A new energy vehicle speed prediction system and method

Abstract

This application provides a new energy vehicle speed prediction system and method, relating to the field of speed prediction technology. It involves determining the driver's driving intention label during driving; based on the driving intention label, determining the behavioral evolution law of the target driving state in behavior identification and the driving trajectory anchor point during driving action feedback; determining the linkage decision elements of the current driving speed in the target driving behavior based on the behavioral evolution law and driving trajectory anchor point; further determining the static disturbance compatibility index and dynamic disturbance compatibility index of the driving speed in the driving behavior feature section; and determining the homomorphic fitting level of the current driving behavior in mountainous areas based on the static disturbance compatibility index and dynamic disturbance compatibility index; and registering and guiding the following speed of the current driving behavior in mountainous scenarios based on the linkage decision elements and homomorphic fitting level. This application can register and guide the following speed of the current driving behavior in complex road scenarios, thereby improving the accuracy of speed prediction in mountainous areas.

Description

Technical Field

[0001] This application relates to the field of vehicle speed prediction technology, and more specifically, to a new energy vehicle speed prediction system and method. Background Technology

[0002] Vehicle speed prediction is a fundamental technology for intelligent driving control of new energy vehicles, and a key prerequisite for realizing core functions such as optimized vehicle power control, driving safety warning, and adaptive cruise control. Vehicle speed prediction uses onboard sensing units, vehicle bus systems, and supporting perception equipment to collect real-time multi-source synchronous data on the new energy vehicle's own operating status, driver behavior, road environment parameters, and surrounding traffic conditions. It analyzes the driving factors and correlation patterns of vehicle speed changes, accurately predicts the vehicle's speed within a set future time domain, and outputs spatiotemporally adaptable speed change trends and prediction results, providing a basis for decision-making in the driving control of new energy vehicles.

[0003] However, traditional new energy vehicle speed prediction in mountainous areas lacks a registration mechanism that integrates driver intent recognition with quantitative constraints on multi-source roadside intrusions. This makes it difficult for the speed prediction process to adapt to the complex road characteristics of dynamic and static intrusions in mountainous areas, resulting in insufficient adaptability of the predicted speed results to the real-time road environment. The prediction accuracy is significantly reduced by interference from road-specific sources, making it difficult to achieve precise speed control in complex road scenarios. Therefore, how to register and guide the following speed of the current driving behavior in complex road scenarios to improve the accuracy of speed prediction in mountainous areas is a problem facing the industry. Summary of the Invention

[0004] This application provides a new energy vehicle speed prediction system and method, which can register and guide the following speed of the current driving behavior in complex road scenarios to improve the accuracy of speed prediction in mountainous areas.

[0005] Firstly, this application provides a method for predicting the speed of a new energy vehicle, the method comprising the following steps:

[0006] Data on driving disturbance sources of new energy vehicles in mountainous road environments are collected, and the driving disturbance source data is used to identify the driving intentions of the driver during driving to obtain the driving intention labels.

[0007] Based on the driving intention label, the behavioral evolution law of the target driving state in behavior identification and the driving trajectory anchor point when driving action feedback are determined, and then the linkage decision elements of the current driving speed in the target driving behavior are determined by the behavioral evolution law and the driving trajectory anchor point.

[0008] The roadside intrusion information of new energy vehicles in mountainous road environments is obtained, and the roadside intrusion information is used to identify waypoints to obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section. Then, the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed is determined by the static disturbance compatibility index and the dynamic disturbance compatibility index.

[0009] Based on the aforementioned linkage decision-making elements and the aforementioned homomorphic fitting hierarchy, the following speed of the current driving behavior in a mountainous scenario is registered and guided.

[0010] In this embodiment, the driving disturbance source data refers to the data set of new energy vehicles themselves and the environment that affect the driver's decision-making during mountain driving.

[0011] In this embodiment, the intention recognition of the driving disturbance source data to obtain the driver's driving intention label during driving specifically includes:

[0012] The driving response characteristics of drivers when driving on mountain roads are analyzed from the driving disturbance source data;

[0013] Based on the driving response characteristics, the driving disturbance source data is compared and mapped to extract the driver's driving decision sequence during driving;

[0014] The driving decision sequence is used to determine the driver's driving intention label during driving.

[0015] In this embodiment, determining the behavioral evolution pattern of the target driving state in behavior recognition and the driving trajectory anchor point during driving action feedback based on the driving intention label specifically includes:

[0016] Based on the driving intent label, the behavioral state attribute of the target driving state in behavior identification is analyzed;

[0017] The behavioral evolution law of the target driving state in behavior identification is determined based on the aforementioned behavioral state attributes;

[0018] Based on the aforementioned behavioral state attributes, the evolution process of driving behavior is collaboratively calibrated to obtain the driving trajectory anchor point at the time of driving action feedback.

[0019] In this embodiment, the driving trajectory anchor point refers to a point that provides spatial positioning for driving action feedback.

[0020] In this embodiment, the linkage decision element refers to the decision criteria of the current driving speed in the target driving behavior.

[0021] In this embodiment, the roadside intrusion information refers to the set of static and dynamic interference data on both sides of mountain roads that affect the driving safety of new energy vehicles.

[0022] In this embodiment, the noise compatibility index refers to the degree to which a fixed road environment adapts to the current driving speed.

[0023] In this embodiment, the disturbance compatibility index refers to the degree to which temporary and moving targets adapt to the current driving speed.

[0024] Secondly, this application provides a new energy vehicle speed prediction system for executing a new energy vehicle speed prediction method, the speed prediction system comprising:

[0025] The data acquisition module is used to collect driving disturbance source data of new energy vehicles in mountainous road environments, and to perform intent recognition on the driving disturbance source data to obtain the driver's driving intent label during driving.

[0026] The behavior decision module is used to determine the behavior evolution law of the target driving state in behavior identification and the driving trajectory anchor point when the driving action feedback is based on the driving intention label, and then determine the linkage decision elements of the current driving speed in the target driving behavior based on the behavior evolution law and the driving trajectory anchor point.

[0027] The waypoint identification module is used to acquire roadside intrusion information of new energy vehicles driving in mountainous road environments, perform waypoint identification on the roadside intrusion information, obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section, and then determine the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed by the static disturbance compatibility index and the dynamic disturbance compatibility index.

[0028] The registration guidance module is used to register and guide the following speed of the current driving behavior in a mountainous scene based on the linkage decision elements and the homomorphic fitting level.

[0029] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0030] Data on driving disturbance sources of new energy vehicles in mountainous road environments is collected, and intent recognition is performed on the driving disturbance source data to obtain the driver's driving intent label during driving. Based on the driving intent label, the behavioral evolution law of the target driving state in behavior identification and the driving trajectory anchor point at the time of driving action feedback are determined. Then, the linkage decision elements of the current driving speed in the target driving behavior are determined by the behavioral evolution law and the driving trajectory anchor point. Roadside intrusion information of new energy vehicles in mountainous road environments is obtained, and waypoint identification is performed on the roadside intrusion information to obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section. Then, the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed is determined by the static disturbance compatibility index and the dynamic disturbance compatibility index. Based on the linkage decision elements and the homomorphic fitting level, the following speed of the current driving behavior in the mountainous scenario is registered and guided.

[0031] Therefore, this application uses the aforementioned linkage decision elements and homomorphic fitting hierarchy to guide the following speed of the current driving behavior in a mountainous scenario. Specifically, determining the linkage decision elements yields the vehicle speed decision benchmark matching the driving intention in mountainous areas and the core control parameters for the collaborative adaptation of driving behavior and speed. This establishes a strong correlation mapping system between driving intention, behavioral evolution patterns, spatial trajectory anchor points, and driving speed, overcoming the technical shortcomings of traditional vehicle speed prediction where speed decisions are disconnected from the driver's subjective driving intention, behavioral temporal logic, and spatial driving trajectory. It provides an intention-driven endogenous decision basis for following speed registration, achieving deep adaptation between vehicle speed prediction results and the driver's actual driving behavior, effectively improving the temporal coherence and behavioral fit of vehicle speed prediction in complex mountainous scenarios. By determining the static disturbance compatibility index and the dynamic disturbance compatibility index, we can obtain the quantitative adaptation index of the static fixed environment and dynamic moving disturbance on the vehicle speed constraint in mountainous roads, as well as the environmental disturbance classification and control benchmark. This enables the dynamic-static separation analysis of multi-source disturbance information on the roadside in mountainous areas and the accurate quantification of the vehicle speed constraint effect. It makes up for the technical defects of traditional vehicle speed prediction, such as the difficulty in quantifying roadside environmental constraints and the coupling interference of dynamic and static disturbances. It provides an exogenous constraint benchmark for car-following speed registration with environmental adaptation, and achieves a high degree of adaptation between the vehicle speed prediction results and the real-time road environment in mountainous areas. This effectively improves the environmental anti-interference and spatial adaptability of vehicle speed prediction in complex scenarios.

[0032] In summary, the technical solution adopted in this application can register and guide the following speed of the current driving behavior in complex road scenarios, thereby improving the accuracy of vehicle speed prediction in mountainous areas. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is an exemplary flowchart of a method for predicting the speed of a new energy vehicle provided in this application;

[0035] Figure 2 This is a flowchart illustrating the process of determining the linkage decision-making elements provided in this application;

[0036] Figure 3 This is a flowchart illustrating the determination of the static disturbance compatibility index and the dynamic disturbance compatibility index based on the information provided in this application.

[0037] Figure 4 This is a diagram of the multi-source input multi-output neural network model for predicting the speed of new energy vehicles in mountainous areas, provided in this application.

[0038] Figure 5 This is a module structure diagram of a new energy vehicle speed prediction system provided in this application. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0040] This application provides a method for predicting the speed of new energy vehicles. The core of this method is to collect driving disturbance source data of new energy vehicles in mountainous road environments, and to perform intent recognition on the driving disturbance source data to obtain the driver's driving intent label during driving. Based on the driving intent label, the method determines the behavioral evolution law of the target driving state in behavior identification and the driving trajectory anchor point at the time of driving action feedback. Then, the method determines the linkage decision elements of the current driving speed in the target driving behavior based on the behavioral evolution law and the driving trajectory anchor point. The method also acquires roadside intrusion information of new energy vehicles in mountainous road environments, performs waypoint identification on the roadside intrusion information, and obtains the static disturbance compatibility index and dynamic disturbance compatibility index of the driving speed in the driving behavior feature section. Then, the method determines the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed based on the static disturbance compatibility index and the dynamic disturbance compatibility index. Finally, the method registers and guides the following speed of the current driving behavior in the mountainous scenario based on the linkage decision elements and the homomorphic fitting level.

[0041] Example 1: To better understand the above technical solution, the following will provide a detailed description of the technical solution in conjunction with the accompanying drawings and specific implementation methods. (Refer to...) Figure 1 As shown in the figure, this is an exemplary flowchart of a new energy vehicle speed prediction method according to this embodiment of the present application. The speed prediction method includes the following steps:

[0042] In step S1, driving disturbance source data of new energy vehicles in mountainous road environment is collected, and the driving disturbance source data is used to identify the driving intention label of the driver during driving.

[0043] In specific implementation, the following methods can be used to collect driving disturbance source data of new energy vehicles in mountainous road environments: The vehicle's own data, such as speed, steering angle, throttle and brake opening, is collected in real time via the vehicle's CAN bus to the ECU; an IMU is installed on the chassis to collect vehicle attitude data; a millimeter-wave radar is installed on the front grille to detect the distance to the vehicle in front and oncoming vehicles; dual cameras are installed on the roof to identify curve signs and pedestrians; a high-precision GPS is installed on the center console to obtain location and slope gradient. All devices are connected to the computing unit via the vehicle's Ethernet, and data is synchronized with GPS timestamps. Then, outliers are removed using the 3σ principle, and disconnected data is filled using linear interpolation. The processed data is used as the driving disturbance source data for new energy vehicles in mountainous road environments. In other embodiments, other methods can also be used to collect driving disturbance source data, which will not be elaborated here.

[0044] It should be noted that, in this application, driving disturbance source data refers to the data set of new energy vehicles themselves and their relationship with the environment that affect drivers' decisions while driving in mountainous areas.

[0045] In this embodiment, the intention recognition of the driving disturbance source data to obtain the driver's driving intention label during driving can be achieved through the following steps:

[0046] The driving response characteristics of drivers when driving on mountain roads are analyzed from the driving disturbance source data;

[0047] Based on the driving response characteristics, the driving disturbance source data is compared and mapped to extract the driver's driving decision sequence during driving;

[0048] The driving decision sequence is used to determine the driver's driving intention label during driving.

[0049] In specific implementation, firstly, in mountainous road scenarios, it is necessary to determine the operational response characteristics of new energy vehicles. These operational response characteristics include: calculating the ratio of the speed difference between two consecutive sampling points to a 0.1-second time interval as the speed change rate; calculating the ratio of the steering angle difference between two consecutive sampling points to a 0.1-second time interval as the steering angle change rate; selecting the average of five consecutive sampling points as the throttle opening average; and selecting the maximum value of ten consecutive sampling points as the brake opening peak value. Secondly, it is necessary to determine the environmental interaction response characteristics, including: calculating the ratio of the distance difference between two consecutive sampling points to a 0.1-second time interval as the distance change rate; and using the degree of matching between the current steering angle and the curvature of the curve to represent the curvature adaptation degree. The higher the curvature adaptation degree, the closer the value is to 1. All the obtained data are used as the driving response characteristics of the driver when driving on mountainous roads. Then, 500 sets of driving disturbance source data for typical mountain driving scenarios can be collected. Multiple experienced mountain drivers or linear fitting methods can be used to label the "feature-single-step decision" correspondence. This correspondence includes "emergency deceleration decision" corresponding to a vehicle speed change rate ≤ -2km / h·s and a peak brake opening ≥ 50%, and "cornering steering decision" corresponding to a steering angle change rate ≥ 3° / s and a curve curvature fit ≥ 0.8. A decision tree algorithm is used to train the mapping model. The maximum depth of the decision tree can be set to 8, and the minimum number of sample splits to 10. Optimization is achieved through 5-fold cross-validation to form a mapping rule base. One driving response feature vector is generated per second from the driving disturbance source data and input into the mapping model to obtain single-step driving decisions. The single-step decisions from 10 consecutive time points are then concatenated in chronological order to form the driver's driving decision sequence during driving. Finally, for driving intentions in mountainous areas, 30 standard decision sequence templates are constructed for each type of driving intention. These templates can be: for example, a curve deceleration template could be "normal deceleration → curve turning → constant speed following → normal acceleration," forming an intention template library. A dynamic time warping algorithm is then used to construct a similarity model, converting the sequence to be matched and the template sequence into decision encoding vectors. The minimum distance between the two vectors is calculated, and a similarity threshold of 0.8 is set. The extracted driving decision sequence is converted into an encoding vector, and dynamic time warping similarity is calculated with all templates in the template library. The number of templates with a similarity ≥ 0.8 for each type of intention is counted, and the one with the highest similarity is selected as the candidate intention. If the number of candidate intention templates accounts for ≥ 70% of the total number of templates in that type, it is directly output as a driving intention label. If the number of candidate intention templates accounts for < 70% of the total number of templates in that type, a new driving decision sequence is added and the calculation is repeated until the ratio reaches the threshold.

[0050] It should be noted that, in this application, driving response characteristics refer to the key characteristics of a driver's operational feedback in response to changes in the mountain road environment; driving decision sequence refers to the set of orderly driving decisions made by the driver over a continuous period of time; and driving intention label refers to the explicit behavioral identifier of the driver while driving on mountain roads.

[0051] In step S2, the behavioral evolution law of the target driving state in behavior recognition and the driving trajectory anchor point when driving action feedback are determined according to the driving intention label. Then, the linkage decision elements of the current driving speed in the target driving behavior are determined by the behavioral evolution law and the driving trajectory anchor point.

[0052] In this embodiment, determining the behavioral evolution pattern of the target driving state in behavior recognition and the driving trajectory anchor point during driving action feedback based on the driving intention label can be achieved through the following steps:

[0053] Based on the driving intent label, the behavioral state attribute of the target driving state in behavior identification is analyzed;

[0054] The behavioral evolution law of the target driving state in behavior identification is determined based on the aforementioned behavioral state attributes;

[0055] Based on the aforementioned behavioral state attributes, the evolution process of driving behavior is collaboratively calibrated to obtain the driving trajectory anchor point at the time of driving action feedback.

[0056] In practice, the process begins by collecting behavioral state attribute categories corresponding to driving intention labels on mountain roads. These categories include attributes such as "acceleration intensity, lane-changing timing, and oncoming vehicle distance threshold" for the "overtaking" intention, and "deceleration advance, steering angle range, and curve curvature adaptation value" for the "curve deceleration" intention. Then, 300 sets of driving disturbance source data under the same driving intention label are collected. Five traffic engineers annotate the attribute values ​​of each set of data. A support vector machine model is then trained using Python's Scikit-learn library. The input is the driving intention label encoding, and the output is the corresponding set of behavioral state attributes. The model penalty parameter C=1.0 and the kernel function is RBF. Optimization is achieved through 5-fold cross-validation. In real-time scenarios, inputting the current driving intention label into the support vector machine model will output the corresponding behavioral state attributes. Then, for the behavioral state attribute corresponding to a certain driving intention label, the behavioral state under that driving intention is divided. The behavioral state can be divided into a preparation state according to the attribute of the "overtaking" intention. Then, 200 sets of behavioral state sequence data under the same map are collected, and a hidden Markov model is used for training. The number of transitions between each state is counted, and the state transition matrix is ​​calculated. The transition probability in the state transition matrix is ​​equal to the number of transitions / the total number of transitions. The observation probability matrix is ​​set as a Gaussian distribution of the behavioral state attribute. The Baum-Welch algorithm is used to optimize the model. 100 sets of test sequences are decoded by the Viterbi algorithm, and the sequences with a repetition rate of ≥85% are extracted as the behavioral evolution law of the target driving state in behavior identification. Finally, spatially related features from the behavioral state attributes were selected. The "deceleration advance" and "curvature adaptation value" for the "curve deceleration" intention, and the "lane merging timing distance" for the "overtaking" intention were used as calibration features. 150 sets of driving trajectory data were collected under the same map. Trajectory segments were extracted according to each state of the behavioral evolution pattern. The DBSCAN clustering algorithm was used to cluster the trajectory segments, setting the neighborhood radius ε=2m (this radius is set based on the spacing between trajectory points on the matching mountain roads), and the minimum sample size MinPts=4. Noise points were removed, and the calibration feature value of each trajectory point within the cluster was calculated. The deceleration advance in this calibration feature value is equal to the distance from the current point to the apex of the curve. Feature values ​​that satisfy the behavioral state attribute threshold (e.g., deceleration advance = 50m, within the 40-60m range of the "curve deceleration" attribute, and with the largest absolute value of the rate of change of speed) were selected as the initial anchor points of the cluster. The mean coordinates of the initial anchor points of 100 sets of test trajectories were calculated, and this mean coordinates were used as the driving trajectory anchor points for driving action feedback.

[0057] It should be noted that, in this application, behavior recognition refers to the process of identifying the target driving state by analyzing driving data; driving action feedback refers to the operational response made by the driver or new energy vehicle in response to driving intention / environment; behavior state attributes refer to the set of attributes that reflect the core characteristics of the target driving state; behavior evolution law refers to the ordered logical sequence of driving behavior changes over time; and driving trajectory anchor point refers to the point that plays a spatial positioning role in driving action feedback.

[0058] Preferably, in this embodiment, the linkage decision-making elements of the current driving speed in the target driving behavior are determined by the behavioral evolution law and the driving trajectory anchor point, with reference to... Figure 2 As shown in the figure, this is a flowchart illustrating the process of determining linkage decision elements in some embodiments of this application. In this embodiment, the determination of linkage decision elements can be achieved through the following steps:

[0059] In step S21, the transition decision-making pattern of the current driving speed under the target driving behavior is analyzed according to the behavior evolution law;

[0060] In step S22, trajectory decision parameters that are dynamically adapted to the driving speed are generated based on the connection decision mode;

[0061] In step S23, the collaborative decision-making sequence of the current driving speed in the target driving behavior is determined based on the driving trajectory anchor points;

[0062] In step S24, the linkage decision elements of the current driving speed in the target driving behavior are determined by the trajectory decision parameters and the collaborative decision sequence.

[0063] In practice, the first step is to break down the behavioral evolution of the target driving behavior into continuous behavioral states. 300 sets of behavioral states and corresponding driving speed data under the same map are collected. Each set of driving speed data includes the start / end speed of each state. A decision tree model can be built using Python's Scikit-learn library. The input is "current behavioral state, previous behavioral state, current driving speed", and the output is "speed connection decision". The maximum depth of the decision tree is set to 6 and the minimum number of sample splits is set to 8. The behavioral evolution pattern to be analyzed is input into the decision tree model, and the speed decision logic when connecting each state is output. This is integrated into the connection decision mode of the current driving speed under the target driving behavior. Next, based on the speed connection logic in the connection decision mode, the trajectory decision parameter type is determined. This trajectory decision parameter type can be "lane-changing speed threshold and trajectory deviation tolerance" for the "overtaking" intention, or "entry speed limit and trajectory curvature adaptation coefficient" for the "curve deceleration" intention. 250 sets of driving speed and corresponding trajectory data under the same map are collected, and the initial parameter values ​​at each speed are calculated using Python's NumPy library. A linear regression model is used to train the dynamic relationship between the parameters and the input driving speed and the output trajectory decision parameter values. The model learning rate is set to 0.001, and the training rounds are 50, forming trajectory decision parameters that dynamically change with speed, i.e., trajectory decision parameters that dynamically adapt to driving speed. Then, the spatial features of the driving trajectory anchor points are extracted. 200 sets of anchor point spatial features and corresponding driving speed data under the same map are collected. Each set of driving speed data contains the suggested driving speed for each anchor point. An LSTM model is built using Python's TensorFlow library. The input is the anchor point coordinates arranged in the order of behavior evolution, and the output is the "speed decision sequence". The number of hidden units of the LSTM is set to 64, the number of training rounds is 30, and the optimizer is Adam (learning rate 0.001). Finally, the output is the speed decision sequence that is coordinated with the anchor point spatial features, that is, the coordinated decision sequence of the current driving speed in the target driving behavior.Finally, a fusion model of "trajectory decision parameters - collaborative decision sequence" is established. First, the weights of the trajectory decision parameters and collaborative decision sequence are determined. These weights can be obtained through scoring by five traffic engineers, which is not limited here. The weight of the trajectory decision parameters is 0.4, and the weight of the collaborative decision sequence is 0.6. Trajectory decision parameters, collaborative decision sequences, and corresponding actual safe driving speed data are collected under 150 sets of the same map. The trajectory decision parameters and collaborative decision sequences are weighted and summed, i.e., comprehensive speed suggestion = 0.4 × parameter speed threshold + 0.6 × sequence anchor point speed, to obtain the preliminary linkage decision elements. Among them, the parameter speed threshold refers to the highest driving speed that meets the safety requirements of the current driving behavior trajectory. This parameter speed threshold is determined by the lane change speed threshold and the upper limit of the curve entry speed in the trajectory decision parameters. The sequence anchor point speed refers to the safe driving speed corresponding to each anchor point position, which is arranged in the order of behavior evolution to form an anchor point speed sequence. Then, by comparing the deviation between the preliminary elements and the actual safe speed, the weights are adjusted to minimize the deviation, and the fused linkage decision elements are determined, i.e., the linkage decision elements of the current driving speed in the target driving behavior.

[0064] It should be noted that, in this application, the connection decision mode refers to the decision logic mode that reflects the connection between the various states of the target driving behavior when the current driving speed is connected; the trajectory decision parameters refer to the trajectory association parameters that are dynamically adjusted with the current driving speed; the collaborative decision sequence refers to the ordered set of decisions for the current driving speed in the target driving behavior; and the linkage decision elements refer to the decision criteria for the current driving speed in the target driving behavior.

[0065] In step S3, roadside intrusion information of new energy vehicles driving in mountainous road environments is obtained, and waypoint identification is performed on the roadside intrusion information to obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section. Then, the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed is determined by the static disturbance compatibility index and the dynamic disturbance compatibility index.

[0066] In practice, obtaining roadside intrusion information of new energy vehicles driving in mountainous road environments can be achieved in the following way: Roadside equipment can be deployed every 500 meters along mountainous roads. Each set of equipment includes a 2-megapixel, 25fps, 120° field-of-view high-definition camera and a 16-line LiDAR to collect static and dynamic disturbance data. Static disturbances include missing guardrails and sharp bends, while dynamic disturbances include pedestrians and falling rocks. The data is transmitted to the vehicle-mounted computing unit via a 5G private network. At the same time, the vehicle-mounted front-view camera uses the YOLOv8 algorithm to identify intrusion targets, and the vehicle-mounted LiDAR measures the distance between the target and the lane. The roadside and vehicle-mounted data are synchronized according to GPS timestamps. For the same target, confidence-weighted fusion is performed, and low-confidence data is eliminated to finally obtain the roadside intrusion information of new energy vehicles driving in mountainous road environments.

[0067] It should be noted that, in this application, roadside intrusion information refers to the set of static and dynamic interference data on both sides of mountain roads that affect the driving safety of new energy vehicles.

[0068] Preferably, in this embodiment, waypoint identification is performed on the roadside intrusion information to obtain the static disturbance compatibility index and dynamic disturbance compatibility index of the driving speed in the driving behavior characteristic section, with reference to... Figure 3 As shown in the figure, this is a flowchart illustrating the determination of the static disturbance compatibility index and the dynamic disturbance compatibility index in some embodiments of this application. In this embodiment, the determination of the static disturbance compatibility index and the dynamic disturbance compatibility index can be achieved by the following steps:

[0069] In step S31, the roadside intrusion information is analyzed to generate waypoint interference features of driving speed in driving behavior;

[0070] In step S32, the dynamic and static modes are separated according to the constraint relationship of the driving speed based on the waypoint interference characteristics to obtain static intrusion clusters and dynamic intrusion clusters;

[0071] In step S33, a static disturbance compatibility index of driving speed in the driving behavior characteristic section is generated based on the static disturbance cluster;

[0072] In step S34, a dynamic disturbance compatibility index of driving speed in the driving behavior characteristic section is generated based on the dynamic disturbance cluster.

[0073] In practice, firstly, roadside intrusion information is divided into waypoints at 10-meter intervals. Intrusion features for each waypoint are extracted: static and dynamic features. The Geopandas library in Python is used to map the intrusion information to the corresponding waypoints. For consecutive waypoints, missing values ​​are supplemented using linear interpolation. The "interference intensity" of each waypoint is calculated, ultimately forming a waypoint interference feature dataset consisting of "waypoint coordinates - static features - dynamic features - interference intensity," representing the waypoint interference features of driving speed in driving behavior. Next, "feature temporal stability" is selected as the separation metric. The waypoint interference features are clustered using the DBSCAN algorithm from the Scikit-learn library in Python: static features with a coefficient of variation ≤ 0.1 are clustered into static intrusion clusters; dynamic features with a coefficient of variation > 0.1 are clustered into dynamic intrusion clusters. Isolated clusters with less than 3 samples after clustering are removed, resulting in static and dynamic intrusion clusters. Next, for static intrusion clusters, evaluation indicators are determined, including curve curvature, slope gradient, and guardrail missing length. Weights are assigned to these indicators, with possible weights of 0.4 for curve curvature, 0.3 for slope gradient, and 0.3 for guardrail missing length; no specific limit is imposed here. Then, 200 sets of static intrusion cluster data and corresponding safe driving speeds are collected, and standardized values ​​for each indicator are calculated. These standardized values ​​include curvature... Corresponding to 0, Corresponding to point 1, the static disturbance compatibility index is calculated using a weighted summation formula: Static Disturbance Compatibility Index = 0.4 × Curvature Standardized Value + 0.3 × Slope Standardized Value + 0.3 × (1 - Guardrail Missing Length / 100), then multiplied by the standardized value of the current driving speed to obtain the static disturbance compatibility index of driving speed in the driving behavior characteristic section. The statistical calculation of each index is based on 200 sets of measured data from mountain static disturbance scenarios covering different curve curvatures, different slopes, and different guardrail integrity levels. Each set of data includes road environment parameters and statistical values ​​of safe driving speeds from multiple experienced mountain drivers in the corresponding scenarios. The "maximum speed at which 95% of drivers can drive safely" is set as the safety benchmark. The upper and lower limits of each index are determined using the percentile method in statistics: Curvature Index: Statistically, when the road curvature ≤ At this time, over 95% of drivers can maintain normal driving speed; when the curvature is ≥ At that time, it is necessary to significantly decelerate to below the safety threshold; therefore, the upper and lower limits of the curvature normalization value are determined as follows: (Corresponding to standardized value 1) and (Corresponding to standardized value 0). Slope index: Statistical analysis shows that when the absolute value of the road slope is ≤5°, there is no significant impact on driving speed; when the absolute value of the slope is ≥15°, a significant reduction in speed is required. Therefore, the upper and lower limits of the standardized slope value are determined to be 5° (corresponding to standardized value 1) and 15° (corresponding to standardized value 0). Guardrail missing length index: Statistical analysis shows that when the missing length of the guardrail is ≥100m, it poses a serious threat to driving safety; when the missing length of the guardrail is 0, there is no safety impact. Therefore, the upper and lower limits of the standardized guardrail missing length value are determined to be 0m (corresponding to standardized value 1) and 100m (corresponding to standardized value 0). Based on the upper and lower limit thresholds of each index of the static disturbance compatibility index, all intermediate values ​​are calculated using linear interpolation. The specific formula is: Curvature standardized value: When <Curvature< When the slope is 5° < absolute slope < 15°, the standardized value is (15 - |slope|) / (15 - 5). The standardized value for the missing guardrail length is 1 - missing guardrail length / 100m when the slope is 0m < missing guardrail length < 100m. Finally, the evaluation indicators for dynamic intrusion clusters are determined. The weights can be determined by expert scoring, with weights including distance (0.5), speed (0.3), and number (0.2), which are not limited here. 150 sets of dynamic intrusion cluster data and their corresponding safe driving speeds are collected. Standardized values ​​for each indicator are calculated. These standardized values ​​include 1 for distance ≥10m and 0 for ≤3m; 0 for speed ≥20km / h and 1 for ≤5km / h. The disturbance compatibility index is calculated using a weighted summation formula: Distance compatibility index = 0.5 × distance standardized value + 0.3 × (1 - speed standardized value) + 0.2 × (1 - number / 5). This is then multiplied by the standardized value of the current driving speed to obtain the driving speed in relation to driving behavior. The disturbance compatibility index in the characteristic section is calculated based on 150 sets of measured data covering dynamic intrusion scenarios in mountainous areas with different distances, speeds, and quantities. These scenarios include typical scenarios such as pedestrians crossing, non-motorized vehicles occupying the road, and falling rocks. The safety benchmark is the "maximum speed at which 95% of drivers can safely avoid the intrusion." The upper and lower limits of each index are determined using the percentile method: Distance index: When the distance between the intruding target and the vehicle is ≥10m, there is sufficient time to avoid it; when the distance is ≤3m, avoidance is extremely difficult. Therefore, the upper and lower limits of the distance standardized value are determined to be 10m (corresponding to standardized value 1) and 3m (corresponding to standardized value 0). Speed ​​index: When the intruding target's moving speed is ≤5km / h, it is easy to avoid; when the speed is ≥20km / h, the difficulty of avoidance increases significantly. Therefore, the upper and lower limits of the speed standardized value are determined to be 5km / h (corresponding to standardized value 1) and 20km / h (corresponding to standardized value 0). Quantity Indicators: When the number of intruding targets is ≥5, multi-target avoidance becomes extremely difficult; when the number is 0, there is no dynamic intrusion. Therefore, the upper and lower limits of the quantity standardization value are determined to be 0 (corresponding to a standardization value of 1) and 5 (corresponding to a standardization value of 0). Based on the upper and lower limit thresholds of each indicator of the disturbance compatibility index, all intermediate values ​​are calculated using linear interpolation. The specific formulas are as follows: Distance Standardization Value: When 3m < distance < 10m, the standardization value = (distance - 3) / (10 - 3). Speed ​​Standardization Value: When 5km / h < speed < 20km / h, the standardization value = (20 - speed) / (20 - 5). Quantity Standardization Value: When 0 < quantity < 5, the standardization value = 1 - quantity / 5.

[0074] It should be noted that, in this application, waypoint identification refers to the process of locating key waypoints that affect the driving of new energy vehicles from roadside information; waypoint interference features are key features reflecting the interference of specific waypoints on driving speed; driving behavior feature profiles refer to a set that centrally reflects the correlation between the driving behavior of new energy vehicles and roadside intrusion in a specific road segment; driving speed constraint relationship refers to the relationship between roadside intrusion-related factors and driving speed, which limits and guides speed adjustment to adapt to road conditions and ensure driving safety; static intrusion cluster refers to the set of intrusions in the waypoint interference features where the constraint relationship on driving speed does not change over time; dynamic intrusion cluster refers to the set of intrusions in the waypoint interference features where the constraint relationship on driving speed changes over time; static disturbance compatibility index refers to the degree of adaptation of a fixed road environment to the current driving speed; dynamic disturbance compatibility index refers to the degree of adaptation of temporary and moving targets to the current driving speed.

[0075] In this embodiment, determining the homomorphic fitting level of the current driving behavior at mountain speeds using the static disturbance compatibility index and the dynamic disturbance compatibility index can be achieved through the following steps:

[0076] The homomorphic response information of the current driving behavior at mountain speeds is determined based on the static disturbance compatibility index and the dynamic disturbance compatibility index.

[0077] The fitted guidance sequence for the vehicle speed evolution process in mountainous areas is determined by the homomorphic response information.

[0078] The homomorphic fitting level of the current driving behavior at mountain speeds is determined by the fitted guidance sequence.

[0079] In practice, firstly, the weights of the static disturbance compatibility index and the dynamic disturbance compatibility index are determined. This can be achieved through traffic engineer scoring or multiple experimental simulations, with a static disturbance weight of 0.45 and a dynamic disturbance weight of 0.55. No specific limit is imposed here. Then, 300 sets of driving data from mountainous areas are collected, each set containing the static disturbance compatibility index, the dynamic disturbance compatibility index, and the corresponding vehicle speed adaptation status. A linear regression model is constructed using Python's Scikit-learn library. The input is a fusion value of "static disturbance compatibility index × 0.45 + dynamic disturbance compatibility index × 0.55", and the output is homomorphic response information. This homomorphic response information includes: a fusion value of 0.8 corresponds to "high vehicle speed adaptation, able to maintain the current speed", and a fusion value of 0.3 corresponds to "low vehicle speed adaptation". "A significant reduction in speed is required." The current static and dynamic disturbance indices are input into the model to obtain the homomorphic response information of the current driving behavior at mountain speeds. The weight coefficients in the fusion value calculation formula "fusion value = static disturbance compatibility index × 0.45 + dynamic disturbance compatibility index × 0.55" are obtained by performing multiple linear regression analysis on 300 sets of comprehensive driving scenario data in mountainous areas that simultaneously include static and dynamic disturbances. The actual safe driving speed is used as the dependent variable, and the static and dynamic disturbance compatibility indices are used as independent variables. The regression coefficients are calculated using the least squares method and then normalized to obtain the weight allocation. This weight allocation can reflect the rule that the impact of dynamic disturbances on driving safety in mountainous roads is slightly greater than that of static disturbances. When the fusion value is ≥0.8, statistics show that in 92% of scenarios, the driver can drive safely without adjusting the vehicle speed, thus corresponding to "high speed compatibility, able to maintain the current speed"; when 0.6≤fusion value<0.8, statistics show that in 87% of scenarios, the driver needs to slightly reduce speed by 5-10 km / h to ensure safety, thus corresponding to "slight speed mismatch, requiring slight reduction of 5-10 km / h"; when 0.3≤fusion value<0.6, statistics show that in 90% of scenarios, the driver needs to reduce speed by 10-20 km / h to ensure safety, thus corresponding to "moderate speed mismatch, requiring reduction of 10-20 km / h"; when the fusion value<0.3, statistics show that in over 95% of scenarios, the driver needs to significantly reduce speed by more than 20 km / h to avoid accidents, thus corresponding to "severe speed mismatch, requiring significant reduction of more than 20 km / h". Then, homomorphic response information for 10 consecutive seconds is extracted in chronological order, and each piece of information is converted into a vehicle speed adjustment command. The vehicle speed adjustment command includes "high fit" corresponding to "maintain current speed", "slight misfit" corresponding to "reduce 5-10 km / h", and "severe misfit" corresponding to "reduce 10-20 km / h". A model is constructed, with the input being the response information encoding of "maintain" code 1 and "slight reduction" code 2 for 10 consecutive seconds, and the output being a fitted guidance sequence of "maintain → slight reduction → slight reduction → maintain". Finally, the fitted guidance sequence of the vehicle speed evolution process in the mountainous area is output.Finally, the hierarchical classification criteria were set: Level 1 is defined as the proportion of "maintain / minor adjustment" instructions in the statistically fitted guidance sequence ≥80%, Level 2 is defined as the proportion 50% to 80%, and Level 3 is defined as the proportion <50%. 200 sets of fitted guidance sequences and corresponding actual vehicle speed limit requirements were collected. Using a decision tree model, the input is the proportion of each instruction in the sequence, and the output is the homomorphic fitting level. The current fitted guidance sequence is input into the model, the proportion of each instruction is statistically analyzed, and the model outputs the corresponding homomorphic fitting level, thus obtaining the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed range.

[0080] It should be noted that in this application, homomorphic response information is comprehensive information reflecting the adaptation state of the current driving behavior and the vehicle speed in the mountainous area; the fitted guidance sequence refers to the orderly adjustment direction sequence that guides the vehicle speed in the mountainous area to change with driving behavior and environmental changes; and the homomorphic fitting level refers to the speed limit level of the degree of adaptation between the current driving behavior and the vehicle speed in the mountainous area.

[0081] In step S4, the following speed of the current driving behavior in the mountainous scene is registered and guided based on the linkage decision elements and the homomorphic fitting level.

[0082] In this embodiment, the registration guidance of the following speed of the current driving behavior in a mountainous scenario based on the linked decision elements and the homomorphic fitting hierarchy can be achieved through the following steps:

[0083] Based on the aforementioned linkage decision-making elements and the aforementioned homomorphic fitting hierarchy, the speed registration interval for the current driving behavior under mountainous following behavior is determined;

[0084] By coupling and matching the speed registration interval with the real-time driving situation, a cascaded decision instruction guided by speed is generated;

[0085] The cascaded decision command triggers coordinated control of the following speed of the current driving behavior in mountainous scenarios, and guides and predicts the driving speed in mountainous areas.

[0086] In practice, the process begins by extracting the "target speed range" from the linkage decision elements. Then, it combines the speed limit rules of the homomorphic fitting hierarchy: Level 1 retains the target speed range; Level 2 reduces the target speed upper limit by 10% to 20%; Level 3 reduces it by 20% to 50%. The adjusted upper and lower speed limits are calculated using Python's NumPy library. 200 sets of historical data on following vehicles in mountainous areas are collected to verify the safety of the interval. If the verification fails, the reduction ratio corresponding to the level is fine-tuned. Finally, the speed registration interval of the current driving behavior under following vehicle behavior in mountainous areas is obtained. Then, the real-time driving situation includes the current driving speed, the distance to the vehicle in front, and changes in roadside intrusions. The Pandas library in Python collects and processes the real-time driving situation every second: if the current speed is within the speed registration range and the distance to the vehicle in front is ≥50m and there are no new intrusions, a normal-level instruction of "maintain current speed" is generated; if the current speed is 5-10km / h higher than the upper limit of the range or the distance to the vehicle in front is 30-50m, a warning-level instruction of "slight deceleration" is generated; if the current speed is ≥10km / h higher than the upper limit of the range or the distance to the vehicle in front is <30m and there are new intrusions, an emergency-level instruction of "emergency deceleration" is generated. The instructions are encoded according to the "normal-warning-emergency" hierarchy, and the encoded results are used as cascaded decision instructions guided by speed. Finally, upon receiving the cascaded decision command, the vehicle's HMI system displays the command content on the instrument panel and flashes red; the audio system plays the corresponding voice prompt; if the vehicle speed does not reach the target within 10 seconds, the ESP system controls the brake pedal to apply slight braking, while the engine ECU reduces torque output to avoid skidding caused by sudden braking, thus completing coordinated control. Using the ARIMA time series model, the vehicle speed data of the past 10 seconds and the type of cascaded decision command are input to predict the vehicle speed trend for the next 5 seconds. If the predicted vehicle speed will exceed the registration range, a prediction command of "decelerating 2-3 km / h in advance" is generated in advance to complete the guidance prediction, which will not be elaborated here.

[0087] It should be noted that, in this application, the speed registration range refers to the reasonable driving speed range under the current mountain following behavior; the following speed refers to the driving speed of the following vehicle when following the preceding vehicle on a mountain road; the cascaded decision command refers to the conversion of the static speed range into a dynamic command that fits the real-time scenario; and the coordinated control guidance refers to the process of converting the command into actual speed adjustment actions to ensure that the speed conforms to the registration range.

[0088] In this embodiment, reference Figure 4As shown in the figure, this is the neural network model architecture diagram used in the new energy vehicle speed prediction method of this invention. It is used to achieve accurate speed prediction and car-following speed registration guidance in complex mountain road scenarios. The model input layer corresponds to the collected multi-source core data, including historical speed data, real-time road condition information, and vehicle status parameters. The intermediate feature extraction layer completes the deep feature mining and fusion of multi-source data through a fully connected network. It is a processing link such as driving response feature analysis, driving behavior evolution law extraction, and static and dynamic disturbance compatibility index calculation. It realizes the feature extraction of linkage decision elements and homomorphic fitting layer. The model output layer outputs the predicted speed value, driving intention decision result, and car-following guidance speed, that is, it completes the accurate speed prediction and car-following speed adaptation guidance in mountain scenarios.

[0089] Therefore, this application uses the aforementioned linkage decision elements and homomorphic fitting hierarchy to guide the following speed of the current driving behavior in a mountainous scenario. Specifically, determining the linkage decision elements yields the vehicle speed decision benchmark matching the driving intention in mountainous areas and the core control parameters for the collaborative adaptation of driving behavior and speed. This establishes a strong correlation mapping system between driving intention, behavioral evolution patterns, spatial trajectory anchor points, and driving speed, overcoming the technical shortcomings of traditional vehicle speed prediction where speed decisions are disconnected from the driver's subjective driving intention, behavioral temporal logic, and spatial driving trajectory. It provides an intention-driven endogenous decision basis for following speed registration, achieving deep adaptation between vehicle speed prediction results and the driver's actual driving behavior, effectively improving the temporal coherence and behavioral fit of vehicle speed prediction in complex mountainous scenarios. By determining the static disturbance compatibility index and the dynamic disturbance compatibility index, we can obtain the quantitative adaptation index of the static fixed environment and dynamic moving disturbance on the vehicle speed constraint in mountainous roads, as well as the environmental disturbance classification and control benchmark. This enables the dynamic-static separation analysis of multi-source disturbance information on the roadside in mountainous areas and the accurate quantification of the vehicle speed constraint effect. It makes up for the technical defects of traditional vehicle speed prediction, such as the difficulty in quantifying roadside environmental constraints and the coupling interference of dynamic and static disturbances. It provides an exogenous constraint benchmark for car-following speed registration with environmental adaptation, and achieves a high degree of adaptation between the vehicle speed prediction results and the real-time road environment in mountainous areas. This effectively improves the environmental anti-interference and spatial adaptability of vehicle speed prediction in complex scenarios.

[0090] In summary, the technical solution adopted in this application can register and guide the following speed of the current driving behavior in complex road scenarios, thereby improving the accuracy of vehicle speed prediction in mountainous areas.

[0091] Example 2: This application provides a new energy vehicle speed prediction system, referring to... Figure 5 As shown in the figure, this is a module structure diagram of a new energy vehicle speed prediction system according to this embodiment of the present application. The speed prediction system includes:

[0092] The data acquisition module 100 is used to collect driving disturbance source data of new energy vehicles in mountainous road environments, and to perform intent recognition on the driving disturbance source data to obtain the driver's driving intent label during driving.

[0093] The behavior decision module 200 is used to determine the behavior evolution law of the target driving state in behavior identification and the driving trajectory anchor point when the driving action feedback is based on the driving intention label, and then determine the linkage decision elements of the current driving speed in the target driving behavior based on the behavior evolution law and the driving trajectory anchor point.

[0094] The waypoint identification module 300 is used to acquire roadside intrusion information of new energy vehicles driving in mountainous road environments, perform waypoint identification on the roadside intrusion information, obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section, and then determine the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed by the static disturbance compatibility index and the dynamic disturbance compatibility index.

[0095] The registration guidance module 400 is used to register and guide the following speed of the current driving behavior in a mountainous scene based on the linkage decision elements and the homomorphic fitting level.

[0096] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0098] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for predicting the speed of a new energy vehicle, characterized in that, The vehicle speed prediction method includes the following steps: Data on driving disturbance sources of new energy vehicles in mountainous road environments are collected, and the driving disturbance source data is used to identify the driving intentions of the driver during driving to obtain the driving intention labels. Based on the driving intention label, the behavioral evolution law of the target driving state in behavior identification and the driving trajectory anchor point when driving action feedback are determined, and then the linkage decision elements of the current driving speed in the target driving behavior are determined by the behavioral evolution law and the driving trajectory anchor point. The roadside intrusion information of new energy vehicles in mountainous road environments is obtained, and the roadside intrusion information is used to identify waypoints to obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section. Then, the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed is determined by the static disturbance compatibility index and the dynamic disturbance compatibility index. Based on the aforementioned linkage decision-making elements and the aforementioned homomorphic fitting hierarchy, the following speed of the current driving behavior in a mountainous scenario is registered and guided.

2. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The driving disturbance data refers to the data set of new energy vehicles themselves and their relationship with the environment that affect drivers' decisions when driving in mountainous areas.

3. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, Intent identification is performed on the driving disturbance source data to obtain the driver's driving intent label during driving, specifically including: The driving response characteristics of drivers when driving on mountain roads are analyzed from the driving disturbance source data; Based on the driving response characteristics, the driving disturbance source data is compared and mapped to extract the driver's driving decision sequence during driving; The driving decision sequence is used to determine the driver's driving intention label during driving.

4. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, Determining the behavioral evolution pattern of the target driving state in behavior recognition and the driving trajectory anchor points during driving action feedback based on the driving intention label specifically includes: Based on the driving intent label, the behavioral state attribute of the target driving state in behavior identification is analyzed; The behavioral evolution law of the target driving state in behavior identification is determined based on the aforementioned behavioral state attributes; Based on the aforementioned behavioral state attributes, the evolution process of driving behavior is collaboratively calibrated to obtain the driving trajectory anchor point at the time of driving action feedback.

5. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The driving trajectory anchor point refers to the point that provides spatial positioning for driving action feedback.

6. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The aforementioned linkage decision-making elements refer to the decision-making criteria for the current driving speed in the target driving behavior.

7. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The roadside intrusion information refers to the set of static and dynamic interference data on both sides of mountain roads that affect the driving safety of new energy vehicles.

8. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The noise compatibility index refers to the degree to which a fixed road environment adapts to the current driving speed.

9. The method for predicting the speed of a new energy vehicle as described in claim 1, characterized in that, The disturbance compatibility index refers to the degree to which temporary and moving targets adapt to the current driving speed.

10. A new energy vehicle speed prediction system, used to execute a new energy vehicle speed prediction method as described in any one of claims 1 to 9, characterized in that, The vehicle speed prediction system includes: The data acquisition module is used to collect driving disturbance source data of new energy vehicles in mountainous road environments, and to perform intent recognition on the driving disturbance source data to obtain the driver's driving intent label during driving. The behavior decision module is used to determine the behavior evolution law of the target driving state in behavior identification and the driving trajectory anchor point when the driving action feedback is based on the driving intention label, and then determine the linkage decision elements of the current driving speed in the target driving behavior based on the behavior evolution law and the driving trajectory anchor point. The waypoint identification module is used to acquire roadside intrusion information of new energy vehicles driving in mountainous road environments, perform waypoint identification on the roadside intrusion information, obtain the static disturbance compatibility index and dynamic disturbance compatibility index of driving speed in the driving behavior feature section, and then determine the homomorphic fitting level of the current driving behavior in the mountainous vehicle speed by the static disturbance compatibility index and the dynamic disturbance compatibility index. The registration guidance module is used to register and guide the following speed of the current driving behavior in a mountainous scene based on the linkage decision elements and the homomorphic fitting level.

Images