An economical driving method for intelligent connected new energy vehicles under traffic light scenarios

Abstract

The present invention relates to an economical driving method for an intelligent network-connected new energy vehicle under a traffic light scenario, belonging to the field of new energy vehicles, and comprising the following steps: constructing a traffic flow prediction model based on a time graph convolutional network, and predicting the traffic flow speed v at a future moment by acquiring historical traffic flow speed information. pre (t); obtain the phase timing information of the signal light, and refer to the predicted traffic flow speed v at the future moment pre (t), plan the global guidance speed vg(t) in the cloud; build an economical driving cruise controller based on model predictive control, referring to the global guidance speed v g (t), real-time tracking of the front vehicle speed, solving the upper target speed v upper (t) and the total required power P of the vehicle d (t); Develop the lower-level energy management strategy, build the vehicle powertrain model, and use the equivalent fuel minimum consumption strategy to calculate the optimal power distribution. The present invention adds global vehicle speed guidance and signal light constraints to the cruise control optimization problem, effectively combines global speed planning with vehicle-following cruise control, and can significantly improve the economy and comfort of the vehicle.

Description

Technical Field

[0001] The present invention belongs to the field of new energy vehicles and relates to an economical driving method for an intelligent networked new energy vehicle under a traffic light scenario. Background Art

[0002] Energy-saving and emission-reduction technology has always been one of the hot spots in the development of the automotive industry. However, as the "marginal effect" of traditional internal combustion engine technology on reducing fuel consumption becomes increasingly significant, economical driving technology that improves vehicle driving strategies has become a new track for energy conservation and emission reduction. This technology encourages the driver or the autonomous driving system to control the vehicle in a more energy-efficient way and rationally allocate vehicle power to adapt to road traffic conditions. Economical driving technology does not rely on traditional internal combustion engine technology, is easy to promote, and is applicable to a variety of vehicle models and traffic scenarios. With the continuous development of intelligent connected vehicle technology, this technology, combined with high-performance computing platforms and vehicle-road-cloud communications, has broad prospects and is expected to be promoted globally to achieve green and sustainable development in the transportation field.

[0003] In complex urban traffic control, economical driving technology relies on real-time acquisition of information about traffic lights and surrounding vehicles to adjust vehicle speed, reduce waiting and frequent speed changes, and reduce energy consumption and emissions. Although advances in communication technology can more accurately obtain this information, improve energy conservation and optimize road network efficiency to a certain extent, existing technologies fail to fully utilize the potential of vehicle control, speed optimization, and vehicle-road-cloud collaboration, and have problems such as insufficient information integration capabilities and lack of personalized driving strategies. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide an economical driving method for intelligent networked new energy vehicles under traffic light scenarios, and to realize an economical driving strategy for hybrid new energy networked vehicles under a vehicle-energy-road-cloud integrated intelligent transportation system. The present invention proposes an economical driving method for intelligent networked new energy vehicles under traffic light scenarios. By integrating the intelligent transportation system and the vehicle-road-cloud collaborative technology, and utilizing advanced on-board sensors and cloud computing technology, it processes information such as traffic signals, flow, and road conditions in real time to achieve more accurate vehicle control and speed optimization. This method not only significantly reduces energy consumption and emissions, but also improves the overall operating efficiency of the road network, makes the driving process more intelligent, and enhances driving safety and comfort, realizing the practical application value of economical driving technology.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for economical driving of an intelligent network-connected new energy vehicle under a traffic light scenario comprises the following steps:

[0007] S1: Construct a traffic flow prediction model based on a time graph convolutional network to predict the future traffic flow speed v by obtaining historical traffic flow speed information pre (t);

[0008] S2: Obtain the phase timing information of the traffic lights, and refer to the predicted traffic flow speed v at the future moment pre (t), planning the global guidance speed v in the cloud g (t);

[0009] S3: Construct an economical driving cruise controller based on model predictive control, referring to the global guidance speed v g (t), real-time tracking of the front vehicle speed, solving the upper target speed v upper (t) and the total required power P of the vehicle d (t);

[0010] S4: Develop the lower-level energy management strategy, build the vehicle powertrain model, and calculate the optimal power distribution using the equivalent fuel minimum consumption strategy.

[0011] Further, in step S1, the following steps are included:

[0012] S11: Construct a traffic flow prediction model based on a temporal convolutional network, which at least includes an input end taking the network topology as input, a two-layer graph convolutional network GCN model for extracting temporal features, and a gated recurrent unit GRU model for extracting spatial features;

[0013] S12: Using historical traffic speed information with road network topology information as network input, the speed prediction results for the current time and future time are obtained through the graph convolutional network GCN model and the gated recurrent unit GRU model, where:

[0014] The unweighted graph G = (V, E) is used to describe the road network topology, where V = {v1, v2, ..., v N}, V represents the road node set, N represents the number of nodes, and E represents the edge set, thereby establishing an adjacency matrix A representing the connections between roads;

[0015] Get traffic flow speed information as feature matrix X N×P The attribute features in are represented by X∈R N×P , P represents the number of node attribute features, that is, the length of the historical time series;

[0016] The GCN model is used to learn spatial features from traffic data. The two-layer GCN model is expressed as:

[0017]

[0018] Where X and A represent the feature matrix and adjacency matrix respectively. represents the preprocessing step, is the self-connected structure matrix, is an angle matrix, σ(·) represents xxx; W0 and W1 represent the weight matrices of the first and second layers respectively; f(A, X t ) represents the graph convolution process, W and b represent the weight and bias in the training process respectively;

[0019] Select the GRU model to capture the temporal dependency of traffic data and use the update gate u t Control the degree to which the previous state information affects the current state, using the reset gate r t Control the degree of ignoring the state information of the previous moment, using the previous moment h t-1 The hidden state and current traffic information are used as input to obtain the current traffic status. The temporal graph convolution network built by combining graph convolution and gated recurrent unit has the following specific calculation process:

[0020] u t =σ(W u [f(A,X t ), h t-1 ]+b u )

[0021] r t =σ(W r [f(A,X t ), h t-1 ]+b r )

[0022] c t =tanh(W c [f(A,X t ), (r t ·h t-1 )]+b c )

[0023] h t =u t ·h t-1 +(1-u t )·c t

[0024] Among them, u t represents the value of the update gate, W u represents the weight matrix of the update gate, X t represents the input features at time t, h t and h t-1 Represent the hidden states of the current moment and the previous moment, respectively, b u represents the bias vector of the update gate; r tRepresents the value of the reset gate, W r represents the weight matrix of the reset gate, b r represents the bias vector of the reset gate; c t represents the candidate hidden state, tanh and σ are activation functions, W c represents the weight matrix of the candidate hidden state, b c represents the bias vector of the candidate hidden state, f(A, X t ) represents the adjacency matrix A and input feature X t The convolution operation performed;

[0025] By processing spatial dependency and temporal dynamics through temporal graph convolution, the topological structure of the urban road network and the periodic characteristics of traffic lights on the traffic flow are captured to complete the task of traffic flow prediction;

[0026] S13: Filter and interpolate the speed prediction result to optimize the prediction result to the speed characteristic v of the vehicle running normally pre (t).

[0027] Further, in step S2, the following steps are included:

[0028] S21: Obtain the phase timing information of the traffic lights and use the improved intelligent driving model to solve the upper and lower limits s of the vehicle's driving trajectory when passing through the entire road section in all green light windows upper 、s lower ;

[0029] S22: Use dynamic programming algorithm to calculate the preliminary global guidance trajectory s of the vehicle dp (t k ) and vehicle speed v dp (t k );

[0030] S23: Based on the optimization result in step S22, use quadratic programming to smooth the trajectory and calculate the global guiding vehicle speed v g (t).

[0031] Further, in step S21, the improved intelligent driving model solves the upper and lower limits s of the driving trajectory according to the signal light phase timing information. upper 、s lower In the process, the model is expressed as:

[0032]

[0033]

[0034] where a IDM is the acceleration of the model, Δv and Δx are the relative velocity and relative position, a max represents the maximum acceleration, v represents the vehicle speed, and vmax represents the maximum vehicle speed, s(v, Δv) represents the expected safety distance, s0 represents the minimum safety distance, t hw represents the time interval. In this process, the red light is regarded as a prohibition of obstructing vehicles. According to this model, the maximum and minimum speed trajectories to reach the next intersection are calculated. If the light is green, the speed remains unchanged.

[0035] In step S22, a dynamic programming algorithm is used to calculate the preliminary global guidance trajectory s of the vehicle. dp (t k ) and vehicle speed v dp (t k ) is:

[0036]

[0037] Among them, Represents the weight coefficient of each item, the first item represents the time efficiency cost, where t end is the total time; the second term represents the reference speed cost, where v is the vehicle speed, v ref is the reference vehicle speed; the third term represents the comfort cost, where a is the ego vehicle acceleration; the fourth term represents the trajectory constraint cost, where s is the ego vehicle trajectory, s upper 、s lower are the upper and lower limits of the reference trajectory; k = 0, 1, 2, ... ∈ [0, t end ] for each time step in the optimization domain.

[0038] In step S23, the global guiding vehicle speed v is calculated using the quadratic programming smooth trajectory g The cost function of (t) is:

[0039]

[0040] Among them, ω1, ω2, ω3, and ω4 represent the weight coefficients of each item. The first item represents the position cost, the second item represents the speed cost, the third item represents the acceleration cost, and the fourth item represents the impact cost; S dp is the reference data after solving the dynamic programming algorithm, s dp ′ is the reference speed for dynamic programming solution, s dp ″ is the reference acceleration for dynamic programming solution, s dp ″′ is the reference impact degree of dynamic programming solution; s is the reference position of dynamic programming solution, s′ is the speed, s″ is the acceleration, and s″′ is the impact degree.

[0041] Further, in step S3, the following steps are specifically included:

[0042] S31: Establishing the vehicle longitudinal dynamics model;

[0043] S32: Constructing an economical driving cruise controller based on model predictive control;

[0044] S33: Design constraints, including the distance constraint from the preceding vehicle, the distance constraint from the traffic light, the road speed limit constraint, and the maximum acceleration constraint;

[0045] S34: Design the controller cost function, including reference speed cost, equivalent fuel consumption cost, control input cost, input change rate cost, and soft constraint cost when the traffic light is yellow;

[0046] S35: Solve the economical cruise driving control strategy in the cloud and convert the upper target vehicle speed v upper (t) and the total required power P of the vehicle d (t) is sent to the vehicle-side controller.

[0047] Further, in step S31, the vehicle longitudinal dynamics model is:

[0048] m·a(t)=F t -F r

[0049]

[0050]

[0051] where F t (t) is the traction force, T w (t) is the wheel torque, r w is the tire radius, represents air resistance, ρ is air density, A is windward area, C d is the air resistance coefficient, mgC r Indicates rolling resistance, C r is the rolling resistance coefficient of the road, and α is the road slope.

[0052] In step S32, the state equation of economical cruise driving is:

[0053]

[0054] The state variable x includes the distance d between the vehicle and the nearest traffic light. light , distance d from the vehicle in front f and the vehicle speed v, v f is the speed of the front vehicle, and the control input u is the wheel torque T w (t), where v f is the speed of the preceding vehicle and the speed of the preceding vehicle is set to the predicted speed v pre (t), r w is the tire radius, Fr is the vehicle resistance, m is the vehicle mass, t s is the time step.

[0055] In step S33, the distance constraint to the preceding vehicle, the distance constraint to the traffic light, the road speed limit constraint, and the maximum acceleration constraint are respectively expressed as:

[0056] d min ≤v(t)·t hw +d0

[0057] 0≤d light (t), if light=red

[0058] 0≤d light (t)+φ(t), if light=yellow

[0059] v min ≤v(t)≤v max

[0060] a min ≤a(t)≤a max

[0061] Where φ(t) is a slack variable and is greater than 0, which restricts vehicles to comply with traffic light rules and cannot pass through traffic light intersections when the light is red or yellow. hw is the time distance of following the vehicle, d0 is the minimum distance of following the vehicle, d min Represents the current minimum following distance of the vehicle, v min 、v max 、a min 、a max Represent the minimum and maximum speed and acceleration of the vehicle respectively.

[0062] In step S4, the controller cost function is expressed as:

[0063]

[0064] Each term in the function represents the reference speed cost, equivalent fuel consumption cost, control input cost, input change rate cost, and soft constraint cost when the traffic light is yellow, ω v ,ω c ,ω u ,ω du ,ω φ Represents the weight factor of each cost, P d (t) is the total power required by the vehicle, P d (t) = T w (t)·ω w (t),ω w is the wheel speed, for Δ∈(t,…,t+NP -1), where x(Δ|t) and u(Δ|t) are the measured state and input control quantity based on time t and the predicted state and input control quantity at time Δ, respectively. P is the prediction time domain, and the optimal control problem is solved by rolling optimization.

[0065] Further, in step S4, the following steps are specifically included:

[0066] S41: Establish a hybrid vehicle power system model, including engine model, motor model and battery model;

[0067] S42: Establish a hybrid vehicle power demand model;

[0068] S43: Establishing the cost function of the energy management problem of the equivalent fuel minimum consumption strategy;

[0069] S44: Combining the cost function and the constraints in the above steps, the optimal power control strategy is solved on the vehicle side, which is also its optimal energy management power allocation strategy.

[0070] Further, in step S41, the engine model is:

[0071]

[0072] P eng =ω eng ·T eng

[0073] in is the instantaneous fuel consumption rate, P eng is the engine power, η eng is the engine efficiency, which is determined by the engine torque T eng and speed ω eng Determine, Q is the lower heating value of the fuel;

[0074] The motor model is:

[0075]

[0076] Where P mot is the motor power, η mot (T mot ,ω mot ) is the motor efficiency, which is determined by the motor torque T mot and speed ω mot Decision, P elec is the power allocated to the motor. The motor power is greater than or equal to zero and less than zero in motion and power generation modes respectively. elec Less than zero means the motor is operating in generating mode;

[0077] The battery model is:

[0078]

[0079] Where P bat is the battery power.

[0080] In step S42, the power demand model of the hybrid vehicle is:

[0081] P mot，out =T mot ·ω mot

[0082] P mot，total =P mot，out +P mot，loss

[0083] P eng，out =T eng ·ω eng

[0084] P eng，total =P eng，out +P eng，loss

[0085] P bat，total =P bat，out +P bat，loss

[0086] Among them, P mot，out , T mot ,ω mot is the motor output power, torque and speed, P eng，out , T eng ,ω eng is the engine output power, torque, and speed, P mot，total , P mot，loss is the total power and power loss of the motor, P bat，total , P bat，out , P bat，loss The total power required by the hybrid vehicle is the sum of the power of the motor and the engine.

[0087] Further, in step S43, the cost function of the energy management problem of the equivalent fuel minimum consumption strategy is expressed as:

[0088]

[0089]

[0090] The upper formula represents the cost function of the energy management strategy, and the lower formula expresses the total fuel consumption rate Equal to the engine fuel consumption rate Equivalent fuel consumption with battery The equivalent fuel consumption of the battery is λ is the equivalent factor, P bat is the battery power; solving the optimal power control strategy on the vehicle side is also its optimal energy management power allocation strategy.

[0091] Further, in step S44, the constraint boundary is expressed as:

[0092]

[0093]

[0094]

[0095] E bat (t)∈[SOC min , SOC max ]·V oc Q

[0096] E bat (0) = E bat (N)

[0097] Among them, the first one represents the constraint of motor output torque, the second one represents the constraint of engine output torque, the third one represents the constraint of battery output power, and the fourth and fifth ones represent the constraint of SOC; E bat is the battery energy, V oc is the open circuit voltage of the battery, Q represents the battery capacity, and the subscripts min and max represent the upper and lower limits of the parameter.

[0098] The beneficial effects of the present invention are:

[0099] The present invention considers historical traffic data, predicts traffic flow speed, and uses the speed as one of the constraints of global planning, thereby improving the accuracy of global speed planning.

[0100] The present invention adds reference vehicle speed guidance and traffic light constraints of global planning to the cruise control optimization problem, which is more in line with actual traffic conditions, reduces vehicle idling and parking behaviors, improves vehicle economy and comfort, and effectively combines global speed planning with vehicle-following cruise control.

[0101] The present invention adopts a hierarchical control strategy, wherein the upper layer performs speed planning and the lower layer performs energy management, thereby ensuring real-time performance, economy and safety.

[0102] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0104] Figure 1 It is the overall framework diagram of the present invention;

[0105] Figure 2 It is a vehicle speed planning framework diagram in the upper cloud planning of the present invention;

[0106] Figure 3 It is a time graph convolution network framework and prediction process diagram of the present invention;

[0107] Figure 4 A simplified schematic diagram of a hybrid vehicle configuration. DETAILED DESCRIPTION

[0108] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0109] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0110] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0111] See also Figure 1 to Figure 4 , which is an economical driving method for intelligent connected new energy vehicles under traffic light scenarios.

[0112] like Figure 1 The overall architecture diagram of an economical driving method for an intelligent networked new energy vehicle in a traffic light scenario of the present invention is shown, which includes the following steps:

[0113] S1: Obtain historical traffic flow speed information, build a traffic flow prediction model based on the time graph convolutional network, and predict the traffic flow speed v at the future moment pre (t);

[0114] S2: Obtain the phase timing information of the traffic lights, refer to the predicted traffic flow speed at the future moment, and plan the global guidance speed v in the cloud g (t);

[0115] S3: Design an economical driving cruise controller based on model predictive control, referring to the global guide speed v g (t), real-time tracking of the front vehicle speed, solving the upper target speed v upper (t) and the total required power P of the vehicle d (t);

[0116] S4: Develop the lower-level energy management strategy, build the vehicle powertrain model, and calculate the optimal power distribution using the equivalent fuel minimum consumption strategy.

[0117] Specifically, it includes upper-layer cloud-side planning and lower-layer vehicle-side optimization, among which the upper-layer cloud-side planning is used for speed planning and the lower-layer vehicle-side optimization is used for energy management.

[0118] In this embodiment, steps S1-S3 describe the upper cloud planning, that is, the process of speed planning. The overall process is as follows: Figure 2 As shown, specifically:

[0119] S1: Obtain historical traffic flow speed information, build a traffic flow prediction model based on the time graph convolutional network, and predict the traffic flow speed v at the future moment pre (t), specifically comprising the following steps:

[0120] S11: Construct a traffic flow prediction model based on temporal graph convolutional network. The network framework and prediction process are as follows Figure 3 As shown, it at least includes an input end that uses road network topology as input, a two-layer graph convolutional network GCN model for extracting temporal features, and a gated recurrent unit GRU model for extracting spatial features.

[0121] S12: Using historical traffic speed information with road network topology information as network input, the speed prediction results for the current time and future time are obtained through the graph convolutional network GCN model and the gated recurrent unit GRU model;

[0122] The unweighted graph G = (V, E) is used to describe the road network topology, where V = {v1, v2, ..., v N}, V represents the road node set, N represents the number of nodes, and E represents the edge set, thereby establishing the adjacency matrix A to represent the connection between roads. Obtain traffic flow speed information as the feature matrix X N×P The attribute features in are represented by X∈R N×P , P represents the number of node attribute features (the length of the historical time series). Using the GCN model to learn spatial features from traffic data, the two-layer GCN model can be expressed as:

[0123]

[0124] Where X and A represent the feature matrix and adjacency matrix respectively. represents the preprocessing step, is the self-connected structure matrix, is a diagonal matrix, W0 and W1 represent the weight matrices of the first and second layers respectively. t ) represents the graph convolution process, W and b represent the weight and bias in the training process respectively;

[0125] Select the GRU model to capture the temporal dependency of traffic data and use the update gate u t Control the degree to which the previous state information affects the current state, using the reset gate r t Control the degree of ignoring the state information of the previous moment, using the previous moment h t-1 The hidden state and current traffic information are used as input to obtain the current traffic status. The temporal graph convolution network built by combining graph convolution and gated recurrent unit has the following specific calculation process:

[0126] ut =σ(W u [f(A,X t ), h t-1 ]+b u )

[0127] r t =σ(W r [f(A,X t ), h t-1 ]+b r )

[0128] c t =tanh(W c [f(A,X t ), (r t ·h t-1 )]+b c )

[0129] h t =u t ·h t-1 +(1-u t )·c t

[0130] Among them, u t represents the value of the update gate, W u represents the weight matrix of the update gate, X t represents the input features at time t, h t and h t-1 Represent the hidden states of the current moment and the previous moment, respectively, b u represents the bias vector of the update gate; r t Represents the value of the reset gate, W r represents the weight matrix of the reset gate, b r represents the bias vector of the reset gate; c t represents the candidate hidden state, tanh and σ are activation functions, W c represents the weight matrix of the candidate hidden state, b c represents the bias vector of the candidate hidden state, f(A, X t ) represents the adjacency matrix A and input feature X t The convolution operation performed;

[0131] By processing spatial dependency and temporal dynamics through temporal graph convolution, the topological structure of the urban road network and the periodic characteristics of traffic lights on the traffic flow are captured to complete the task of traffic flow prediction;

[0132] S13: Filter and interpolate the speed prediction result to optimize the prediction result to the speed characteristic v of the vehicle running normally pre(t). This is because of the unequal step length of road segmentation and the time sampling accuracy. The speed prediction result is a step signal, which does not conform to the continuous signal characteristics of the normal vehicle speed. Therefore, filtering and interpolation processing are required.

[0133] S2: Obtain the phase timing information of the traffic lights, refer to the predicted traffic flow speed at the future moment, and plan the global guidance speed v in the cloud g (t), specifically comprising the following steps:

[0134] S21: Obtain the phase timing information of the traffic lights and use the improved intelligent driving model (Intelligent DriverModel) to solve the upper and lower limits s of the vehicle's driving trajectory through the entire road section in all green light windows upper 、s lower , its model is expressed as:

[0135]

[0136]

[0137] where a IDM is the acceleration of the model, Δv and Δx are the relative velocity and relative position, a max represents the maximum acceleration, v represents the vehicle speed, and v max represents the maximum vehicle speed, s(v, Δv) represents the expected safety distance, s0 represents the minimum safety distance, t hw represents the time interval. In this process, the red light is regarded as a prohibition of obstructing vehicles. According to this model, the maximum and minimum speed trajectories to reach the next intersection are calculated. If the light is green, the speed remains unchanged.

[0138] S22: Use dynamic programming algorithm to calculate the preliminary global guidance trajectory s of the vehicle dp (t k ) and vehicle speed v dp (t k ), and its cost function is:

[0139]

[0140] Among them, Represents the weight coefficient of this item, the first item represents the time efficiency cost, where t end is the total time, and the second term represents the reference speed cost, where v is the vehicle speed, v ref is the reference vehicle speed, the third term represents the comfort cost, where a is the ego vehicle acceleration, and the fourth term represents the trajectory constraint cost, where s is the ego vehicle trajectory, s upper 、s lower are the upper and lower limits of the reference trajectory. Where k = 0, 1, 2, ... ∈ [0, tend ] is each time step in the optimization domain;

[0141] S23: Based on the optimization result in step S22, use quadratic programming to smooth the trajectory and calculate the global guiding vehicle speed v g (t), its cost function is:

[0142]

[0143] Among them, ω1, ω2, ω3, and ω4 represent the weight coefficients of each item. The first item represents the position cost, the second item represents the speed cost, the third item represents the acceleration cost, and the fourth item represents the impact cost; s dp is the reference data after solving the dynamic programming algorithm, s dp ′ is the reference speed for dynamic programming solution, s dp ″ is the reference acceleration for dynamic programming solution, s dp ″′ is the reference impact degree of dynamic programming solution; s is the reference position of dynamic programming solution, s′ is the speed, s″ is the acceleration, and s″′ is the impact degree.

[0144] S3: Design an economical driving cruise controller based on model predictive control, referring to the global guide speed v g (t), real-time tracking of the front vehicle speed, solving the upper target speed v upper (t) and the total required power P of the vehicle d (t), specifically comprising the following steps:

[0145] S31: Establishing the vehicle longitudinal dynamics model:

[0146] m·a(t)=F t -F r

[0147]

[0148]

[0149] where F t (t) is the traction force, T w (t) is the wheel torque, r w is the tire radius, represents air resistance, ρ is air density, A is windward area, C d is the air resistance coefficient, mgC r Indicates rolling resistance, C r is the rolling resistance coefficient of the road, α is the road slope;

[0150] S32: Design of a cruise controller for economical driving based on model predictive control:

[0151] S321: Set the front vehicle speed to the predicted vehicle speed v pre (t);

[0152] S322: Design the state equation for economical cruising:

[0153]

[0154] The state variable x includes the distance d between the vehicle and the nearest traffic light. light , distance d from the vehicle in front f and the vehicle speed v, v f is the speed of the front vehicle, and the control input u is the wheel torque T w (t), where v f is the speed of the preceding vehicle, r w is the tire radius, F r is the vehicle resistance, m is the vehicle mass, t s is the time step;

[0155] S33: Design constraints, including the distance from the vehicle in front, the distance from the traffic light, the road speed limit and the maximum acceleration:

[0156] d min ≤v(t)·t hw +d0

[0157] 0≤d light (t), if light=red

[0158] 0≤d light (t)+φ(t), if light=yellow

[0159] v min ≤v(t)≤v max

[0160] a min ≤a(t)≤a max

[0161] Where φ(t) is a slack variable and is greater than 0, which restricts vehicles to comply with traffic light rules and cannot pass through traffic light intersections when the light is red or yellow. hw is the time distance of following the vehicle, d0 is the minimum distance of following the vehicle, d min Represents the current minimum following distance of the vehicle, v min 、v max 、a min 、a max Respectively represent the minimum and maximum speed and acceleration of the vehicle;

[0162] S34: Design the controller cost function, which represents the cost of reference speed, the cost of equivalent fuel consumption, the cost of control input, the cost of input change rate, and the cost of violating the soft constraint when the traffic light is yellow:

[0163]

[0164] where ω v ,ω c ,ω u ,ω du ,ω φ Represents the weight factor of each cost, P d (t) is the total power required by the vehicle, P d (t) = T w (t)·ω w (t),ω w is the wheel speed, for Δ∈(t,…,t+N P -1), where x(Δ|t) and u(Δ|t) are the measured state and input control quantity based on time t and the predicted state and input control quantity at time Δ, respectively. P It is the prediction time domain, and the optimal control problem is solved by rolling optimization;

[0165] S35: Solve the economical cruise driving control strategy in the cloud and convert the upper target vehicle speed v upper (t) and the total required power P of the vehicle d (t) is sent to the vehicle-side controller.

[0166] In this embodiment, step S4 performs lower-level vehicle-side optimization, that is, performs energy management. In this embodiment, a hybrid vehicle among new energy vehicles is used as a driving target. A simplified schematic diagram of a hybrid vehicle configuration is shown in FIG. Figure 4 As shown, it at least includes a fuel tank, an engine, a clutch, a power battery, a motor, a transmission, a differential and other power structural components. On this basis, the specific content of step S4 in this embodiment is as follows:

[0167] S4: Develop the lower-level energy management strategy, build the vehicle powertrain model, and calculate the optimal power distribution using the equivalent fuel minimum consumption strategy, which specifically includes the following steps:

[0168] S41: Establishing a hybrid vehicle power system model, which includes the following specific steps:

[0169] S411: Build engine model:

[0170]

[0171] P eng =ω eng ·Teng

[0172] in is the instantaneous fuel consumption rate, P eng is the engine power, η eng is the engine efficiency, which is determined by the engine torque T eng and speed ω eng Determine, Q is the lower heating value of the fuel;

[0173] S412: Establish motor model:

[0174]

[0175] Where P mot is the motor power, η mot (T mot ,ω mot ) is the motor efficiency, which is determined by the motor torque T mot and speed ω mot Decision, P elec is the power allocated to the motor. The motor power is greater than or equal to zero and less than zero, which are the motion and power generation modes respectively. elec Less than zero means the motor is operating in generating mode;

[0176] S413: Establishing battery model:

[0177]

[0178] Where P bat is the battery power;

[0179] S42: Build a hybrid vehicle power demand model:

[0180] P mot，out =T mot ·ω mot

[0181] P mot，total =P mot，out +P mot，loss

[0182] P eng，out =T eng ·ω eng

[0183] P eng，total =P eng，out +P eng，loss

[0184] P bat，total =P bat，out +P bat，loss

[0185] Among them, Pmot，out , T mot ,ω mot is the motor output power, torque and speed, P eng，out , T eng ,ω eng is the engine output power, torque, and speed, P mot，total , P mot，loss is the total power and power loss of the motor, P bat，total , P bat，out , P bat，loss is the total battery power, output power, and loss power. The total power required by the hybrid is the sum of the motor and engine power;

[0186] S43: Establish the cost function of the energy management problem of the equivalent fuel minimum consumption strategy:

[0187]

[0188]

[0189] The upper formula represents the cost function of the energy management strategy, and the lower formula expresses the total fuel consumption rate Equal to the engine fuel consumption rate Equivalent fuel consumption with battery The equivalent fuel consumption of the battery is λ is the equivalent factor, P bat is the battery power. Combining the above steps, the optimal power control strategy is solved on the vehicle side, which is also its optimal energy management power allocation strategy.

[0190] S44: Establish constraint boundaries:

[0191]

[0192]

[0193]

[0194] E bat (f)∈[SOC min , SOC max ]·V oc Q

[0195] E bat (0) = E bat (N)

[0196] The first one represents the constraint of motor output torque, the second one represents the constraint of engine output torque, the third one represents the constraint of battery output power, the fourth and fifth ones represent the constraint of SOC, and E bat is the battery energy, Voc is the open circuit voltage of the battery, Q represents the battery capacity, and the subscripts min and max represent the upper and lower limits of the parameter;

[0197] S44: Combining the cost function and the constraints in the above steps, the optimal power control strategy is solved on the vehicle side, which is also its optimal energy management power allocation strategy.

[0198] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. An economical driving method for an intelligent network-connected new energy vehicle under a traffic light scenario, characterized in that: It includes the following steps: S1: Construct a traffic flow prediction model based on a time graph convolutional network to predict the future traffic flow speed v by obtaining historical traffic flow speed information pre (t); S2: Obtain the phase timing information of the traffic lights, and refer to the predicted traffic flow speed v at the future moment pre (t), planning the global guidance speed v in the cloud g (t); S3: Construct an economical driving cruise controller based on model predictive control, referring to the global guidance speed v g (t), real-time tracking of the front vehicle speed, solving the upper target speed v upper (t) and the total required power P of the vehicle d (t); S4: Develop lower-level energy management strategies, build vehicle powertrain models, and calculate optimal power allocation using equivalent fuel minimum consumption strategies; In step S2, the following steps are included: S21: Obtain the phase timing information of the traffic lights and use the improved intelligent driving model to solve the upper and lower limits s of the vehicle's driving trajectory when passing through the entire road section in all green light windows upper 、s lower ; S22: Use dynamic programming algorithm to calculate the preliminary global guidance trajectory S of the vehicle dp (t k ) and vehicle speed v dp (t k ); S23: Based on the optimization result in step S22, use quadratic programming to smooth the trajectory and calculate the global guiding vehicle speed v g (t); In step S3, the following steps are specifically included: S31: Establishing the vehicle longitudinal dynamics model; S32: Constructing an economical driving cruise controller based on model predictive control; S33: Design constraints, including the distance constraint from the preceding vehicle, the distance constraint from the traffic light, the road speed limit constraint, and the maximum acceleration constraint; S34: Design the controller cost function, including reference speed cost, equivalent fuel consumption cost, control input cost, input change rate cost, and soft constraint cost when the traffic light is yellow; S35: Solve the economical cruise driving control strategy in the cloud and convert the upper target vehicle speed v upper (t) and the total required power P of the vehicle d (t) sent to the vehicle-side controller; In step S4, the following steps are specifically included: S41: Establish a hybrid vehicle power system model, including engine model, motor model and battery model; S42: Establish a hybrid vehicle power demand model; S43: Establishing the cost function of the energy management problem of the equivalent fuel minimum consumption strategy; S44: Combining the cost function and the constraints in the above steps, the optimal power control strategy is solved on the vehicle side, which is also its optimal energy management power allocation strategy.

2. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 1, characterized in that: In step S1, the following steps are included: S11: Construct a traffic flow prediction model based on a temporal convolutional network, which at least includes an input end taking the network topology as input, a two-layer graph convolutional network GCN model for extracting temporal features, and a gated recurrent unit GRU model for extracting spatial features; S12: Using historical traffic speed information with road network topology information as network input, the speed prediction results for the current time and future time are obtained through the graph convolutional network GCN model and the gated recurrent unit GRU model, where: The unweighted graph G = (V, E) is used to describe the road network topology, where V = {v1, v2, ..., v N }, V represents the road node set, N represents the number of nodes, and E represents the edge set, thereby establishing an adjacency matrix A representing the connections between roads; Get traffic flow speed information as feature matrix X N×P The attribute features in are represented by X∈R N×P , P represents the number of node attribute features, that is, the length of the historical time series; The GCN model is used to learn spatial features from traffic data. The two-layer GCN model is expressed as: Where X and A represent the feature matrix and adjacency matrix respectively. represents the preprocessing step, is the self-connected structure matrix, is an angle matrix, σ(·) represents xxx; W0 and W1 represent the weight matrices of the first and second layers respectively; f(A, X t ) represents the graph convolution process, W and b represent the weight and bias in the training process respectively; Select the GRU model to capture the temporal dependency of traffic data and use the update gate u t Control the degree to which the previous state information affects the current state, using the reset gate r t Control the degree of ignoring the state information of the previous moment, using the previous moment h t-1 The hidden state and current traffic information are used as input to obtain the current traffic status. The time graph convolution network built by combining graph convolution and gated recurrent unit has the following specific calculation process: u t =σ(W u [f(A,X t ),h t-1 ]+b u ) r t =σ(W r [f(A,X t ),h t-1 ]+b r ) c t =tanh(W c [f(A,X t ),(r t ·h t-1 )]+b c ) h t =u t ·h t-1 +(1-u t )·c t Among them, u t represents the value of the update gate, W u represents the weight matrix of the update gate, X t represents the input features at time t, h t and h t-1 Represent the hidden states of the current moment and the previous moment, respectively, b u represents the bias vector of the update gate; r t Represents the value of the reset gate, W r represents the weight matrix of the reset gate, b r represents the bias vector of the reset gate; c t represents the candidate hidden state, tanh and σ are activation functions, W c represents the weight matrix of the candidate hidden state, b c represents the bias vector of the candidate hidden state, f(A, X t ) represents the adjacency matrix A and input feature X t The convolution operation performed; By processing spatial dependency and temporal dynamics through temporal graph convolution, the topological structure of the urban road network and the periodic characteristics of traffic lights on the traffic flow are captured to complete the task of traffic flow prediction; S13: Filter and interpolate the speed prediction result to optimize the prediction result to the speed characteristic v of the vehicle running normally pre (t).

3. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 1, characterized in that: In step S21, the improved intelligent driving model solves the upper and lower limits s of the driving trajectory according to the signal light phase timing information. upper 、s lower In the process, the model is expressed as: where a IDM is the acceleration of the model, Δv and Δx are the relative velocity and relative position, a max represents the maximum acceleration, v represents the vehicle speed, and v max represents the maximum vehicle speed, s(v,Δv) represents the expected safety distance, s0 represents the minimum safety distance, and t hw represents the time interval. In this process, the red light is regarded as a prohibition of obstructing vehicles. According to this model, the maximum and minimum speed trajectories to reach the next intersection are calculated. If the light is green, the speed remains unchanged. In step S22, a dynamic programming algorithm is used to calculate the preliminary global guidance trajectory s of the vehicle. dp (t k ) and vehicle speed v dp (t k ) is: Among them, Represents the weight coefficient of each item, the first item represents the time efficiency cost, where t end is the total time; the second term represents the reference speed cost, where v is the vehicle speed, v ref is the reference vehicle speed; the third term represents the comfort cost, where a is the ego vehicle acceleration; the fourth term represents the trajectory constraint cost, where s is the ego vehicle trajectory, s upper 、s lower are the upper and lower limits of the reference trajectory; k = 0, 1, 2, ... ∈ [0, t end ] is each time step in the optimization domain; In step S23, the global guiding vehicle speed v is calculated using the quadratic programming smooth trajectory g The cost function of (t) is: Among them, ω1, ω2, ω3, and ω4 represent the weight coefficients of each item. The first item represents the position cost, the second item represents the speed cost, the third item represents the acceleration cost, and the fourth item represents the impact cost. s dp is the reference data after solving the dynamic programming algorithm, s dp ′ is the reference speed for dynamic programming solution, s dp ″′ is the reference acceleration for dynamic programming solution, s dp ″′ is the reference impact degree of dynamic programming solution; s is the reference position of dynamic programming solution, s′ is the speed, s″ is the acceleration, and s″′ is the impact degree.

4. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 1, characterized in that: In step S31, the vehicle longitudinal dynamics model is: m·a(t)=F t -F r Among them, F t (t) is the traction force, T w (t) is the wheel torque, r w is the tire radius, represents air resistance, ρ is air density, A is windward area, C d is the air resistance coefficient, mgC r Indicates rolling resistance, C r is the rolling resistance coefficient of the road, α is the road slope; In step S32, the state equation of economical cruise driving is: The state variable x includes the distance d between the vehicle and the nearest traffic light. light , distance d from the vehicle in front f and the vehicle speed v, v f is the speed of the front vehicle, and the control input u is the wheel torque T w (t), where v f is the speed of the preceding vehicle and the speed of the preceding vehicle is set to the predicted speed v pre (t), r w is the tire radius, F r is the vehicle resistance, m is the vehicle mass, t s is the time step; In step S33, the distance constraint to the preceding vehicle, the distance constraint to the traffic light, the road speed limit constraint, and the maximum acceleration constraint are respectively expressed as: d min ≤v(t)·t hw +d0 0≤d light (t),if light=red 0≤d light (t)+φ(t),iflight=yellow v min ≤v(t)≤v max a min ≤a(t)≤a max Where φ(t) is a slack variable and is greater than 0, which restricts vehicles to comply with traffic light rules and cannot pass through traffic light intersections when the light is red or yellow. hw is the time distance of following the vehicle, d0 is the minimum distance of following the vehicle, d min Represents the current minimum following distance of the vehicle, v min 、v max 、a min 、a max Respectively represent the minimum and maximum speed and acceleration of the vehicle; In step S4, the controller cost function is expressed as: Each term in the function represents the reference speed cost, equivalent fuel consumption cost, control input cost, input change rate cost, and soft constraint cost when the traffic light is yellow. v ,ω c ,ω u ,ω du ,ω φ Represents the weight factor of each cost, P d (t) is the total power required by the vehicle, P d (t) = T w (t)·ω w (t),ω w is the wheel speed, for Δ∈(t,...,t+N P -1), where x(Δ|t) and u(Δ|t) are the measured state and input control quantity based on time t and the predicted state and input control quantity at time Δ, respectively. P is the prediction time domain, and the optimal control problem is solved by rolling optimization.

5. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 1, characterized in that: In step S41, the engine model is: P eng =ω eng ·T eng in is the instantaneous fuel consumption rate, P eng is the engine power, n eng is the engine efficiency, which is determined by the engine torque T eng and speed ω eng Determine, Q is the lower heating value of the fuel; The motor model is: Where P mot is the motor power, η mot (T mot ,ω mot ) is the motor efficiency, which is determined by the motor torque T mot and speed ω mot Decision, P elec is the power allocated to the motor. The motor power is greater than or equal to zero and less than zero in motion and power generation modes respectively. elec Less than zero means the motor is operating in generating mode; The battery model is: Where P bat is the battery power; In step S42, the power demand model of the hybrid vehicle is: P mot,out =T mot ·oh mot P mot,total =P mot,out +P mot,loss P eng,out =T eng ·oh eng P eng,toal =P eng,out +P eng,loss P bat,total =P bat,out +P bat,loss Among them, P mot，out , T mot ,ω mot is the motor output power, torque and speed, P eng，out ,t eng ,ω eng is the engine output power, torque, and speed, P mot，total , P mot，loss is the total power and power loss of the motor, P bat，total , P bat，out , P bat，loss is the total battery power, output power, loss power, and the total power required by the hybrid is the sum of the motor and engine power.

6. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 5, characterized in that: In step S43, the cost function of the energy management problem of the equivalent fuel minimum consumption strategy is expressed as: The upper formula represents the cost function of the energy management strategy, and the lower formula expresses the total fuel consumption rate Equal to the engine fuel consumption rate Equivalent fuel consumption with battery The equivalent fuel consumption of the battery is λ is the equivalent factor, P bat is the battery power; solving the optimal power control strategy on the vehicle side is also its optimal energy management power allocation strategy.

7. The economical driving method of an intelligent network-connected new energy vehicle in a traffic light scenario according to claim 6, characterized in that: In step S44, the constraint boundary is expressed as: From bat (t)∈[SOC min ,SOC max ]·V oc ·Q AND bat (0)=And bat (N) Among them, the first one represents the constraint of motor output torque, the second one represents the constraint of engine output torque, the third one represents the constraint of battery output power, and the fourth and fifth ones represent the constraint of SOC; E bat is the battery energy, V oc is the open circuit voltage of the battery, Q represents the battery capacity, and the subscripts min and max represent the upper and lower limits of the parameter.

Images