || is the data point x i With the center of mass μ j The Euclidean distance between .

[0123] 3.1.3) Randomly select K data points as the initial centroid, or use the K-Means algorithm to initialize the centroid: μ j =argmin j D(x). Where D(x) is the cost function for selecting data x as the centroid, which is usually proportional to the square of the distance to the nearest centroid.

[0124] 3.1.4) For each data point x i Assign to the nearest centroid μ j :y i =argmin||x i -μ j || 2 .

[0125] 3.1.5) Recalculate the centroid. For each cluster, recalculate the centroid Among them, C j is the set of all data points in the jth cluster, |C j ∣ is the number of data points in cluster j.

[0126] 3.1.6) Repeat steps 3.1.4) and 3.1.5) until one of the following conditions is met:

[0127] (a) The change in the center of mass is less than the preset threshold:

[0128] (b) Reaching the preset number of iterations.

[0129] 3.1.7) Repeat 3.1.2) to 3.1.6) and select the K value with the slowest SSE decrease rate as the optimal number of clusters.

[0130] 3.1.8) Result interpretation. Drivers are divided into different style levels s, s∈|K| according to the value of the cluster center, with the sequence (vehicle speed, accelerator pedal opening, following distance, acceleration, braking frequency) as the reference priority. The higher s is, the more conservative the driver's driving style is, which means more residual computing power for the vehicle.

[0131] 3.2) Based on the driving style time series data, in the future m time slots, refer to the owner's historical number of car usage n m , the total number of days is n, and the probability of the owner using the car in the future m time slots is: p = n m / n.

[0132] 3.3) We can further obtain the computing power startup priority of each vehicle in the future m time slots: S = sa (2-λ 1 )(λ 2 +1)(1-p). The larger the S value, the higher the vehicle's computing power startup priority.

[0133] 4) After the car owner completes the computing power sharing, the car automatically sends the shared period data to the cloud. The cloud quickly calculates and updates the car owner's income, and settles it to the car owner regularly in a secure way. The car owner can view detailed sharing records and income through the mobile app.

Claims

1. A vehicle computing power sharing scheduling method based on user driving behavior analysis, characterized in that: The following steps are involved: 1) Build a vehicle-side computing power sharing system, including the cloud and the vehicle side; The cloud stores the historical driving behavior data of vehicle-side users and obtains the current status of the vehicle in real time. 2) Evaluate the priority of vehicle computing power sharing based on the vehicle-side user's historical driving behavior data and the vehicle's current status; 3) The client sends a computing request to the cloud; 4) Based on the vehicle computing power sharing priority, the cloud dispatches one or more vehicle-side computing powers with the highest priority, responds to the computing power request, and uploads the response result to the cloud; 5) The cloud summarizes the response results and transmits them to the user end.

2. According to claim 1, a method for sharing and scheduling vehicle computing power based on user driving behavior analysis is characterized in that: In step 1), the current state of the vehicle includes the current use state of the vehicle λ1, the remaining power of the vehicle a, and the state of the vehicle connected to the power supply λ2; λ2=1 indicates that the vehicle is connected to the power supply, and λ2=0 indicates that the vehicle is not connected to the power supply.

3. The vehicle-mounted computing power sharing scheduling method based on user driving behavior analysis according to claim 1 is characterized in that: In step 2), after the user sends a computing power request to the cloud, the cloud determines the response service type corresponding to the computing power request; the response service type includes continuous response service and one-time response service; If the response service type is continuous response service, the cloud allocates computing power with higher priority.

4. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 1 is characterized in that: In step 3), the rules for evaluating vehicle computing power sharing priorities include: The more remaining power a vehicle has or the more power it has connected to, the higher the priority of computing power startup. The greater the probability that the vehicle will not be used in the next m time slots, the greater the priority of computing power startup; The more conservative the owner's driving style, the higher the priority of computing power startup.

5. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 1 is characterized in that: In step 3), the steps of evaluating the vehicle computing power sharing priority include: 3.1) Perform cluster analysis on the user's historical driving behavior data to obtain multiple driving style cluster centers; 3.2) Based on the driving style cluster center sequence number, the vehicle-side driving style level s∈|K| is assigned; the higher the vehicle-side driving style level s, the greater the vehicle computing power sharing priority; 3.3) Based on the driving style time series data, calculate the probability p = n that the owner will use the car in the future m time slots m / n;n m The number of times the owner has used the car in history n m , n is the total number of days; 3.4) Calculate the vehicle computing power sharing priority S = sa(2-λ1)(λ2+1)(1-p); the larger the S value is, the higher the vehicle's computing power startup priority is.

6. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 1 is characterized in that: In step 3.1), the method for clustering the user's historical driving behavior data includes the K-Means method.

7. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 1 is characterized in that: In step 3.1), the step of clustering the user's historical driving behavior data includes: 3.1.1) Collect driving behavior data, denoted as x i ={x i1 ,x i2 ,...,x i5 }; x i1 ,x i2 ,...,x i5 They represent the vehicle speed, accelerator pedal opening, following vehicle distance, acceleration, and braking frequency respectively; 3.1.2) Use the elbow rule to determine the value of K and calculate the total inner sum of squares under different values ​​of K Where N is the total number of data points, K is the number of clusters, μ j is the centroid of the jth cluster, δ(y i ,μ j ) is an indicator function, if x i Belongs to cluster i is 1, otherwise it is 0, ||x i -μ j || is the data point x i With the center of mass μ j The Euclidean distance between 3.1.3) Set the initial center of mass Initialize the number of iterations t = 1; 3.1.4) For each data point x i Assign to the nearest centroid y i =argmin||x i -μ j || 2 ; 3.1.5) Recalculate the centroid Among them, C j is the set of all data points in the jth cluster, |C j ∣ is the number of data points in cluster j; 3.1.6) Judgment Or whether t is greater than tmax. If not, set t = t + 1 and update the centroid. And return to step 3.1.4), if yes, then end the iteration and select the K value with the slowest SSE decrease rate as the optimal number of clusters; ∈ is the error threshold; tmax is the maximum number of iterations; 3.1.7) Based on the cluster number K, cluster the user's historical driving behavior data.

8. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 7 is characterized in that: In step 3.1.3), the initial centroid is set by randomly selecting K data points as the initial centroid, or using the K-Means algorithm to initialize the centroid. D(x) is the cost function for selecting data x as the centroid.

9. The method for sharing and scheduling vehicle computing power based on user driving behavior analysis according to claim 1 is characterized in that: In step 4), the cloud periodically schedules one or more vehicle-side computing forces with the highest priority, and the scheduling period is n time slots.

10. The vehicle-mounted computing power sharing and scheduling method based on user driving behavior analysis according to claim 1 is characterized in that: In step 4), after the cloud dispatches one or more vehicle-side computing power with the highest priority, the vehicle-side revenue is settled periodically.

Images