Electric automobile motor driving device and control method

Abstract

The invention relates to the technical field of vehicle control, in particular to an electric vehicle motor driving device and a control method. A road surface image acquisition module acquires a current driving road surface image through a camera on a vehicle; the road surface icing grade judgment module is used for comparing the walking road surface images based on the image feature database so as to judge the grade of an iced road surface; the turning radius calculation module obtains the nearest turning radius data in front of the vehicle; the power parameter matching module adjusts power parameters of the two motors based on the iced road surface grade, the turning radius data and the vehicle data; the torque distribution parameter module adjusts the torque distribution parameters of the two motors based on the inclination gradient of the front road surface; and the display early warning module displays the current pavement grade evaluation result and the corresponding driving torque distribution state, and carries out early warning on the danger grade. The driving force of the driving system can be better distributed and controlled according to the road condition, so that the driving stability is improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicle control technology, and in particular to an electric vehicle motor drive device and a control method. Background Art

[0002] Electric vehicles do face a unique set of challenges when driving on icy roads, which not only affect the basic performance and driving safety of the vehicle, but also have an adverse effect on the efficiency of the battery. When the temperature drops below freezing, especially on icy or snow-covered roads, the friction between the tires and the ground is greatly reduced, making the vehicle more prone to slipping, making driving more difficult. Summary of the invention

[0003] The object of the present invention is to provide an electric vehicle motor drive device and a control method, so as to better distribute and control the driving force of the drive system according to the road conditions, thereby improving driving stability.

[0004] To achieve the above-mentioned object, in a first aspect, the present invention provides an electric vehicle motor drive device, comprising a road surface image acquisition module, a road surface icing level judgment module, a turning radius calculation module, a power parameter matching module, a torque distribution parameter module and a display warning module;

[0005] The road surface image acquisition module is used to acquire the current driving road surface image through the camera on the vehicle;

[0006] The road icing grade judgment module is used to compare the road surface image based on the image feature database to judge the icy road surface grade;

[0007] The turning radius calculation module is used to obtain the nearest turning radius data in front of the vehicle;

[0008] The power parameter matching module is used to adjust the power parameters of the two motors based on the icy road grade, turning radius data and vehicle data;

[0009] The torque distribution parameter module is used to adjust the torque distribution parameters of the two motors based on the inclination gradient of the road ahead;

[0010] The display warning module is used to display the current road surface grade assessment result and the corresponding driving torque distribution state, and to issue a warning on the danger level.

[0011] Wherein, the road surface image acquisition module includes a camera parameter setting unit, an image acquisition unit and an image processing unit;

[0012] The camera parameter setting unit is used to set the camera parameters;

[0013] The image acquisition unit is used to acquire a road surface image based on the camera parameters;

[0014] The image processing unit is used to pre-process the road surface image to obtain a processed image.

[0015] Wherein, the road icing grade judgment module includes a data annotation unit, a feature extraction unit, a similarity calculation unit and a classification unit;

[0016] The data annotation unit is used to collect road surface images under different conditions as training data sets;

[0017] The feature extraction unit is used to mark the icing level of each image to obtain a labeled data set;

[0018] The similarity calculation unit is used to extract a feature vector from the labeled data set using the SIFT method;

[0019] The classification unit is used to calculate the cosine similarity between the feature vector of the new image and the feature vector already in the database;

[0020] New images are classified based on cosine similarity to determine the icing level they belong to.

[0021] Wherein, the feature extraction unit includes a conversion subunit, an extreme value calculation subunit, a filtering subunit, and a histogram calculation subunit;

[0022] The conversion subunit is used to convert the processed image into a grayscale image;

[0023] The extreme value calculation subunit is used to generate images of different resolutions by performing blurring and smoothing processing on the grayscale image multiple times, and calculate the difference between each pair of Gaussian images of adjacent scales to obtain a set of second images;

[0024] The filtering subunit is used to find a local maximum or minimum value in each second image to obtain a key point group;

[0025] The histogram calculation subunit is used to apply a threshold to filter out the key point group to obtain feature points, then calculate the gradient direction histogram of the surrounding area of ​​each feature point, and determine the direction of the key point according to the histogram peak to obtain a feature vector.

[0026] Wherein, the similarity calculation unit includes a dot product calculation subunit, a modulus length product subunit and a similarity calculation subunit;

[0027] The dot product calculation subunit is used to calculate the dot product between two vectors;

[0028] The module length product subunit is used to calculate the module length product of two vectors;

[0029] The similarity calculation subunit is used to obtain cosine similarity by dividing the dot product by the module length product.

[0030] Wherein, the turning radius calculation module includes a position acquisition unit, a turning image query unit and a radius calculation unit;

[0031] The position acquisition unit is used to acquire the current position information of the vehicle;

[0032] The turning image query unit is used to query the nearest turning image in the map based on the position information;

[0033] The radius calculation unit is used to calculate turning radius data based on the turning image.

[0034] Wherein, the power parameter matching module includes a weight setting unit, a risk score calculation unit and a power parameter calculation unit;

[0035] The weight setting unit is used to calculate the influence weight of each input parameter on the slipperiness by a machine learning method;

[0036] The risk score calculation unit is used to calculate the risk score based on the input parameters and the impact weight;

[0037] The power parameter calculation unit is used to match the corresponding power parameter based on the risk score.

[0038] Wherein, the torque distribution parameter module includes a slope acquisition unit, a center of gravity calculation unit and a torque adjustment unit;

[0039] The slope acquisition unit is used to monitor the slope value in front of the vehicle in real time using a vehicle-mounted sensor;

[0040] The center of gravity calculation unit is used to calculate the center of gravity value of the vehicle in real time based on the slope value;

[0041] The torque adjustment unit is used to adjust the torque ratio of the front and rear motors based on the center of gravity value of the vehicle.

[0042] In a second aspect, the present invention further provides an electric vehicle motor drive control method, comprising:

[0043] Acquire the current road surface image through the camera on the vehicle;

[0044] Compare the road surface image based on the image feature database to determine the icy road surface grade;

[0045] Obtain the nearest turning radius data ahead of the vehicle;

[0046] Adjust the power parameters of the two motors based on icy road grade, turning radius data and vehicle data;

[0047] Adjust the torque distribution parameters of the two motors based on the inclination of the road ahead;

[0048] Displays the current road grade assessment results and the corresponding drive torque distribution status, and issues warnings on the danger level.

[0049] In an electric vehicle motor drive device and control method of the present invention, a road image acquisition module is equipped with a high-precision camera to capture real-time road images in front of and around the vehicle. Through these image information, the specific conditions of the current driving road surface can be understood, providing basic data for subsequent analysis. The road icing level judgment module can automatically identify and evaluate the degree of icing on the road surface based on a machine learning algorithm and a pre-stored image feature database. In order to ensure the safety of the vehicle when turning, the turning radius calculation module predicts and calculates the curvature of the closest turning point that the vehicle is about to encounter. By combining GPS positioning, map information and sensor data, the radius of each curve can be accurately estimated to help the system better plan the driving path. According to the road icing level, the expected turning radius and the vehicle's own performance parameters (such as speed, acceleration, etc.), the power parameter matching module dynamically adjusts the power distribution between the front and rear or left and right drive motors. The purpose is to maintain optimal traction and stability, and to ensure good handling performance even in severe weather conditions. Considering that different road slopes will affect vehicle driving, especially when the road surface is inclined, the torque distribution parameter module senses the slope change in front of the vehicle and adjusts the driving torque ratio between the two wheels accordingly. This not only improves the climbing ability, but also prevents skidding when going downhill. The display warning module not only intuitively shows the driver the current road conditions (including ice level) and power distribution, but also issues different levels of alarm prompts based on the system's comprehensive evaluation. If potential dangers are detected, such as severe icy sections or sharp curves, the driver will be reminded to take appropriate measures in time to avoid accidents.

[0050] Through highly sensitive capture and intelligent processing of environmental information, effective control of vehicle behavior is achieved, significantly improving driving safety, especially in winter when roads are covered with ice and snow. In addition, as technology develops, the system also reserves interfaces to facilitate future upgrades and expansion of more functions, such as interconnection and collaboration with other intelligent transportation systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 It is a structural diagram of an electric vehicle motor drive device of the present invention.

[0053] Figure 2 It is a structural diagram of the road surface image acquisition module of the present invention.

[0054] Figure 3 It is a structural diagram of the road icing grade judgment module of the present invention.

[0055] Figure 4 It is a structural diagram of the feature extraction unit of the present invention.

[0056] Figure 5 It is a structural diagram of the similarity calculation unit of the present invention.

[0057] Figure 6 It is a structural diagram of the turning radius calculation module of the present invention.

[0058] Figure 7 It is a structural diagram of the power parameter matching module of the present invention.

[0059] Figure 8 It is a structural diagram of the torque distribution parameter module of the present invention.

[0060] Road surface image acquisition module 101, road surface icing level judgment module 102, turning radius calculation module 103, power parameter matching module 104, torque allocation parameter module 105, display warning module 106, camera parameter setting unit 107, image acquisition unit 108, image processing unit 109, data annotation unit 110, feature extraction unit 111, similarity calculation unit 112, classification unit 113, conversion subunit 114, extreme value calculation subunit 115, filtering subunit 116, histogram calculation subunit 117, dot product calculation subunit 118, module length product subunit 119, similarity calculation subunit 120, position acquisition unit 121, turning image query unit 122, radius calculation unit 123, weight setting unit 124, risk score calculation unit 125, power parameter calculation unit 126, slope acquisition unit 127, center of gravity calculation unit 128, torque adjustment unit 129. DETAILED DESCRIPTION

[0061] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0062] First embodiment

[0063] See also Figures 1 to 8 The present invention provides an electric vehicle motor drive device, comprising a road surface image acquisition module 101, a road surface icing level judgment module 102, a turning radius calculation module 103, a power parameter matching module 104, a torque distribution parameter module 105 and a display warning module 106; the road surface image acquisition module 101 is used to obtain the current driving road surface image through a camera on the vehicle; the road surface icing level judgment module 102 is used to compare the driving road surface image based on an image feature database to judge the icy road surface level; the turning radius calculation module 103 is used to obtain the nearest turning radius data in front of the vehicle; the power parameter matching module 104 is used to adjust the power parameters of the two motors based on the icy road surface level, the turning radius data and the vehicle data; the torque distribution parameter module 105 is used to adjust the torque distribution parameters of the two motors based on the inclination slope of the road surface in front; the display warning module 106 is used to display the current road surface level evaluation result and the corresponding driving torque distribution state, and to warn of the danger level.

[0064] In this embodiment, the road image acquisition module 101 is equipped with a high-precision camera to capture real-time road images in front of and around the vehicle. Through these image information, the specific conditions of the current road surface can be understood, providing basic data for subsequent analysis. The road icing level judgment module 102 can automatically identify and evaluate the degree of icing on the road surface based on a machine learning algorithm and a pre-stored image feature database. In order to ensure the safety of the vehicle when turning, the turning radius calculation module 103 predicts and calculates the curvature of the closest turning point that the vehicle is about to encounter. By combining GPS positioning, map information and sensor data, the radius of each curve can be accurately estimated to help the system better plan the driving path. According to the road icing level, the expected turning radius and the vehicle's own performance parameters (such as speed, acceleration, etc.), the power parameter matching module 104 dynamically adjusts the power distribution between the front and rear or left and right drive motors. The purpose is to maintain optimal traction and stability, and to ensure good handling performance even in severe weather conditions. Considering that different road slopes will affect vehicle driving, especially when the road surface is inclined, the torque distribution parameter module 105 senses the slope change in front of the vehicle and adjusts the driving torque ratio between the two wheels accordingly. This can not only improve the climbing ability, but also prevent slipping when going downhill. The display warning module 106 not only intuitively displays the current road conditions (including ice level) and power distribution to the driver, but also issues different levels of alarm prompts based on the comprehensive evaluation of the system. If potential dangers are detected, such as severe icy sections or sharp curves, the driver will be reminded in time to take appropriate measures to avoid accidents.

[0065] Through highly sensitive capture and intelligent processing of environmental information, effective control of vehicle behavior is achieved, significantly improving driving safety, especially in winter when roads are covered with ice and snow. In addition, as technology develops, the system also reserves interfaces to facilitate future upgrades and expansion of more functions, such as interconnection and collaboration with other intelligent transportation systems.

[0066] The road surface image acquisition module 101 includes a camera parameter setting unit 107, an image acquisition unit 108 and an image processing unit 109; the camera parameter setting unit 107 is used to set camera parameters; the image acquisition unit 108 is used to acquire a road surface image based on the camera parameters; the image processing unit 109 is used to pre-process the road surface image to obtain a processed image.

[0067] The camera parameter setting unit 107 configures various parameters of the camera to adapt to different driving environments and lighting conditions. Its functions include but are not limited to adjusting key parameters such as resolution, frame rate, exposure time, white balance, focal length, etc. These settings are crucial to ensuring the quality of the collected images, especially in low-light or high-contrast environments, where appropriate camera parameters can significantly improve image clarity and detail expression.

[0068] The image acquisition unit 108 performs the actual image capture task based on the optimal parameter configuration determined by the camera parameter setting unit 107. It uses one or more high-definition cameras installed on the vehicle to continuously take real-scene photos or video streams of the road ahead at set time intervals.

[0069] The image processing unit 109 performs necessary preprocessing operations on the original image in order to provide a cleaner and more readable data source for subsequent analysis and judgment. The preprocessing steps include denoising, contrast enhancement, color correction, edge detection, size adjustment, etc., all of which are intended to remove unnecessary interference factors and highlight key features.

[0070] The road icing level judgment module 102 includes a data annotation unit 110, a feature extraction unit 111, a similarity calculation unit 112 and a classification unit 113; the data annotation unit 110 is used to collect road images under different conditions as a training data set; the feature extraction unit 111 is used to annotate the icing level of each image to obtain a labeled data set; the similarity calculation unit 112 is used to extract a feature vector from the labeled data set using the SIFT method; the classification unit 113 is used to calculate the cosine similarity between the feature vector of a new image and the feature vector already in the database; the new image is classified according to the cosine similarity to determine the icing level to which it belongs.

[0071] The data annotation unit 110 constructs a large and diverse training data set. This process involves extensively collecting road images under different weather conditions, lighting environments, and geographic locations. To ensure the quality and representativeness of the data set, the collected images not only cover various types of road surfaces (such as asphalt roads, concrete roads, etc.), but also include various degrees of icing conditions, from slightly wet to completely frozen. Each image is manually or semi-automatically annotated by professionals to record its actual icing level and other relevant information to form a labeled data set.

[0072] Based on the labeled data set prepared by the data annotation unit 110, the feature extraction unit 111 performs in-depth analysis on each image to identify key visual features that can reflect the characteristics of icing. This process involves image processing techniques and algorithms, such as using convolutional neural networks (CNNs) to automatically capture texture, color distribution, and shape patterns in images.

[0073] In order to effectively compare the similarity between the newly acquired image and the existing data, the similarity calculation unit 112 uses the scale-invariant feature transform (SIFT) method. SIFT is a powerful computer vision technology that can stably detect and describe local feature points at different scales and rotation angles. By applying SIFT, this unit can extract a set of representative feature vectors from the labeled data set, which encode important geometric and photometric information in the image. When a new road image is received, its feature vector is also extracted, and the degree of match between it and the known samples in the database is calculated by metrics such as cosine similarity.

[0074] The classification unit 113 uses the similarity results calculated above to classify the icing level of the new image. It first compares the feature vector of the new image with all feature vectors in the database based on the cosine similarity value to find the closest one or more groups of samples. Then, based on the average or weighted average icing level of these samples, it predicts the specific category to which the new image belongs. In order to improve the classification accuracy, the unit can also combine other auxiliary information, such as temperature, humidity sensor readings, and even historical data trends, and make a final judgment by comprehensively considering multiple factors.

[0075] The feature extraction unit 111 includes a conversion subunit 114, an extreme value calculation subunit 115, a filtering subunit 116, and a histogram calculation subunit 117; the conversion subunit 114 is used to convert the processed image into a grayscale image; the extreme value calculation subunit 115 is used to generate images of different resolutions by blurring and smoothing the grayscale image for multiple times, and calculate the difference between each pair of Gaussian images of adjacent scales to obtain a group of second images; the filtering subunit 116 is used to find a local maximum or minimum value in each second image to obtain a key point group;

[0076] The histogram calculation subunit 117 is used to apply a threshold to filter out the key point group to obtain feature points, then calculate the gradient direction histogram of the surrounding area of ​​each feature point, and determine the direction of the key point according to the histogram peak to obtain a feature vector.

[0077] The conversion subunit 114 first converts the color processed image into a grayscale image. This step simplifies the image data and makes subsequent processing more efficient. Since the grayscale image only contains brightness information and eliminates color interference, it is more effective in detecting texture and other brightness-based features.

[0078] The extreme value calculation subunit 115 performs multiple blurring and smoothing processes on the grayscale image to generate a series of images with different resolutions. This process usually involves constructing a Gaussian pyramid, in which each layer represents an image at a specific scale. For each pair of Gaussian images of adjacent scales, the difference between them is calculated to obtain a "Difference of Gaussians" (DoG). DoG images can highlight areas that change significantly at different scales, which often correspond to important structures or feature points in the image.

[0079] In each second image, the filtering subunit 116 searches for local maxima or minima to identify potential keypoint groups. These keypoints are usually stable and unique locations in the image, such as corners, edges or other significant features.

[0080] The histogram calculation subunit 117 applies a threshold to filter out key points that do not meet the conditions, ensuring that only the most representative feature points are retained. For each filtered feature point, the gradient direction histogram in its surrounding neighborhood is calculated. This involves evaluating the directional changes of each pixel relative to its neighbors and quantizing these changes into a number of discrete directional intervals. In this way, the texture and shape information around each feature point can be captured.

[0081] The similarity calculation unit 112 includes a dot product calculation subunit 118, a modulus product subunit 119 and a similarity calculation subunit 120; the dot product calculation subunit 118 is used to calculate the dot product between two vectors; the modulus product subunit 119 is used to calculate the modulus product of two vectors; the similarity calculation subunit 120 is used to obtain cosine similarity by dividing the dot product by the modulus product.

[0082] The dot product calculation subunit 118 performs a dot product (or inner product) operation on a pair of feature vectors. The dot product is a mathematical tool to measure the similarity of two vectors in direction. Specifically, if the directions of the two vectors are very close, their dot product value is large; conversely, if the directions of the two vectors are very different, the dot product value is small. Through the dot product operation, the complex feature vector in the original high-dimensional space can be simplified into a single value, which can reflect the extent to which the two vectors point in the same direction.

[0083] The modulus product subunit 119 calculates the modulus (i.e., the length of the vector) of each of the two feature vectors involved in the comparison, and then multiplies the two moduli. The modulus is a measure of the size of the vector, reflecting the overall size of the vector in its dimension.

[0084] The similarity calculation subunit 120 calculates the cosine similarity between two vectors using the result of the dot product and the modulus product.

[0085] The turning radius calculation module 103 includes a position acquisition unit 121, a turning image query unit 122 and a radius calculation unit 123; the position acquisition unit 121 is used to obtain the current position information of the vehicle; the turning image query unit 122 is used to query the nearest turning image in the map based on the position information; the radius calculation unit 123 is used to calculate the turning radius data based on the turning image.

[0086] The location acquisition unit 121 uses a global positioning system (GPS), an inertial measurement unit (IMU), and other auxiliary positioning technologies, such as differential GPS or cellular network positioning, to determine the current precise geographic location of the vehicle.

[0087] Based on the current position information provided by the position acquisition unit 121 , the turning image query unit 122 searches for the nearest curved road segment in a pre-constructed map database.

[0088] Once a suitable turning image is obtained, the radius calculation unit 123 starts to perform geometric analysis on the curve in the image. It identifies and extracts key geometric features of the curve, such as the center line, inner and outer boundaries, etc., and builds a mathematical model based on these features. This process usually involves techniques such as curve fitting and arc detection, with the goal of describing the shape of the curve with a concise mathematical expression.

[0089] The power parameter matching module 104 includes a weight setting unit 124, a risk score calculation unit 125 and a power parameter calculation unit 126; the weight setting unit 124 is used to calculate the influence weight of each input parameter on the slipperiness through a machine learning method; the risk score calculation unit 125 is used to calculate the risk score based on the input parameters and the influence weight; the power parameter calculation unit 126 is used to match the corresponding power parameter based on the risk score.

[0090] The weight setting unit 124 uses a machine learning method to evaluate the influence of each input parameter on the slipperiness and assigns corresponding weights to them. To this end, the unit first needs to build a data set containing a large amount of historical data, which covers actual slip event records under various variables such as different weather conditions, road surface types (such as dry, slippery, icy), vehicle speed, acceleration, steering angle, etc.

[0091] Based on the above data set, the weight setting unit 124 applies a variety of machine learning algorithms (such as random forest, gradient boosting tree, neural network, etc.) for training to identify which factors are most likely to predict the slip. For example, some algorithms can automatically extract several key features that have the greatest impact on slip and quantify their relative importance. In addition, the generalization ability and reliability of the model can also be improved through cross-validation and other techniques.

[0092] The risk score calculation unit 125 receives the influence weights from the weight setting unit 124 and various input parameters of the current vehicle and environment, including but not limited to the road friction coefficient, vehicle speed, acceleration, steering angle, tire pressure, etc. It multiplies these parameters by their corresponding influence weights and then summarizes them to obtain an overall risk score. This score reflects the probability of a skidding accident under the current conditions.

[0093] Based on the risk score provided by the risk score calculation unit 125, the power parameter calculation unit 126 is responsible for determining the optimal motor output power. Specifically, it selects different power matching strategies according to the score. For low risk scores, the system allows higher power output to ensure the acceleration performance of the vehicle; for high risk scores, the power will be reduced accordingly, limiting the maximum speed and acceleration, thereby reducing the risk of skidding.

[0094] The torque distribution parameter module 105 includes a slope acquisition unit 127, a center of gravity calculation unit 128 and a torque adjustment unit 129;

[0095] The slope acquisition unit 127 is used to monitor the slope value in front of the vehicle in real time using the vehicle-mounted sensor; the center of gravity calculation unit 128 is used to calculate the center of gravity value of the vehicle in real time based on the slope value; the torque adjustment unit 129 is used to adjust the torque ratio of the front and rear motors based on the center of gravity value of the vehicle.

[0096] The slope acquisition unit 127 uses a variety of on-board sensors (such as inertial measurement unit IMU, laser radar LiDAR, camera, etc.) to monitor the slope value of the road in front of the vehicle in real time. IMU can provide high-precision posture information, including pitch angle (i.e. longitudinal tilt angle), which is a key parameter for judging the slope. LiDAR and cameras can capture the three-dimensional structure of the road surface and assist in identifying subtle slope changes. In addition to directly measuring the current slope, the slope acquisition unit 127 also combines advanced environmental perception technologies, such as deep learning models, to predict upcoming slope changes. For example, when approaching a tunnel or bridge, the system can estimate the slope trend in advance based on historical data and map information, and make corresponding preparations.

[0097] Based on the real-time slope value provided by the slope acquisition unit 127, the center of gravity calculation unit 128 dynamically calculates the center of gravity position of the vehicle. When the vehicle is traveling on an inclined road, the center of gravity calculation unit 128 pays special attention to the effect of the slope on the center of gravity height and the load distribution of the front and rear axles. It adjusts the relevant parameters in the center of gravity calculation formula according to the slope direction (uphill or downhill) and the slope size to ensure that the obtained result is accurate.

[0098] The torque adjustment unit 129 intelligently adjusts the torque ratio of the front and rear motors according to the center of gravity value provided by the center of gravity calculation unit 128. Specifically, when the vehicle is on an uphill slope, the front motor will be given a larger torque to provide stronger traction; while when going downhill, the output of the front motor is reduced, and more braking control is relied on the rear motor to prevent the front of the vehicle from sinking too much.

[0099] Second embodiment

[0100] The present invention also provides an electric vehicle motor drive control method, comprising:

[0101] S201 obtains a current driving road image through a camera on the vehicle;

[0102] Use multiple high-definition cameras installed in different locations of the vehicle (such as the front windshield, side mirrors, etc.) to capture real-time road images in front of and around the vehicle. Ensure that the cameras have sufficient resolution and viewing angle to provide detailed and comprehensive road information, even at high speeds or on narrow roads.

[0103] S202 compares the road surface image based on the image feature database to determine the icy road surface grade;

[0104] Apply deep learning algorithms, such as convolutional neural networks (CNNs), to build a powerful image recognition model that can distinguish different levels of icy surfaces. Use a large number of labeled road image datasets for model training, which cover various types of roads (asphalt, cement, etc.) and different degrees of icing. When receiving a new road image, the system will quickly compare it with the pre-stored image feature database to determine the most similar sample and output the icing level assessment result accordingly.

[0105] S203 obtains the nearest turning radius data ahead of the vehicle;

[0106] Combined with GPS positioning and electronic map information, the current position of the vehicle is accurately matched with the position of the upcoming curve, thereby predicting the nearest turning radius. Sensors such as LiDAR and cameras are used to further verify and refine the turning radius data, especially in the absence of detailed map information, to ensure the accuracy and reliability of the data.

[0107] As the vehicle moves forward, information about the road ahead is continuously updated, maintaining the latest turning radius estimate.

[0108] S204 adjusts power parameters of the two motors based on the icy road grade, the turning radius data, and the vehicle data;

[0109] Taking into account factors such as ice level, turning radius, vehicle speed, acceleration, etc., the power output of the front and rear motors is dynamically adjusted to ensure optimal traction and stability. Customized power adjustment strategies are formulated according to different scenarios, such as allowing higher power output in low-risk situations and appropriately reducing power in high-risk situations. By monitoring vehicle behavior in real time, such as whether there are signs of slight slippage, power parameters are adjusted in a timely manner to form a closed-loop control system.

[0110] S205 adjusts torque distribution parameters of the two motors based on the inclination of the road ahead;

[0111] The inertial measurement unit (IMU) and other sensors are used to monitor the slope changes of the road ahead in real time, especially when going uphill or downhill, to ensure that the torque distribution meets the current needs. The center of gravity position of the vehicle is calculated based on the slope information, and then the torque ratio between the front and rear motors is adjusted to prevent safety hazards caused by the center of gravity shift. Torque vectoring is implemented to improve steering response and grip by fine-tuning the torque difference between the left and right motors, thereby improving handling performance.

[0112] S206 displays the current road surface grade assessment result and the corresponding driving torque distribution status, and issues a warning of the danger level.

[0113] Develop an intuitive and easy-to-understand human-machine interface (HMI) to graphically display the road surface grade assessment results, drive torque distribution status and other key information. Set different warning levels (such as low, medium and high) according to the assessed risk level, and remind the driver of potential dangers through visual (such as warning lights) and auditory (such as voice prompts).

[0114] In summary, this electric vehicle motor drive control method not only improves the safety of the vehicle in severe weather and complex road conditions, but also provides the driver with a more secure and comfortable driving experience through intelligent perception, precise analysis and targeted adjustment of power parameters of road conditions.

[0115] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. An electric vehicle motor drive device, characterized in that: It includes a road image acquisition module, a road icing level judgment module, a turning radius calculation module, a power parameter matching module, a torque distribution parameter module and a display warning module; The road surface image acquisition module is used to acquire the current driving road surface image through the camera on the vehicle; The road icing grade judgment module is used to compare the road surface image based on the image feature database to judge the icy road surface grade; The turning radius calculation module is used to obtain the nearest turning radius data in front of the vehicle; The power parameter matching module is used to adjust the power parameters of the two motors based on the icy road grade, turning radius data and vehicle data; The torque distribution parameter module is used to adjust the torque distribution parameters of the two motors based on the inclination gradient of the road ahead; The display warning module is used to display the current road surface grade assessment result and the corresponding driving torque distribution state, and to issue a warning on the danger level.

2. The electric vehicle motor drive device according to claim 1, characterized in that: The road surface image acquisition module includes a camera parameter setting unit, an image acquisition unit and an image processing unit; The camera parameter setting unit is used to set the camera parameters; The image acquisition unit is used to acquire a road surface image based on the camera parameters; The image processing unit is used to pre-process the road surface image to obtain a processed image.

3. The electric vehicle motor drive device according to claim 2, characterized in that: The road icing grade judgment module includes a data annotation unit, a feature extraction unit, a similarity calculation unit and a classification unit; The data annotation unit is used to collect road surface images under different conditions as training data sets; The feature extraction unit is used to mark the icing level of each image to obtain a labeled data set; The similarity calculation unit is used to extract a feature vector from the labeled data set using the SIFT method; The classification unit is used to calculate the cosine similarity between the feature vector of the new image and the feature vector already in the database, and classify the new image according to the cosine similarity to determine the icing level to which the new image belongs.

4. The electric vehicle motor drive device according to claim 3, characterized in that: The feature extraction unit includes a conversion subunit, an extreme value calculation subunit, a filtering subunit, and a histogram calculation subunit; The conversion subunit is used to convert the processed image into a grayscale image; The extreme value calculation subunit is used to generate images of different resolutions by performing blurring and smoothing processing on the grayscale image multiple times, and calculate the difference between each pair of Gaussian images of adjacent scales to obtain a set of second images; The filtering subunit is used to find a local maximum or minimum value in each second image to obtain a key point group; The histogram calculation subunit is used to apply a threshold to filter out the key point group to obtain feature points, then calculate the gradient direction histogram of the surrounding area of ​​each feature point, and determine the direction of the key point according to the histogram peak value to obtain a feature vector.

5. The electric vehicle motor drive device according to claim 4, characterized in that: The similarity calculation unit includes a dot product calculation subunit, a modulus length product subunit and a similarity calculation subunit; The dot product calculation subunit is used to calculate the dot product between two vectors; The module length product subunit is used to calculate the module length product of two vectors; The similarity calculation subunit is used to obtain cosine similarity by dividing the dot product by the module length product.

6. The electric vehicle motor drive device according to claim 5, characterized in that: The turning radius calculation module includes a position acquisition unit, a turning image query unit and a radius calculation unit; The position acquisition unit is used to acquire the current position information of the vehicle; The turning image query unit is used to query the nearest turning image in the map based on the position information; The radius calculation unit is used to calculate turning radius data based on the turning image.

7. The electric vehicle motor drive device according to claim 6, characterized in that: The power parameter matching module includes a weight setting unit, a risk score calculation unit and a power parameter calculation unit; The weight setting unit is used to calculate the influence weight of each input parameter on the slipperiness by a machine learning method; The risk score calculation unit is used to calculate the risk score based on the input parameters and the impact weight; The power parameter calculation unit is used to match the corresponding power parameter based on the risk score.

8. The electric vehicle motor drive device according to claim 7, characterized in that: The torque distribution parameter module includes a slope acquisition unit, a center of gravity calculation unit and a torque adjustment unit; The slope acquisition unit is used to monitor the slope value in front of the vehicle in real time using a vehicle-mounted sensor; The center of gravity calculation unit is used to calculate the center of gravity value of the vehicle in real time based on the slope value; The torque adjustment unit is used to adjust the torque ratio of the front and rear motors based on the center of gravity value of the vehicle.

9. A method for controlling a motor drive of an electric vehicle, using a motor drive device of an electric vehicle according to any one of claims 1 to 8, characterized in that: include: Acquire the current road surface image through the camera on the vehicle; Compare the road surface image based on the image feature database to determine the icy road surface grade; Obtain the nearest turning radius data ahead of the vehicle; Adjust the power parameters of the two motors based on icy road grade, turning radius data and vehicle data; Adjust the torque distribution parameters of the two motors based on the inclination of the road ahead; Displays the current road surface grade assessment results and the corresponding drive torque distribution status, and issues warnings on the danger level.

Images