An ecological speed management method based on dynamic programming

Abstract

An ecological speed management method based on dynamic programming. The present invention belongs to the technical field of vehicle speed management, and specifically relates to an ecological speed management method based on dynamic programming. The purpose of the present invention is to reduce vehicle fuel consumption while ensuring efficiency and safety, and to propose an ecological speed management method based on dynamic programming. The specific process of an ecological speed management method based on dynamic programming is as follows: Step 1, collect vehicle driving trajectory data and vehicle fuel consumption data, process the collected vehicle driving trajectory data and vehicle fuel consumption data to obtain processed data; Step 2, based on the processed data, establish a vehicle operation transient fuel consumption estimation model based on the multivariate regression method; Step 3, construct an ecological speed management model based on the model of Step 2, use the dynamic programming algorithm to solve the ecological speed management model, and obtain a speed plan.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle speed management, and in particular relates to an ecological vehicle speed management method based on dynamic programming. Background Art

[0002] While driving on the road, a vehicle's energy consumption is directly affected by its speed. Ecological speed management is a viable approach to reducing vehicle energy consumption. Properly managing vehicle speed not only improves operational efficiency and safety, but also effectively reduces energy consumption. Establishing a scientific speed management approach has both theoretical and practical significance for promoting energy conservation, emission reduction, and sustainable development in transportation. Summary of the Invention

[0003] The purpose of the present invention is to reduce vehicle fuel consumption while ensuring efficiency and safety, and to propose an ecological vehicle speed management method based on dynamic programming.

[0004] The specific process of an ecological speed management method based on dynamic programming is as follows:

[0005] Step 1: Collect vehicle driving trajectory data and vehicle fuel consumption data, and process the collected vehicle driving trajectory data and vehicle fuel consumption data to obtain processed data;

[0006] Step 2: Based on the processed data, a vehicle operation transient fuel consumption estimation model based on the multivariate regression method is established;

[0007] Step 3: Based on the model in step 2, an ecological speed management model is constructed, and a dynamic programming algorithm is used to solve the ecological speed management model to obtain a speed plan.

[0008] The beneficial effects of the present invention are:

[0009] This paper proposes an ecological speed management method based on dynamic programming, aiming to achieve energy conservation and emission reduction through rational speed management. The method first collects and processes trajectory and fuel consumption data. Then, a transient fuel consumption estimation model is established based on trajectory features and their combination. Furthermore, an ecological speed management model is constructed, optimizing trip fuel consumption while taking into account road speed limits, ride comfort, and travel time constraints. Finally, a dynamic programming algorithm is used to solve the optimal speed solution.

[0010] This paper addresses the issue of eco-driving for vehicles and proposes an eco-speed management method based on dynamic programming. This method accurately estimates fuel consumption, reduces total fuel consumption and driving costs without increasing travel time or ensuring ride comfort, thereby reducing carbon emissions from road transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Flowchart of the present invention;

[0012] Figure 2 The figure is a comparison chart of the original data and the smoothed data of vehicle speed;

[0013] Figure 3 A comparison chart of the fuel consumption estimated by the transient fuel consumption estimation model and the actual fuel consumption;

[0014] Figure 4 This is a speed optimization process diagram of the ecological vehicle speed management model, where v0 represents the initial speed value; v1 represents the speed value of the first state; v2 represents the speed value of the second state; and v n-1 Indicates the speed value of the n-1th state; v n Indicates the speed value of the nth state; v t Indicates the final speed value;

[0015] Figure 5 This is the ecological speed plan obtained by dynamic programming. DETAILED DESCRIPTION

[0016] Specific implementation method 1: This implementation method is an ecological vehicle speed management method based on dynamic programming. The specific process is as follows:

[0017] Step 1: Collect vehicle driving trajectory data and vehicle fuel consumption data, and process the collected vehicle driving trajectory data and vehicle fuel consumption data to obtain processed data;

[0018] Step 2: Based on the processed data, a vehicle operation transient fuel consumption estimation model based on the multivariate regression method is established;

[0019] Step 3: Based on the model in step 2, an ecological speed management model is constructed, and a dynamic programming algorithm is used to solve the ecological speed management model to obtain a speed plan.

[0020] Specific embodiment 2: This embodiment differs from specific embodiment 1 in that: in step 1, vehicle driving trajectory data and vehicle fuel consumption data are collected; the specific process is:

[0021] The vehicle's driving trajectory data is collected through high-precision GPS equipment, and the vehicle's overall control data is read from the vehicle's OBD port using the on-board diagnostic system to obtain fuel consumption data;

[0022] The vehicle driving trajectory data mainly includes information such as vehicle latitude and longitude coordinates, vehicle movement posture and collection accuracy;

[0023] The vehicle motion posture is yaw angle, pitch angle, roll angle, gyro X axis, gyro Y axis, gyro Z axis, X axis (vehicle X axis acceleration), Y axis (vehicle Y axis acceleration), Z axis (vehicle Z axis acceleration), east speed, north speed, sky speed, vehicle speed;

[0024] The vehicle OBD port includes sampling time, fuel consumption rate (fuel consumption data), engine speed, vehicle speed, accelerator pedal position, engine torque and vehicle acceleration;

[0025] The vehicle fuel consumption data mainly includes fuel consumption rate, engine speed, vehicle speed, accelerator pedal position, engine torque and vehicle acceleration data;

[0026] The OBD port can provide various vehicle operating condition data, mainly including sampling time, fuel consumption rate, engine speed, vehicle speed, accelerator pedal position, engine torque and vehicle acceleration;

[0027] In this example, a total of about 700 minutes of raw driving data and a driving mileage of about 800 km were acquired.

[0028] Other steps and parameters are the same as those in the first embodiment.

[0029] Specific embodiment three: This embodiment differs from specific embodiment one or two in that: in step one, the collected vehicle driving trajectory data and vehicle fuel consumption data are processed; the specific process is:

[0030] 1) Delete abnormal values ​​of vehicle fuel consumption data less than 0ml / s;

[0031] 2) Based on the result of 1), according to the GPS positioning data, delete the vehicle location data outside the actual road section and the corresponding vehicle fuel consumption data;

[0032] 3) Based on the result of 2), according to the GPS positioning data, delete the vehicle location data and corresponding vehicle fuel consumption data under GPS failure;

[0033] 4) Based on the result of 3), calculate the instantaneous vehicle specific power according to formula (1). The vehicle specific power VSP is defined as the instantaneous power per unit mass of the vehicle;

[0034] VSP=v(1.1a+9.81i r +0.132)+0.000302v 3 (1)

[0035] Where VSP is the vehicle specific power in kW / s; v is the vehicle speed in m / s; a is the vehicle acceleration in m / s 2 ;ir Indicates the road slope ratio, in %;

[0036] The vehicle specific power is divided into 2 kW / s intervals (for example, [0 kW / s-2 kW / s] is the first interval, [2 kW / s-4 kW / s] is the second interval, and [4 kW / s-6 kW / s] is the third interval). This results in L intervals. The fuel consumption value within each interval roughly follows a normal distribution. Then, according to the 3σ principle, low-probability fuel consumption data exceeding three standard deviations within each interval are eliminated. (The VSP involves speed, acceleration, and road slope ratio, and speed, acceleration, and road slope ratio correspond to the vehicle fuel consumption data at the same moment).

[0037] The VSP-3σ preprocessing method calculates the instantaneous vehicle specific power and divides it into intervals of 2 kW / t. If the fuel consumption values ​​within each interval show a normal distribution, the low-probability fuel consumption data that exceeds three times the standard deviation will be eliminated.

[0038] 5) Based on the results of 4), the Ramer-Douglas-Peucker algorithm is used to identify the segments in the trajectory. The endpoints of each segment are used as segmentation points. The ratio of the height difference between the two endpoints of the segment to the distance is calculated as the road slope of this segment.

[0039] 6) Based on the result of 5), the vehicle speed and acceleration data of step 1 are smoothed using a Savitzky-Golay filter.

[0040] The effect after speed smoothing is as follows Figure 2 shown.

[0041] Other steps and parameters are the same as those in the first or second embodiment.

[0042] Specific embodiment 4: This embodiment differs from any one of specific embodiments 1 to 3 in that: in step 2, a vehicle operation transient fuel consumption estimation model based on a multiple regression method is established based on the processed data; the specific process is:

[0043] Step 2: 1. Starting from engine power, select characteristic parameters related to fuel consumption, and obtain an expression for vehicle power by analyzing the force applied to the vehicle when it is moving forward;

[0044] λP=ρC D Av 3 / 2+C r mgvcosθ+mgvsinθ+mva

[0045] Where P is the power of the vehicle engine; λ is the loss coefficient of the vehicle engine energy in the transmission system, which is affected by the vehicle structure and vehicle state; ρ is the air density; C Dis the aerodynamic drag coefficient; A represents the frontal area of ​​the vehicle; v is the speed of the ambient wind relative to the vehicle; C r is the rolling resistance coefficient of the tire; m is the vehicle weight; g is the acceleration due to gravity; θ is the road slope; a is the vehicle acceleration;

[0046] The main influencing factors of vehicle power (such as vehicle speed, vehicle acceleration and road slope) and their combined characteristics are determined as the main parameters of the vehicle operation transient fuel consumption estimation model, and the basic structure of the transient fuel consumption estimation model based on the multivariate regression method is formed;

[0047] The basic structure of the transient fuel consumption estimation model based on the multiple regression method is constructed as follows:

[0048]

[0049] Where v is the speed of the ambient wind relative to the vehicle; θ is the road slope; a is the vehicle acceleration; p air Indicates v 3 ;p r represents vcosθ; p g represents vsinθ; p acc means va;

[0050] F p,1 、F p,2 、F p,3 Indicates the power provided by the engine;

[0051] c(1) represents the polynomial degree of the ambient wind speed v relative to the vehicle; c(2) represents the polynomial degree of the vehicle acceleration a; c(3) represents the polynomial degree of the road slope θ; c(4) represents the polynomial degree of cos(θ); c(5) represents the polynomial degree of sin(θ);

[0052] N(1) represents the maximum value of the polynomial degree of the ambient wind speed v relative to the vehicle; N(2) represents the maximum value of the polynomial degree of the vehicle acceleration a; N(3) represents the maximum value of the polynomial degree of the road slope θ; N(4) represents the maximum value of the polynomial degree of cos(θ); N(5) represents the maximum value of the polynomial degree of sin(θ);

[0053] k c(1),c(2),c(3),c(4),c(5) Indicates v c(1) a c(2) θ c(3) cos c(4) (θ)sin c(5) (θ) simulation fitting coefficients;

[0054] c(6) represents p air The degree of the polynomial; c(7) represents p rThe degree of the polynomial; c(8) represents p g The degree of the polynomial; c(9) represents p acc The degree of the polynomial;

[0055] N(6) represents p air The maximum value of the polynomial degree; N(7) represents p r The maximum value of the polynomial degree; N(8) represents p g The maximum value of the polynomial degree; N(9) represents p acc The maximum value of the polynomial degree;

[0056] k c(6),c(7),c(8),c(9) express The simulation fitting coefficients of

[0057] k c(1),c(2),c(3),c(4),c(5),c(6),c(7),c(8),c(9) express The simulation fitting coefficients of

[0058] Step 22: Calculate the correlation between the features in each basic structure and remove the features with high correlation;

[0059] Based on the elimination of highly correlated features, regularization penalties are added to each basic structure to obtain filtered features; features with higher contribution to predictive ability are selected;

[0060] Based on the obtained filtered features, N vehicle operation transient fuel consumption estimation models based on a multiple regression method are obtained; wherein N is greater than or equal to 2;

[0061] Calculate the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method, and select the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum Bayesian Information Criterion (BIC) value to obtain the optimal transient fuel consumption estimation model;

[0062] The specific process is:

[0063] Step 221: The feature is the product of multiple terms of each variable in the same basic structure;

[0064] The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,1 The variable is v c(1) 、a c(2) ,θ c(3) 、cos c(4) (θ), sin c(5) (θ);

[0065] The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,2 The variables are

[0066] The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,3 The variables are v c(1) 、a c(2) ,θ c(3) 、cos c(4) (θ), sin c(5) (θ);

[0067] Taking the first basic structure of the transient fuel consumption estimation model based on the multiple regression method as an example,

[0068]

[0069] The characteristic is v c(1) a c(2) θ c(3) cos c(4) (θ)sin c(5) (θ), an example is as follows: If c(1) = 2, c(2) = 1, c(3) = 1, c(4) = 1, c(5) = 1, then the feature is v 2 aθcos(θ)sin(θ);

[0070] The correlation between features is analyzed using the Pearson correlation coefficient or the Spearman correlation coefficient. When the correlation between features is greater than 0.7, the correlation between features is considered high, and the corresponding two features are eliminated.

[0071] Step 222: In the basic structure F p,1 、F p,2 、F p,3 Add regular term penalties (addition) to obtain the filtered features;

[0072] Step 223: Based on the filtered features obtained in step 222, obtain N vehicle operation transient fuel consumption estimation models based on the multivariate regression method (similar to the formula in step 23, there are several, and the optimal transient fuel consumption estimation model based on the multivariate regression method is finally selected using step 224 with the minimum BIC value);

[0073] Step 224: Calculate the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method obtained in step 223, and select the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum Bayesian Information Criterion (BIC) value to obtain the optimal transient fuel consumption estimation model;

[0074] Step 2 and 3: Use the variable weighted composite acceleration to replace the acceleration term in the transient fuel consumption estimation model to obtain the final form of the vehicle operation transient fuel consumption estimation model based on the multiple regression method:

[0075]

[0076] Where, F p The power provided to the engine; k0, k1, k2, k3, k4, k5, k6, k7, k8, k9, k 10 、k 11 、k 12 is the fitting coefficient; is the weighted composite acceleration, which is defined as the weighted composite value of the instantaneous acceleration of the first 5 items and the last 5 items at time t (all are instantaneous accelerations. Assuming that the current time is t, the first 5 items are the instantaneous accelerations at times t-5, t-4, t-3, t-2, and t-1, and the last 5 items are the minus signs instead of the plus signs).

[0077] In this embodiment, if Figure 3 As shown in Figure 2, the transient fuel consumption estimation model based on the multivariate regression method has a good fitting effect and can basically capture the overall trend of fuel consumption. The accuracy difference between the training set and the test set is small, and the R 2 The RMSE and MAE of the model are 0.158 ml / s and 0.120 ml / s respectively.

[0078] Table 1 Performance evaluation of transient fuel consumption estimation model

[0079]

[0080] The other steps and parameters are the same as those in the first to third embodiments.

[0081] Specific embodiment 5: This embodiment differs from any one of specific embodiments 1 to 4 in that the regularization term penalty in step 222 is expressed as:

[0082]

[0083] Where L represents the regularization penalty, represents the regularization coefficient, K is the number of coefficients, k i is the i-th coefficient (the coefficient is one more than the feature, and the coefficient is a constant term); the L regularization term takes the sum of the absolute values ​​of the coefficients as the regularization term, which can cause some coefficients in the model to become 0, thereby achieving the effect of feature selection.

[0084] The other steps and parameters are the same as those in the first to fourth embodiments.

[0085] Specific embodiment 6: This embodiment differs from any one of specific embodiments 1 to 5 in that: in step 224, the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method obtained in step 223 is calculated, and the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum Bayesian Information Criterion (BIC) is selected as the optimal transient fuel consumption estimation model;

[0086] The specific process is:

[0087] BIC is the Bayesian Information Criterion, which is calculated as follows:

[0088]

[0089] Where, represents the likelihood function value of the vehicle operation transient fuel consumption estimation model based on the multiple regression method, x represents the number of parameters (coefficients) of the vehicle operation transient fuel consumption estimation model based on the multiple regression method, and y represents the number of samples.

[0090] The other steps and parameters are the same as those in the first to fifth embodiments.

[0091] Specific embodiment 7: This embodiment differs from any one of specific embodiments 1 to 6 in that the weighted composite acceleration in steps 2 and 3 is obtained by the following formula:

[0092]

[0093] Where, Represents weight (different a(t) corresponds to different weights); represents a constant, m is a constant, with a value of 5; a(t) is the vehicle acceleration at time t.

[0094] The other steps and parameters are the same as those in the first to sixth embodiments.

[0095] Specific embodiment 8: This embodiment differs from any one of specific embodiments 1 to 7 in that: in step 3, an ecological vehicle speed management model is constructed based on the model in step 2, and a dynamic programming algorithm is used to solve the ecological vehicle speed management model to obtain a vehicle speed plan; the specific process is as follows:

[0096] Step 3: Figure 4 As shown in the figure, the target journey is evenly divided into M stages according to a certain spatial distance. At the same time, the upper and lower speed limits of each stage are evenly discretized to obtain n speed states, and the travel time and acceleration of each stage are calculated.

[0097] The specific calculation method is:

[0098] T m,i,j =2Δs / (v m,i +v m+1,j )

[0099] a m,i,j =(v m+1,j 2 -v m,i 2 ) / 2Δs

[0100] Where, T m,i,j It represents the travel time consumed when the speed state i in stage m is transferred to the speed state j in stage m+1. m, i, j are counter variables, 1≤m≤M, 1≤i≤n, 1≤j≤n;

[0101] a m,i,j It represents the acceleration consumed when the speed state i in stage m is transferred to the speed state j in stage m+1;

[0102] Δs represents the length of each stage;

[0103] v m,i Indicates the speed value of the i-th state in stage m;

[0104] v m+1,j Indicates the speed value of the jth state in stage m+1;

[0105] Step 3.2: Establish an ecological vehicle speed management model with the minimum total fuel consumption as the optimization goal

[0106] Step 33: Use the dynamic programming algorithm to solve the ecological vehicle speed management model established in step 32 to obtain the vehicle speed plan with the minimum fuel consumption.

[0107] The other steps and parameters are the same as those in the first to seventh embodiments.

[0108] Specific embodiment 9: This embodiment differs from specific embodiments 1 to 8 in that: in step 32, an ecological vehicle speed management model is established with the minimum total trip fuel consumption as the optimization goal. The specific process is:

[0109]

[0110]

[0111] Where, FC m,i,j represents the amount of fuel consumed when the speed state i in stage m is transferred to the speed state j in stage m+1; f(v m,i ,am,i,j ,s m ) is the transient fuel consumption rate, indicating that the vehicle is at v m,i 、a m,i,j , road slope s m The fuel consumption rate under combined conditions is estimated by the vehicle operation transient fuel consumption model based on the multiple regression method Calculated;

[0112] f(v m,i ,a m,i,j ,s m )=F p / (λ×unit fuel calorific value)

[0113] λ is the loss coefficient;

[0114] s m represents the road slope at stage m;

[0115] v m,min 、v m,max are the minimum and maximum speed limits of stage m respectively;

[0116] v m+1,min 、v m+1,max are the minimum and maximum speed limits of stage m+1 respectively;

[0117] a max 、a min They are the maximum and minimum acceleration values ​​allowed while ensuring driving comfort.

[0118] The other steps and parameters are the same as those in Specific Embodiments 1 to 8.

[0119] Specific embodiment 10: This embodiment differs from specific embodiments 1 to 9 in that: in step 33, a dynamic programming algorithm is used to solve the ecological vehicle speed management model established in step 32 to obtain a vehicle speed plan with the minimum fuel consumption during the trip; the specific process is as follows:

[0120] Step 331: Set stage m=M, i=1, j=1, Q M+1,j =0; execute step 332;

[0121] Where Q M+1,j is the optimal objective function value of stage M+1;

[0122] Step 332: Set the current stage speed to v m,i , the speed of the next stage is v m+1,j , determine the speed v m,i and v m+1,j And whether the acceleration required for the speed change satisfies the constraint (v m,min <vm,i <v m,max 、v m+1,min <v m+1,j <v m+1,max 、a min m,i,j max );

[0123] If the constraints are met, the ecological speed management model (Ecological Speed ​​Management Model

[0124] The travel time T consumed when the speed state i in stage m is transferred to the speed state j in stage m+1 m,i,j , calculate the fuel consumption when the speed state i in stage m is transferred to the speed state j in stage m+1

[0125] If the constraint is not satisfied, set j = j + 1 and repeat step 332 until j = n; execute step 333;

[0126] Step 333: Let the current stage speed v m,i Transfer to next stage speed v m+1,j The minimum cost when transferring to the end speed v0 is recorded as Q m,i,j , Q m,i,j =FC m,i,j +Q m+1,i ;

[0127] Let the current stage speed v m,i The minimum cost when transferring to the end speed v0 is recorded as Q m,i , Q m,i =min{Q m,i,1 ,Q m,i,2 ,...,Q m,i,j ,...,Q m,i,N};

[0128] Q m,i The corresponding Q m,i,j The j value is recorded in the optimal transfer record R m,i , execute steps 334;

[0129] Step 334: Let i = i + 1, j = 1, repeat steps 332 and 333, and calculate the minimum cost Q for each vehicle speed in the current stage to the end speed state v0 of the trip. m,i ;

[0130] Until i=n; execute step 335;

[0131] ​​Step 335: Let m = m-1, i = 1, j = 1, repeat steps 332 to 334 until m = 0, and get the minimum cost Q m,i , the minimum cost Q m,i Corresponding Q m,i,j The j value corresponding to Medium v m,i,j is the vehicle speed at stage m in the target journey.

[0132] The other steps and parameters are the same as those in Specific Embodiments 1 to 9.

[0133]

[0134] Q n+1 (x n+1 )=0

[0135] Where Q m (x m ) is the optimal objective function of stage m; n is the total number of stages; x m represents the state variable of stage m; u m (x m ) represents the value of the state variable x in the mth stage m The decision variable when y m (x m ,u m ) represents the objective function value of stage m; D m (x m ) indicates that the mth stage starts from state x m The set of allowed decisions to start; Q n+1 (x n+1 ) represents the optimal objective function of stage n+1; x n+1 Represents the state variable of stage n+1;

[0136] In this embodiment, if Figure 5 As shown in the figure, the speed plan obtained by solving the ecological speed management model using dynamic programming achieved a reduction in fuel consumption. Compared with the cruise control mode, the total fuel consumption was reduced by 2.05%. At the same time, the average speed was slightly improved, which reduced the total travel time by 2.37%, achieving the goal of fuel saving while ensuring the travel time.

[0137] Table 2 Comparison of results of ecological speed management model

[0138]

[0139] The present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. An ecological vehicle speed management method based on dynamic programming, characterized by: The specific process of the method is: Step 1: Collect vehicle driving trajectory data and vehicle fuel consumption data, and process the collected vehicle driving trajectory data and vehicle fuel consumption data to obtain processed data; Step 2: Based on the processed data, a vehicle operation transient fuel consumption estimation model based on the multivariate regression method is established; The specific process is: Step 2: Build the basic structure of the transient fuel consumption estimation model based on the multiple regression method. The expression is: Where, v is the speed of the ambient wind relative to the vehicle; θ is the road slope; a is the vehicle acceleration; p air Indicates v 3 ;p r represents vcosθ; p g represents vsinθ; p acc means va; F p,1 、F p,2 、F p,3 Indicates the power provided by the engine; c(1) represents the polynomial degree of the ambient wind speed v relative to the vehicle; c(2) represents the polynomial degree of the vehicle acceleration a; c(3) represents the polynomial degree of the road slope θ; c(4) represents the polynomial degree of cos(θ); c(5) represents the polynomial degree of sin(θ); N(1) represents the maximum value of the polynomial degree of the ambient wind speed v relative to the vehicle; N(2) represents the maximum value of the polynomial degree of the vehicle acceleration a; N(3) represents the maximum value of the polynomial degree of the road slope θ; N(4) represents the maximum value of the polynomial degree of cos(θ); N(5) represents the maximum value of the polynomial degree of sin(θ); k c(1),c(2),c(3),c(4),c(5) Indicates v c(1) a c(2) θ c(3) cos c(4) (θ)sin c(5) (θ) simulation fitting coefficients; c(6) represents p air The degree of the polynomial; c(7) represents p r The degree of the polynomial; c(8) represents p g The degree of the polynomial; c(9) represents p acc The degree of the polynomial; N(6) represents p air The maximum value of the polynomial degree; N(7) represents p r The maximum value of the polynomial degree; N(8) represents p g The maximum value of the polynomial degree; N(9) represents p acc The maximum value of the polynomial degree; k c(6),c(7),c(8),c(9) express The simulation fitting coefficients of k c(1),c(2),c(3),c(4),c(5),c(6),c(7),c(8),c(9) express v c(1) a c(2) θ c(3) cos c(4) (θ)sin c(5) (θ) simulation fitting coefficients; Step 22: Calculate the correlation between the features in each basic structure and remove the features with high correlation; Based on eliminating highly correlated features, a regularization penalty is added to each basic structure to obtain filtered features; Based on the obtained filtered features, N vehicle operation transient fuel consumption estimation models based on a multiple regression method are obtained; wherein N is greater than or equal to 2; Calculate the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method, and select the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum Bayesian Information Criterion (BIC) value to obtain the optimal transient fuel consumption estimation model; Step 2 and 3: Use the variable weighted composite acceleration to replace the acceleration term in the transient fuel consumption estimation model to obtain the final form of the vehicle operation transient fuel consumption estimation model based on the multiple regression method: Where, F p The power provided to the engine; k0, k1, k2, k3, k4, k5, k6, k7, k8, k9, k 10 、k 11 、k 12 is the fitting coefficient; is the variable weighted composite acceleration; Step 3: Based on the model in step 2, an ecological speed management model is constructed, and a dynamic programming algorithm is used to solve the ecological speed management model to obtain a speed plan.

2. The ecological vehicle speed management method based on dynamic programming according to claim 1, characterized in that: In step 1, the vehicle driving trajectory data and vehicle fuel consumption data are collected; the specific process is: Collect vehicle driving trajectory data through GPS equipment, and use the on-board diagnostic system to read vehicle master control data from the vehicle OBD port to obtain fuel consumption data; The vehicle driving trajectory data includes the vehicle's latitude and longitude coordinates, vehicle motion posture, and acquisition accuracy information; The vehicle motion posture is yaw angle, pitch angle, roll angle, gyro X axis, gyro Y axis, gyro Z axis, plus table X axis, plus table Y axis, plus table Z axis, east speed, north speed, sky speed, vehicle speed; The X-axis is the vehicle's X-axis acceleration, the Y-axis is the vehicle's Y-axis acceleration, and the Z-axis is the vehicle's Z-axis acceleration. The vehicle OBD port includes sampling time, fuel consumption rate, engine speed, vehicle speed, accelerator pedal position, engine torque and vehicle acceleration; The vehicle fuel consumption data includes fuel consumption rate, engine speed, vehicle speed, accelerator pedal position, engine torque and vehicle acceleration data.

3. The ecological vehicle speed management method based on dynamic programming according to claim 2, characterized in that: In step 1, the collected vehicle driving trajectory data and vehicle fuel consumption data are processed; the specific process is as follows: 1) Delete abnormal values ​​of vehicle fuel consumption data less than 0ml / s; 2) Based on the result of 1), according to the GPS positioning data, delete the vehicle location data outside the actual road section and the corresponding vehicle fuel consumption data; 3) Based on the result of 2), according to the GPS positioning data, delete the vehicle location data and corresponding vehicle fuel consumption data under GPS failure; 4) Based on the result of 3), calculate the instantaneous vehicle specific power according to formula (1); VSP=v(1.1a+9.81i r +0.132)+0.000302v 3 (1) Where VSP is the vehicle specific power in kW / s; v is the vehicle speed in m / s; a is the vehicle acceleration in m / s 2 ; i r Indicates the road slope ratio, in %; The vehicle specific power is divided into L intervals with an interval of 2kW / s. Then, the fuel consumption data exceeding three times the standard deviation in each interval is eliminated according to the 3σ principle. 5) Based on the results of 4), the Ramer-Douglas-Peucker algorithm is used to identify the segments in the trajectory. The endpoints of each segment are used as segmentation points. The ratio of the height difference between the two endpoints of the segment to the distance is calculated as the road slope of this segment. 6) Based on the result of 5), the vehicle speed and acceleration data of step 1 are smoothed using a Savitzky-Golay filter.

4. The ecological vehicle speed management method based on dynamic programming according to claim 3 is characterized by: In step 22, the correlation between the features in each basic structure is calculated, and the features with high correlation are eliminated; Based on eliminating highly correlated features, a regularization penalty is added to each basic structure to obtain filtered features; Based on the obtained filtered features, N vehicle operation transient fuel consumption estimation models based on a multiple regression method are obtained; wherein N is greater than or equal to 2; Calculate the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method, and select the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum Bayesian Information Criterion (BIC) value to obtain the optimal transient fuel consumption estimation model; The specific process is: Step 221: The feature is the product of multiple terms of each variable in the same basic structure; The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,1 The variable is v c(1) 、a c(2) ,θ c(3) 、cos c (4) (θ), sin c(5) (θ); The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,2 The variables are The basic structure of the transient fuel consumption estimation model based on the multiple regression method F p,3 The variables are v c(1) 、a c(2) ,θ c(3) 、cos c(4) (θ), sin c(5) (θ); The correlation between features is analyzed using the Pearson correlation coefficient or the Spearman correlation coefficient. When the correlation between features is greater than 0.7, the correlation between features is considered high, and the corresponding two features are eliminated. Step 222: In the basic structure F p,1 、F p,2 、F p,3 Regularization penalties are added to obtain the filtered features; Step 223: Based on the filtered features obtained in step 222, obtain N vehicle operation transient fuel consumption estimation models based on the multivariate regression method; Step 224: Calculate the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multivariate regression method obtained in step 223, and select the optimal transient fuel consumption estimation model based on the multivariate regression method corresponding to the minimum value of the Bayesian Information Criterion (BIC).

5. The ecological vehicle speed management method based on dynamic programming according to claim 4 is characterized by: The regularization term penalty in step 222 is expressed as: Where L represents the regularization penalty, represents the regularization coefficient, K is the number of coefficients, k i is the i-th coefficient.

6. The ecological vehicle speed management method based on dynamic programming according to claim 5, characterized in that: In step 224, the Bayesian Information Criterion (BIC) of the N vehicle operation transient fuel consumption estimation models based on the multiple regression method obtained in step 223 is calculated, and the vehicle operation transient fuel consumption estimation model based on the multiple regression method corresponding to the minimum value of the Bayesian Information Criterion (BIC) is selected as the optimal transient fuel consumption estimation model; The specific process is: BIC is the Bayesian Information Criterion and is calculated as: Where, represents the likelihood function value of the vehicle operation transient fuel consumption estimation model based on the multiple regression method, x represents the number of parameters of the vehicle operation transient fuel consumption estimation model based on the multiple regression method, and y represents the number of samples.

7. The ecological vehicle speed management method based on dynamic programming according to claim 6, characterized in that: The variable weighted composite acceleration in steps 2 and 3 is obtained by the following formula: Where, represents weight; represents a constant, m is a constant, with a value of 5; a(t) is the vehicle acceleration at time t.

8. The ecological vehicle speed management method based on dynamic programming according to claim 7 is characterized by: In step 3, an ecological speed management model is constructed based on the model in step 2, and a dynamic programming algorithm is used to solve the ecological speed management model to obtain a speed plan. The specific process is as follows: Step 3: Divide the target journey into M stages at equal intervals. Discretize the upper and lower speed limits of each stage to obtain n speed states. Calculate the travel time and acceleration of each stage. The specific calculation method is: T m,i,j =2Δs / (v m,i +v m+1,j ) a m,i,j =(v m+1,j 2 -v m,i 2 ) / 2Δs Where, T m,i,j It represents the travel time consumed when the speed state i in stage m is transferred to the speed state j in stage m+1. m, i, j are counter variables, 1≤m≤M, 1≤i≤n, 1≤j≤n; a m,i,j It represents the acceleration consumed when the speed state i in stage m is transferred to the speed state j in stage m+1; Δs represents the length of each stage; v m,i Indicates the speed value of the i-th state in stage m; v m+1,j Indicates the speed value of the jth state in stage m+1; Step 32: Establish an ecological vehicle speed management model with the minimum total trip fuel consumption as the optimization goal; Step 33: Use the dynamic programming algorithm to solve the ecological vehicle speed management model established in step 32 to obtain the vehicle speed plan with the minimum fuel consumption.

9. The ecological vehicle speed management method based on dynamic programming according to claim 8, characterized in that: In step 32, an ecological vehicle speed management model is established with the minimum total trip fuel consumption as the optimization goal; the specific process is as follows: s.t.v m,min <v m,i <v m,max v m+1,min <v m+1,j <v m+1,max a min <a m,i,j <a max Where, FC m,i,j represents the amount of fuel consumed when the speed state i in stage m is transferred to the speed state j in stage m+1; f(v m,i ,a m,i,j ,s m ) is the transient fuel consumption rate, indicating that the vehicle is at v m,i 、a m,i,j , road slope s m The fuel consumption rate under combined conditions is estimated by the vehicle operation transient fuel consumption model based on the multiple regression method Calculated; f(v m,i ,a m,i,j ,s m )=F p / (λ×unit fuel calorific value) λ is the loss coefficient; s m represents the road slope at stage m; v m,min 、v m,max are the minimum and maximum speed limits of stage m respectively; v m+1,min 、v m+1,max are the minimum and maximum speed limits of stage m+1 respectively; a max 、a min They are the maximum and minimum acceleration values ​​allowed respectively.

10. The ecological vehicle speed management method based on dynamic programming according to claim 9, characterized in that: In step 33, the dynamic programming algorithm is used to solve the ecological vehicle speed management model established in step 32 to obtain the vehicle speed plan with the minimum fuel consumption. The specific process is as follows: Step 331: Set stage m=M, i=1, j=1, Q M+1,j =0; execute step 332; Where Q M+1,j is the optimal objective function value of stage M+1; Step 332: Set the current stage speed to v m,i , the speed of the next stage is v m+1,j , determine the speed v m,i and v m+1,j and whether the acceleration required for the speed change satisfies the constraints; If the constraints are met, the driving time T consumed when the speed state i in stage m is transferred to the speed state j in stage m+1 through the ecological vehicle speed management model is m,i,j , calculate the fuel consumption when the speed state i in stage m is transferred to the speed state j in stage m+1 If the constraint is not satisfied, set j = j + 1 and repeat step 332 until j = n; execute step 333; Step 333: Let the current stage speed v m,i Transfer to next stage speed v m+1,j The minimum cost when transferring to the end speed v0 is recorded as Q m,i,j , Q m,i,j =FC m,i,j +Q m+1,i ; Let the current stage speed v m,i The minimum cost when transferring to the end speed v0 is recorded as Q m,i , Q m,i =min{Q m,i,1 ,Q m,i,2 ,...,Q m,i,j ,...,Q m,i,N }; Q m,i The corresponding Q m,i,j The j value is recorded in the optimal transfer record R m,i , execute steps 334; Step 334: Let i = i + 1, j = 1, repeat steps 332 and 333, and calculate the minimum cost Q for each vehicle speed in the current stage to the end speed state v0 of the trip. m,i ; Until i=n; execute step 335; Step 335: Let m = m-1, i = 1, j = 1, repeat steps 332 to 334 until m = 0, and get the minimum cost Q m,i , the minimum cost Q m,i Corresponding Q m,i,j The j value corresponding to Medium v m,i,j is the vehicle speed at stage m in the target journey.

Images