An adaptive energy-saving speed planning method for following a vehicle

Abstract

The application discloses a kind of self-adapting energy-saving speed planning method of following car, comprising: vehicle information acquisition and processing: through vehicle-mounted sensor and vehicle networking equipment, front car state and road information are acquired, equidistant sampling is carried out on spatial domain, and slope prediction information is acquired;Front car speed prediction: short-term prediction is carried out to front car speed using neural network;Energy-saving following car speed optimization problem establishment: select vehicle nominal kinetic energy, passing time, acceleration as optimization variable, with the lowest energy consumption as target, consider comfort and safety, design model predictive control;Optimization problem solution and online speed control: determine prediction domain, call commercial IPOPT solver to calculate energy-saving following car speed optimization problem, use the first acceleration signal in prediction domain to control speed, and roll forward optimization and control.The application can reduce the energy consumption of following car cruise in slope scene, ensure the timeliness and practicability of following car speed planning.

Description

Technical Field

[0001] This invention belongs to the field of automotive adaptive cruise control technology and relates to an adaptive energy-saving following speed planning method. Background Technology

[0002] With the application of vehicle-mounted sensors and the development of vehicle-road-network technology, ground vehicles can acquire information about the status of preceding vehicles and roads. This information is used to design predictive motion control, enabling advanced driver assistance functions and even high-level autonomous driving technologies. Among these, speed planning that considers following safety is a prerequisite for adaptive cruise control, and its control effect directly impacts the safety, comfort, and economy of following the vehicle. Therefore, speed planning methods that integrate information about preceding vehicles and roads are of great significance for achieving safe, energy-efficient, and effective assisted cruise control.

[0003] Currently, vehicle speed planning technology has been extensively studied. Chinese patent "Speed ​​Control Method, Device, Unmanned Driving Machine and Electronic Equipment for Unmanned Driving" (CN202310460815.7) provides a speed control method for unmanned driving. However, this invention ignores road gradient and the state of the vehicle in front, failing to achieve energy-efficient driving in following scenarios. Furthermore, traditional following speed planning methods assume the preceding vehicle's constant speed as an ideal condition, neglecting its dynamic characteristics. In addition, existing speed planning methods represent road gradient as a polynomial function of distance in the time domain, resulting in high online solution costs for the optimization problem. Therefore, existing following speed planning methods cannot guarantee timeliness and good overall performance, urgently requiring improvement. Summary of the Invention

[0004] The purpose of this invention is to address the performance shortcomings of current vehicle following speed planning methods by providing an adaptive and energy-saving vehicle following speed planning method, reducing energy consumption during vehicle following in gradient scenarios, and ensuring the timeliness and practicality of the vehicle following speed planning method.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] An adaptive energy-saving vehicle following speed planning method includes the following steps:

[0007] Step 1, Vehicle Information Acquisition and Processing: Acquire the status of the vehicle in front and road information through onboard sensors and vehicle networking equipment, and obtain slope prediction information by equidistant sampling in the spatial domain;

[0008] Step 2, Predicting the speed of the vehicle in front: Using a neural network to make short-term predictions of the speed of the vehicle in front;

[0009] Step 3: Establishing the energy-saving following speed optimization problem: Select the vehicle's nominal kinetic energy, travel time, and acceleration as optimization variables, with the goal of minimizing driving energy consumption, and considering comfort and safety, design a model predictive control;

[0010] Step 4: Optimization Problem Solving and Online Vehicle Speed ​​Control: Determine the prediction domain, call the commercial IPOPT solver to calculate the energy-saving following speed optimization problem, use the first acceleration signal in the prediction domain to control the vehicle speed, and perform rolling forward optimization and control.

[0011] Specifically, step 1 includes:

[0012] Step 1.1. Obtain the status of the vehicle in front: Based on the onboard sensors, obtain the speed of the vehicle in front and the distance between the two vehicles;

[0013] Step 1.2. Obtain vehicle location: The vehicle uses a GPS receiver to receive satellite signals to determine its current location's longitude and latitude coordinates; it also uses maps and geographic information systems to obtain road network, terrain, and elevation information.

[0014] Step 1.3. Offline Data Acquisition: Vehicles can obtain speed limit information for roads ahead by integrating navigation system data with map data;

[0015] Step 1.4. Distance Interval Selection: Discretize the road ahead with an equal interval Δs of 10m, and use linear interpolation to obtain the elevation data at each point.

[0016] Specifically, step 2 includes:

[0017] Step 2.1. Neural Network Training: Collect offline vehicle speed data, construct a neural network using Long Short-Term Memory (LSTM), and perform offline training of the neural network;

[0018] Step 2.2. Online vehicle speed prediction: Based on the speed and acceleration of the vehicle in front, an LSTM neural network is used to predict the vehicle speed within the next 15 seconds. In this embodiment, the average root mean square error of the vehicle speed prediction is 0.26 m / s.

[0019] Further, step 2.1, neural network training, includes:

[0020] Step 2.1.1. Collect offline vehicle speed data: Obtain vehicle speed information at different times on the corresponding roads through the intelligent transportation system, select a certain scale, and generate offline vehicle speed data;

[0021] Step 2.1.2. Neural Network Construction: LSTM is selected as the specific neural network for velocity prediction; the time series variables velocity and acceleration are selected as the input layer, and the expected velocity sequence is used as the output layer; the sequence length of the input layer is set to 30, the sequence length of the output layer is set to 15, and the number of hidden layers is set to 200.

[0022] Step 2.1.3. Offline training of neural network: Group the offline vehicle speed data, train the LSTM network, and predict the speed in the next 15 seconds based on the speed and acceleration information in the past 30 seconds.

[0023] Specifically, step 3 includes:

[0024] Step 3.1. Establish the state transition equation: At the current system time k, for road segment i, select the vehicle's nominal kinetic energy. Travel time t k (i) As the state variable, the vehicle acceleration a(i) is selected as the control variable; for adjacent road segments i+1, let the vehicle's kinetic energy, travel time, and acceleration action be respectively... t k (i+1) and a(i+1); using vehicle longitudinal dynamics, establish the state transition equation between the vehicle's nominal kinetic energy, travel time, and acceleration:

[0025]

[0026] Among them, nominal kinetic energy Defined as v(i) 2 / 2;

[0027] Step 3.2. Establish system variable constraint equations: The vehicle's kinetic energy cannot exceed the upper limit corresponding to the road speed limit; to ensure comfort, the vehicle's acceleration must be kept within a certain range, and the acceleration derivative impact Jerk(i) must also be kept within a certain range; to ensure vehicle traffic efficiency, a minimum road speed is set; establish the following system variable constraint equations, where the underscores and overscores of the variables represent the lower and upper limits of the variables:

[0028]

[0029] Step 3.3. Establish following safety constraint equations: The vehicle needs to meet following safety requirements to avoid rear-end collisions; define the travel time t for the vehicle and the vehicle in front through road segment i as follows. k (i) and The following vehicle safety constraint equation is:

[0030]

[0031] in For safe headway;

[0032] Step 3.4. Design Optimization Objective: Use vehicle driving energy consumption as the optimization objective; for gasoline vehicles, let m, g, f, C D A f ρ and ρ represent the vehicle mass, gravitational acceleration, rolling resistance coefficient, air resistance coefficient, frontal area, and air density, respectively; vehicle power requirement. It can be expressed as a function of system variables and vehicle parameters:

[0033]

[0034] At the same time, fuel consumption rate This can be expressed as the vehicle's power requirement. The polynomial function is shown below.

[0035]

[0036] Among them, β2, β1, and β0 represent energy consumption characteristic parameters, which are determined by the specific vehicle model;

[0037] Fuel consumption rate Multiply by the equivalent travel time The fuel consumption on road segment i can be obtained; therefore, the optimization objective is the sum of fuel consumption within the prediction domain:

[0038]

[0039] Step 3.5. Constructing the spatial domain model predictive control problem: Combining formulas (1)-(6), with the goal of minimizing energy consumption, construct the model predictive control problem in the spatial domain:

[0040]

[0041] st

[0042]

[0043] Specifically, step 4 includes:

[0044] Step 4.1. Solving the energy-saving following speed optimization problem: Taking into account both optimization performance and solution timeliness, the prediction domain size of the model predictive control is determined, and the mature commercial solver IPOPT is used to solve the speed optimization problem;

[0045] Step 4.2. Forward Rolling Optimization Control: Update vehicle position and prediction information, repeat the energy-saving following speed optimization problem and online vehicle speed control.

[0046] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0047] 1. This invention maintains good comfort, high energy efficiency, and strong optimization performance while ensuring following safety: In real-world adaptive cruise control, a fixed headway is typically used, neglecting terrain changes and the status of the vehicle ahead. This leads to drastic changes in vehicle load due to terrain variations and susceptibility to the preceding vehicle's condition, exacerbating energy consumption. This invention, by acquiring road information ahead and predicting the speed of the vehicle ahead, aims to minimize cruise energy consumption by establishing a spatial domain model predictive control problem to plan vehicle speed. Compared with intelligent driver following methods, the adaptive energy-saving following speed planning method effectively predicts the speed of the vehicle ahead, achieving control performance that meets both safety and comfort requirements.

[0048] 2. High online computation efficiency: This invention adopts a spatial domain modeling method, which avoids the polynomial fitting of road slope by traditional time domain methods, shortens the computation time of vehicle speed planning, and reduces the solution complexity of the rolling optimization problem.

[0049] 3. Experimental verification shows that this invention has the following advantages compared with the prior art:

[0050] 1) Under the adaptive energy-saving following speed planning method of the present invention, the vehicle speed is optimized according to the altitude, showing a trend of deceleration before going uphill, deceleration during going uphill, and acceleration during going downhill. However, under the intelligent driver following model method, the vehicle follows the speed of the vehicle in front, maintaining a stable headway, and cannot adjust to the terrain.

[0051] 2) Under the adaptive energy-saving following speed planning method of the present invention, the acceleration of the vehicle is maintained at [-2,2] m / s². 2 The magnitude of the change is similar to that of the intelligent driver following model method, maintaining driving comfort.

[0052] 3) Under the adaptive energy-saving following speed planning method of the present invention, the headway between the vehicle and the vehicle in front is always maintained above the safety threshold, ensuring following safety.

[0053] 4) The total fuel consumption of the intelligent driver following model method is 441.15g, while the adaptive energy-saving following speed planning method of this invention reduces the total fuel consumption to 402.32g. In experimental verification of following scenarios, the adaptive energy-saving following cruise strategy can save up to 8.80% of fuel.

[0054] 5) The computation time of the vehicle speed planning method affects the online implementation of optimized control: The average computation time required for each update of the adaptive energy-saving following speed planning method of this invention is 25.7ms; for vehicle cruise systems, the control sampling frequency generally does not exceed 5Hz, that is, the shortest update interval is 0.2s. Therefore, the method of this invention has good potential for online application. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating an adaptive energy-saving following speed planning method according to the present invention.

[0056] Figure 2 This is a flowchart illustrating the prediction of the speed of the preceding vehicle in an adaptive energy-saving following speed planning method of the present invention.

[0057] Figure 3 This is a schematic diagram illustrating the process of establishing the energy-saving following speed optimization problem in an adaptive energy-saving following speed planning method of the present invention.

[0058] Figure 4 This is a diagram of the speed data of the preceding vehicle in a simulation test of an adaptive energy-saving following speed planning method of the present invention.

[0059] Figure 5 This is a speed result diagram of an adaptive energy-saving following speed planning method of the present invention.

[0060] Figure 6 This is a speed result graph of the existing intelligent driver following model method.

[0061] Figure 7 This is an acceleration result diagram of an adaptive energy-saving vehicle following speed planning method and an intelligent driver following model method according to the present invention.

[0062] Figure 8 This is a time-distance diagram of the vehicle headway in an adaptive energy-saving following speed planning method of the present invention. Detailed Implementation

[0063] The present invention will now be described in further detail with reference to the accompanying drawings.

[0064] Figure 1 The flowchart of an adaptive energy-saving following speed planning method of the present invention is shown.

[0065] In this embodiment, as Figure 1 As shown, an adaptive energy-saving following speed planning method includes:

[0066] Step 1: Vehicle Information Acquisition and Processing: Acquire the status of the vehicle in front and road information through onboard sensors and vehicle networking equipment, and obtain slope prediction information by equidistant sampling in the spatial domain.

[0067] Step 1, the vehicle information acquisition and processing steps, includes:

[0068] Step 1.1: Obtain the status of the vehicle in front: Based on the onboard sensors, obtain the speed of the vehicle in front and the distance between the two vehicles.

[0069] Step 1.2: Obtain Vehicle Location: The vehicle uses a GPS receiver to receive satellite signals to determine its current location's longitude and latitude coordinates. Maps and Geographic Information Systems (GIS) are then used to obtain road network, terrain, and elevation information.

[0070] Step 1.3: Offline data acquisition: Vehicles can obtain speed limit information for the road ahead by integrating navigation system with map data.

[0071] Step 1.4: Distance interval selection: Discretize the road ahead with an equal interval Δs of 10m, and use linear interpolation to obtain the elevation data at each point.

[0072] Step 2: Predicting the speed of the vehicle in front: Using a neural network to make short-term predictions of the speed of the vehicle in front;

[0073] Figure 2 The flowchart illustrates the process of predicting the speed of the vehicle in front in an adaptive energy-saving following speed planning method according to the present invention. For example... Figure 2 As shown, step 2, the forward vehicle speed prediction step, includes:

[0074] Step 2.1: Neural Network Training: Collect offline vehicle speed data, construct a neural network using Long Short-Term Memory (LSTM), and perform offline training. Select time-series variables speed and acceleration as input layers, and use the expected speed sequence as the output layer. The input layer sequence length is 30, the output layer sequence length is 15, and the number of hidden layers is set to 200.

[0075] Step 2.2: Online vehicle speed prediction: Based on the speed and acceleration of the vehicle in front, an LSTM neural network is used to predict the vehicle speed within the next 15 seconds. In this embodiment, the average root mean square error of the vehicle speed prediction is 0.26 m / s.

[0076] Step 2.1, the neural network training step, includes:

[0077] Step 2.1.1: Collect offline vehicle speed data: Obtain vehicle speed information at different times on the corresponding roads through the intelligent transportation system, select a certain scale, and generate offline vehicle speed data.

[0078] Step 2.1.2: Neural Network Construction: LSTM is selected as the specific neural network for velocity prediction. The time-series variables velocity and acceleration are chosen as the input layer, and the expected velocity sequence is used as the output layer. The input sequence length is set to 30, the output sequence length to 15, and the number of hidden layers to 200.

[0079] Step 2.1.3: Offline training of neural network: Group the offline vehicle speed data, train the LSTM network, and predict the speed in the next 15 seconds based on the speed and acceleration information in the past 30 seconds.

[0080] Step 3: Establishing the energy-saving following speed optimization problem: Select the vehicle's nominal kinetic energy, travel time, and acceleration as optimization variables, with the goal of minimizing driving energy consumption, and considering comfort and safety, design a model predictive control;

[0081] like Figure 3 As shown, step 3, the steps for establishing the energy-saving following speed optimization problem, include:

[0082] Step 3.1: Establish the state transition equation: At the current system time k, for road segment i, select the vehicle's nominal kinetic energy. Travel time t k (i) is selected as the state variable, and the vehicle acceleration a(i) is chosen as the control variable. For adjacent road segments i+1, let the vehicle's kinetic energy, travel time, and acceleration action be respectively... t k (i+1) and a(i+1). Using the longitudinal dynamics of the vehicle, the state transition equations between the vehicle's nominal kinetic energy, travel time, and acceleration are established as follows.

[0083]

[0084] Among them, nominal kinetic energy Defined as v(i) 2 / 2.

[0085] Step 3.2: Establish system variable constraints: Vehicle kinetic energy cannot exceed the upper limit corresponding to the road speed limit. To ensure comfort, vehicle acceleration must be kept within a certain range, and the acceleration derivative impact Jerk(i) must also be kept within a certain range. To ensure vehicle traffic efficiency, a minimum road speed is set. The following system variable constraint equations are established, where the underscores and overscores of the variables represent the lower and upper limits of the variables.

[0086]

[0087] Step 3.3: Establish following safety constraints: The vehicle must meet following safety requirements to avoid rear-end collisions. Define the travel time t for the vehicle and the vehicle in front of it through road segment i as follows. k (i) and The following are the following safety constraints for following vehicles.

[0088]

[0089] in For safe headway.

[0090] Step 3.4: Design Optimization Objective: Use vehicle driving energy consumption as the optimization objective. For gasoline vehicles, let m, g, f, and C... D Af Let ρ be the vehicle mass, gravitational acceleration, rolling drag coefficient, air drag coefficient, frontal area, and air density, respectively. Vehicle power requirement. It can be represented as a function of system variables and vehicle parameters, as shown below:

[0091]

[0092] At the same time, fuel consumption rate This can be expressed as the vehicle's power requirement. The polynomial function is shown below.

[0093]

[0094] Among them, β2, β1, and β0 represent energy consumption characteristic parameters, which are determined by the specific vehicle model.

[0095] Fuel consumption rate Multiply by the equivalent travel time The fuel consumption on road segment i can be obtained. Therefore, the optimization objective is the sum of fuel consumption within the prediction domain, as shown below.

[0096]

[0097] Step 3.5: Construct the spatial domain model predictive control problem: Combine formulas (1)-(6) to construct the model predictive control problem in the spatial domain with the goal of minimizing energy consumption, as shown below.

[0098]

[0099] st

[0100]

[0101] Step 4: Solving the energy-saving following speed optimization problem and online vehicle speed control: Determine the prediction domain, call the commercial IPOPT solver to calculate the energy-saving following speed optimization problem (see formula (7)), use the first acceleration signal in the prediction domain to control the vehicle speed, and perform rolling forward optimization and control.

[0102] Step 4, the solution to the energy-saving following speed optimization problem and the online vehicle speed control steps, include:

[0103] Step 4.1: Solving the speed optimization problem for energy-saving vehicle following: Taking into account both optimization performance and solution timeliness, the prediction domain size of the model predictive control is determined, and the mature commercial solver IPOPT is used to solve the speed optimization problem.

[0104] Step 4.2: Forward Rolling Optimization Control: Update vehicle position and prediction information, repeat the energy-saving following speed optimization problem and online vehicle speed control.

[0105] Figure 4 The embodiment of the present invention shows the speed information of the vehicle in front and the speed limit information along the route, which were collected in a real-world cruising scenario.

[0106] To further verify the advantages of the present invention, an intelligent driver following model was used to simulate human driving behavior, and the results were compared with the adaptive energy-saving following speed planning method of the present invention.

[0107] Figure 5 The speed results of the adaptive energy-saving following speed planning method are shown in the figure.

[0108] Figure 6 The speed results of the intelligent driver following model method are shown in the figure.

[0109] Figure 7 The acceleration results of the intelligent driver following model method and the adaptive energy-saving following speed planning method are shown in the figure.

[0110] Figure 8 The headway diagram of the adaptive energy-saving following speed planning method is shown.

[0111] like Figure 5 As shown, under the adaptive energy-saving following speed planning method of the present invention, the vehicle speed is optimized with altitude, showing a trend of deceleration before going uphill, deceleration during going uphill, and acceleration during going downhill.

[0112] like Figure 6 As shown, under the intelligent driver following model method, the vehicle follows the speed of the vehicle in front and maintains a stable headway, but fails to adapt to the terrain.

[0113] like Figure 7 As shown, under the adaptive energy-saving following speed planning method of the present invention, the acceleration of the vehicle is maintained in the range of [-2, 2] m / s². 2 The magnitude of the change is similar to that of the intelligent driver model method, maintaining driving comfort.

[0114] like Figure 8 As shown, under the adaptive energy-saving following speed planning method of the present invention, the headway between the vehicle and the vehicle in front is always maintained above the safety threshold, ensuring following safety.

[0115] Table 1 is a comparison table of the results of the adaptive energy-saving following speed planning method and the intelligent driver following model method of the present invention.

[0116] Table 1

[0117]

[0118] Table 1 shows a comparison of the results of different following cruise methods. The total fuel consumption of the intelligent driver model method is 441.15g, while the adaptive energy-saving following speed planning method reduces the total fuel consumption to 402.32g. In the following scenario of this embodiment, the energy-saving following cruise strategy saves 8.80% of fuel.

[0119] The computation time of the vehicle speed planning method is crucial for the online implementation of optimized control. The adaptive energy-saving following speed planning method of this invention requires an average computation time of 25.7 ms for each update. For vehicle cruise systems, the control sampling frequency generally does not exceed 5 Hz, meaning the shortest update interval is 0.2 s. Therefore, the method of this invention has good potential for online application.

Claims

1. An adaptive energy-saving vehicle following speed planning method, characterized in that, Includes the following steps: Step 1, Vehicle Information Acquisition and Processing: Acquire the status of the vehicle in front and road information through onboard sensors and vehicle networking equipment, and obtain slope prediction information by equidistant sampling in the spatial domain; Step 2, Predicting the speed of the vehicle in front: Using a neural network to make short-term predictions of the speed of the vehicle in front; Step 3: Establishing the energy-saving following speed optimization problem: Select the vehicle's nominal kinetic energy, travel time, and acceleration as optimization variables, with the goal of minimizing driving energy consumption, and considering comfort and safety, design a model predictive control; Step 4, Optimization Problem Solving and Online Vehicle Speed ​​Control: Determine the prediction domain, call the commercial IPOPT solver to calculate the energy-saving following speed optimization problem, use the first acceleration signal in the prediction domain to control the vehicle speed, and perform rolling forward optimization and control; Step 1 includes: Step 1.

1. Obtain the status of the vehicle in front: Based on the vehicle's onboard sensors, obtain the speed of the vehicle in front and the distance between the two vehicles; Step 1.

2. Obtain vehicle location: The vehicle uses a GPS receiver to receive satellite signals to determine its current location's longitude and latitude coordinates; it also uses maps and geographic information systems to obtain road network, terrain, and elevation information. Step 1.

3. Offline Data Acquisition: Vehicles can obtain speed limit information for roads ahead by integrating navigation system data with map data; Step 1.

4. Distance Interval Selection: Using equal intervals... For a 10m discretized road ahead, use linear interpolation to obtain the elevation data at each point; Step 2 includes: Step 2.

1. Neural Network Training: Collect offline vehicle speed data, construct a neural network using Long Short-Term Memory (LSTM), and perform offline training of the neural network; Step 2.

2. Online vehicle speed prediction: Based on the speed and acceleration of the vehicle in front, an LSTM neural network is used to predict the vehicle speed within the next 15 seconds. The average root mean square error of the vehicle speed prediction is 0.26 m / s. Step 2.1, neural network training, includes: Step 2.1.

1. Collect offline vehicle speed data: Obtain vehicle speed information at different times on the corresponding roads through the intelligent transportation system, select a certain scale, and generate offline vehicle speed data; Step 2.1.

2. Neural Network Construction: LSTM is selected as the specific neural network for velocity prediction; the time series variables velocity and acceleration are selected as the input layer, and the expected velocity sequence is used as the output layer; the sequence length of the input layer is set to 30, the sequence length of the output layer is set to 15, and the number of hidden layers is set to 200. Step 2.1.

3. Offline training of neural network: Group the offline vehicle speed data, train the LSTM network, and predict the speed in the next 15 seconds based on the speed and acceleration information in the past 30 seconds; Step 3 includes: Step 3.

1. Establish the state transition equation: at the current system time... Below, for road segments Select vehicle nominal kinetic energy Travel time As a state variable, vehicle acceleration is selected. To control the quantity; for adjacent road segments Let the vehicle's kinetic energy, travel time, and acceleration be respectively... , and Using vehicle longitudinal dynamics, establish the state transition equations between the vehicle's nominal kinetic energy, travel time, and acceleration: （1） Among them, nominal kinetic energy Defined as ; Step 3.

2. Establish system variable constraint equations: the vehicle's kinetic energy cannot exceed the upper limit corresponding to the road speed limit; to ensure comfort, the vehicle's acceleration must be kept within a certain range, and the acceleration derivative impact degree... It also needs to be kept within a certain range; to ensure vehicle traffic efficiency, a minimum road speed is set; the following system variable constraint equations are established, where the underscores and overscores of the variables represent the lower and upper limits of the variables: （2） Step 3.

3. Establish following safety constraint equations: The vehicle needs to meet following safety requirements to avoid rear-end collisions; define the road segments traversed by the vehicle and the vehicle in front. The passage times are respectively and Then the following vehicle safety constraint equation is: （3） in For safe headway; Step 3.

4. Design Optimization Objective: Use vehicle driving energy consumption as the optimization objective; for gasoline vehicles, let... , , , , and These are, respectively, vehicle mass, gravitational acceleration, rolling drag coefficient, air drag coefficient, frontal area, and air density; vehicle power requirement. It can be expressed as a function of system variables and vehicle parameters: （4） At the same time, fuel consumption rate This can be expressed as the vehicle's power requirement. The polynomial function is shown below. （5） in, , , The characteristic parameters representing energy consumption are determined by the specific vehicle model; Fuel consumption rate Multiply by the equivalent travel time Road fragments can be obtained The goal is to predict the total fuel consumption over the specified range; therefore, the optimization objective is to predict the total fuel consumption within the prediction domain. （6） Step 3.

5. Constructing the spatial domain model predictive control problem: Combining formulas (1)-(6), with the goal of minimizing energy consumption, construct the model predictive control problem in the spatial domain: (7)。 2. The adaptive energy-saving following speed planning method according to claim 1, characterized in that, Step 4 includes: Step 4.

1. Solving the speed optimization problem for energy-saving vehicle following: Taking into account both optimization performance and solution timeliness, the prediction domain size of the model predictive control is determined, and the mature commercial solver IPOPT is used to solve the speed optimization problem; Step 4.

2. Forward Rolling Optimization Control: Update vehicle position and prediction information, repeat the energy-saving following speed optimization problem and online vehicle speed control.

Images