CN-121989947-A - Driving intention recognition improvement method, device and equipment based on fuel-saving cruise control of heavy truck and storage medium

CN121989947ACN 121989947 ACN121989947 ACN 121989947ACN-121989947-A

Abstract

The invention discloses a driving intention recognition improvement method, device, equipment and storage medium based on fuel-saving cruise control of a heavy truck, relates to the technical field of fuel-saving control of vehicles, and is used for solving the problems of low fuel efficiency, inaccurate interaction intention recognition and high calculation resource consumption in the prior art. The method comprises the steps of constructing a prediction model fused by a dynamic graph neural network and an LSTM, obtaining interaction information of surrounding vehicles and extracting potential driving intention characteristics, realizing balance control between fuel saving and safety through a multi-level mode switching mechanism on the basis of a prediction result, introducing an event-triggered attention allocation method, dividing a prediction area to improve calculation efficiency, combining real-time road network data, and completing solving of a fuel consumption optimization control strategy based on a quadratic programming method. The method can improve the prediction accuracy of the vehicle driving intention, can reduce the fuel consumption per kilometer compared with the existing self-adaptive cruise control system, and has good energy conservation, stability and practical application value.

Inventors

HE SHUILONG
CHEN CAIWEI
ZHOU FU
YE HONGLIN
TANG TAO
LI HUI
LIU HUIQI
CHEN XUE
BAO JIADING

Assignees

南宁桂电电子科技研究院有限公司
桂林电子科技大学

Dates

Publication Date: 20260508
Application Date: 20260327

Claims (7)

1. A driving intention recognition improvement method, a device, equipment and a storage medium based on a heavy truck fuel-saving cruise control are characterized by comprising the following steps: S1, a data acquisition module is used for acquiring running state information and road environment information of a target vehicle and surrounding vehicles in real time; s2, a dynamic diagram construction module is used for modeling vehicles as diagram nodes, modeling inter-vehicle interaction as diagram edges and dynamically generating a vehicle interaction diagram; s3, a Long Short-Term Memory (LSTM) prediction module is connected with the dynamic diagram construction module and used for performing time sequence intention prediction based on the interaction characteristics extracted by the diagram neural network; s4, a three-mode switching control module is used for switching among an energy-saving mode, an adaptive mode and a safety mode according to a prediction result; S5, a road condition optimization module is used for carrying out path fitting and segmentation by combining real-time road network information; And S6, the fuel consumption optimization module is used for solving the cruise control output based on the prediction result and the fitting road condition, and realizing a fuel saving control strategy.
2. The control system of claim 1, wherein the S2 module models interaction relationships between nodes using a dynamic graph neural network (DYNAMIC GRAPH Neural Network, DGNN) and supports updating adjacency matrices in real time to accommodate traffic environment changes.
3. The control system of claim 1, wherein the S3 module is configured to perform a vehicle intent timing prediction based on the extracted map-embedded features, the prediction accuracy being not less than 94%.
4. The method of claim 3, further comprising an event-triggered attention allocation module for triggering attention weight reassignment upon occurrence of a particular event, thereby dynamically partitioning the prediction area, reducing ineffective computations.
5. The control system of claim 1, wherein the S4 module comprises: S41, an energy-saving mode, which is used for adopting an optimal oil consumption control strategy when the working condition is stable; S42, an adaptive mode, which is used for adjusting control parameters according to the intention of surrounding vehicles under the condition of medium traffic complexity; s43, a safety mode is used for guaranteeing the running safety of the vehicle preferentially when the potential collision risk exists.
6. The control system of claim 1, wherein the S5 module performs fitting modeling on road gradient, lane width, speed limit information, and the like based on the high-precision map data, and inputs the result to the fuel consumption optimization module.
7. The control system of claim 1, wherein the S6 module minimizes fuel consumption per unit distance based on a quadratic programming algorithm, while satisfying vehicle dynamic constraints and safety boundaries.

Description

Driving intention recognition improvement method, device and equipment based on fuel-saving cruise control of heavy truck and storage medium Technical Field The patent relates to the field of cruise control and fuel-saving control of commercial vehicles, in particular to a problem of prediction of the running intention of a commercial vehicle, and specifically relates to a method, a device, equipment and a storage medium for identifying and improving the running intention based on fuel-saving cruise control of a heavy commercial vehicle. Background With the development of intelligent traffic systems, automatic driving and advanced driving assistance systems are increasingly used in the field of commercial vehicles. The heavy commercial vehicle is used as the main force of road transportation, and the fuel consumption occupies an important proportion of the operation cost of enterprises, so the realization of efficient, energy-saving and safe driving control in the actual road environment is the important research point. Currently, the mainstream fuel-saving Cruise Control system (PREDICTIVE ADAPTIVE Cruise Control, PACC) depends on a fixed route, static rules or preset controller parameters, and is difficult to adapt to complex and changeable traffic environments and front vehicle behaviors. Conventional approaches are typically based on preset vehicle following models and rule controllers, lacking efficient recognition and predictive capabilities for the intent of the lead vehicle. In addition, when facing complex working conditions such as multi-vehicle interaction, ramp road conditions, emergencies and the like, the response of the existing system is often lagged, and the fuel economy and the driving safety are difficult to be considered. Meanwhile, with the development of deep learning and graph neural networks, how to effectively combine the graph neural networks with a time sequence prediction model and apply the graph neural networks to a vehicle control system becomes a key problem to be solved urgently. Therefore, it is necessary to construct a novel fuel-saving control system capable of dynamically modeling the interaction relation of vehicles, accurately predicting the intention of surrounding vehicles and adjusting the control strategy in real time by combining road information, so as to improve the adaptability, stability and energy-saving performance of the system. The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art. Disclosure of Invention The application mainly aims to provide a layered cruise control method, device, equipment and storage medium based on the driving intention of surrounding vehicles, which aim to solve the technical problem of how to plan safe and economical vehicle speed by considering the changing situation of the driving intention of the surrounding vehicles so as to save fuel with high efficiency. In order to achieve the above object, the present application provides a layered cruise control method based on the driving intention of a surrounding vehicle, the method comprising: predicting potential running behaviors of surrounding vehicles according to running information of the surrounding vehicles when the emergency occurs, and obtaining a running intention prediction result of the surrounding vehicles; And determining a vehicle cruise control mode according to the relative distance between the front vehicles, the relative speed and/or the prediction result of the running intention of the surrounding vehicles. Further, the vehicle control modes include an energy saving mode, an adaptive mode, and a safety mode. And completing fuel saving control according to the vehicle cruise control mode. In one embodiment, the predictive adaptive cruise control method based on the dynamic map-LSTM comprises the following steps: Determining a cruise control mode according to the vehicle running state and the traffic environment information, wherein the cruise control mode comprises an economic mode, an adaptive mode and a safety mode; Under the self-adaptive mode, a dynamic graph structure is constructed, the self-vehicle and surrounding target vehicles are modeled as nodes in a graph neural network, and the dynamic interaction relationship among the nodes is used as an edge vector; extracting interaction mode features among surrounding vehicles according to the graph neural network, and generating intermediate potential driving intents; Inputting the intermediate potential intention representation into an LSTM network to complete the sequence prediction of the future motion state of surrounding vehicles and generate multi-time-domain driving intention information; When the identified intention of the front vehicle is a specific event such as acceleration, deceleration or lane change, dividing an intention region by combining an