An automatic tracking control method for an engineering vehicle

Abstract

The application discloses an automatic tracking control method for engineering vehicles, and relates to the field of engineering vehicle control, which comprises collecting real-time data of the engineering vehicle, and constructing an engineering vehicle operation characteristic model by using a deep learning algorithm; inputting the real-time data into the engineering vehicle operation characteristic model, outputting dynamic behavior indexes, and performing segmented processing on the driving path of the engineering vehicle; based on the segmented processing result, calculating the posture angle and the road surface condition of the engineering vehicle through multi-sensor data fusion, and establishing an engineering vehicle trajectory prediction model; predicting and planning the driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model, and performing collaborative control on the engineering vehicle system through a vehicle-mounted controller to complete automatic tracking control. The application not only improves the precision and reliability of automatic control, but also enhances the environmental adaptability of the system under complex working conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering vehicle control, and particularly relates to an automatic tracking control method for engineering vehicles. BACKGROUND

[0002] Traditional engineering vehicle control systems mainly rely on manual operation, which is easily affected by subjective factors such as driver experience and fatigue level under complex working conditions, and it is difficult to ensure the efficiency and safety of operation. At present, scholars at home and abroad have carried out a large number of researches on automatic control technology of engineering vehicles, and have proposed various automatic tracking control schemes based on visual navigation, laser radar navigation and the like. However, these schemes still have problems such as insufficient perception accuracy, poor environmental adaptability and single control strategy in practical application, especially under complex road conditions and adverse weather conditions, the existing technology is difficult to accurately perceive the running state of the engineering vehicle and the surrounding environment information, and it is difficult to realize stable and reliable automatic tracking control.

[0003] With the rise of deep learning technology, intelligent control methods based on deep neural networks show good development prospects. However, most of the existing deep learning control methods are designed for ordinary vehicles, and the particularity of engineering vehicles such as large changes in vehicle posture, complex road conditions and dynamic changes in load is not fully considered. At the same time, the existing technology often uses a single sensor data source for control decision-making, lacks effective fusion of multi-source data, and leads to insufficient system robustness and reliability. In addition, traditional trajectory planning algorithms fail to fully utilize historical data and environmental information, making it difficult to accurately predict and optimize the driving trajectory of the engineering vehicle.

[0004] The existing automatic control technology of engineering vehicles mainly has the technical problems of limited perception ability, poor environmental adaptability and single control strategy. The present application proposes an automatic tracking control method for engineering vehicles, which constructs a running feature model through a deep learning algorithm, realizes accurate perception of the posture of the engineering vehicle and the road conditions by combining multi-sensor data fusion technology, and makes control decisions based on a trajectory prediction model, effectively solving the problems of insufficient control accuracy and reliability of the existing technology under complex working conditions. SUMMARY

[0005] In view of the problems of insufficient perception accuracy, poor environmental adaptability and single control strategy of the existing automatic tracking control system of engineering vehicles under complex working conditions, the present application is proposed.

[0006] Therefore, the problem to be solved by the present application is how to provide a method capable of accurately perceiving the running state of the engineering vehicle and adapting to complex road conditions, and realizing stable and reliable automatic tracking control.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the embodiments of the present application provide an automatic tracking control method for engineering vehicles, which comprises collecting real-time data of the engineering vehicle and constructing an engineering vehicle running feature model by using a deep learning algorithm; inputting the real-time data into the engineering vehicle running feature model, outputting dynamic behavior indexes, and performing segmented processing on the driving path of the engineering vehicle; based on the segmented processing result, calculating the attitude angle and the road surface condition of the engineering vehicle through multi-sensor data fusion, and establishing an engineering vehicle trajectory prediction model; predicting and planning the driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model, and performing collaborative control on the engineering vehicle system through a vehicle-mounted controller to complete the automatic tracking control.

[0009] As a preferred scheme of the automatic tracking control method for engineering vehicles according to the present application, the engineering vehicle system comprises an engineering vehicle steering system, a braking system and a power system; the automatic tracking control is completed by predicting and planning the driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model and performing collaborative control on the engineering vehicle system through the vehicle-mounted controller, which comprises: calculating the optimal driving trajectory of the engineering vehicle by using a recursive least squares algorithm according to the predicted trajectory data output by the engineering vehicle trajectory prediction model; generating steering angle instructions, acceleration instructions and braking instructions according to the optimal driving trajectory; sending the steering angle instructions to the electric steering system through the vehicle-mounted controller to execute steering adjustment, and sending the acceleration instructions and the braking instructions to the power system and the braking system respectively to execute power output adjustment and braking force adjustment; simultaneously collecting real-time response data of the execution of the engineering vehicle system, and performing deviation analysis on the real-time response data and the optimal driving trajectory; when the deviation is less than or equal to a second preset threshold, the current control instructions remain unchanged, and the automatic tracking control is continued to be executed; when the deviation is greater than the second preset threshold, the steering angle instructions, the acceleration instructions and the braking instructions are regenerated according to the deviation until the automatic tracking control of the engineering vehicle is completed.

[0010] As a preferred scheme of the automatic tracking control method for engineering vehicles according to the present application, the specific formula of the optimal driving trajectory is as follows:

[0011]

[0012] wherein, T opt is the optimal driving trajectory value, x i is the current position coordinate, x p (i) is the predicted trajectory coordinate, n is the prediction time domain length, λ is the forgetting factor, v is the current vehicle speed, μ is the smoothing factor, δ is the trajectory curvature, and σ is the curvature standard deviation threshold.

[0013] As a preferred scheme of the automatic tracking control method for the engineering vehicle, the construction method of the engineering vehicle trajectory prediction model is that, based on the result of the segmented processing, Kalman filtering algorithm is used for fusion operation on real-time data to obtain three-dimensional attitude angle data of the engineering vehicle; meanwhile, road surface image data obtained by the camera array is input into a pre-trained semantic segmentation network to extract road surface texture features, slope features and wetness degree features to form road surface condition data; the three-dimensional attitude angle data and the road surface condition data are input into a long short-term memory network to establish the engineering vehicle trajectory prediction model.

[0014] As a preferred scheme of the automatic tracking control method for the engineering vehicle, the method of the segmented processing is that, real-time data is input into an engineering vehicle running feature model, wherein the engineering vehicle running feature model comprises a distance prediction module, an attitude evaluation module and an environment perception module; the distance prediction module is used for calculating safety distance parameters of the engineering vehicle and surrounding environment obstacles; the attitude evaluation module is used for analyzing pitch angle parameters and roll angle parameters of the engineering vehicle; the environment perception module is used for identifying road surface type parameters and road condition complexity; the safety distance parameters, the pitch angle parameters, the roll angle parameters, the road surface type parameters and the road condition complexity are combined to form dynamic behavior indexes; according to the dynamic behavior indexes, the driving path of the engineering vehicle is divided, when the road condition complexity is greater than a first preset threshold, the corresponding road section is divided into a complex road section, and tracking control parameters are dynamically adjusted, and multi-sensor data fusion calculation is started to obtain attitude angle data and road surface condition data; when the road condition complexity is less than or equal to the first preset threshold, the corresponding road section is divided into a standard road section, and preset tracking control parameters are used for regular tracking control.

[0015] As a preferred scheme of the automatic tracking control method for the engineering vehicle, the specific formula of the dynamic behavior index is as follows:

[0016]

[0017] wherein, DBI is an output value of the dynamic behavior index, d s is a safety distance parameter, d max is a preset maximum safety distance threshold, θ p is a pitch angle parameter, θ r is a roll angle parameter, θ th is an angle threshold, R is a road surface type parameter, C is a road condition complexity, S is an environmental state feature value, and α, β and γ are adjustment coefficients.

[0018] As a preferred scheme of the automatic tracking control method for the engineering vehicle, the real-time data comprises driving track data, vehicle state data and surrounding environment image data; the engineering vehicle running feature model is established by collecting real-time data obtained by a camera array, a GPS positioning module, a wheel speed sensor and an inertial measurement unit (IMU) carried by the engineering vehicle; the real-time data is preprocessed, and the preprocessed real-time data is input into a pre-trained deep convolutional neural network to extract running feature information in the running process of the engineering vehicle; and the engineering vehicle running feature model is constructed according to the running feature information, wherein the engineering vehicle running feature model comprises a graph neural network (GNN) structure representing the running state of the engineering vehicle and the environmental interaction relationship, and the model parameters are optimized by a contrast learning method.

[0019] In a second aspect, an embodiment of the present application provides an automatic tracking control system for an engineering vehicle, comprising: a collection module configured to collect real-time data of the engineering vehicle and construct an engineering vehicle running feature model by using a deep learning algorithm; a segmented processing module configured to input the real-time data into the engineering vehicle running feature model, output dynamic behavior indicators, and perform segmented processing on a driving path of the engineering vehicle; an establishment module configured to calculate a posture angle and a road surface condition of the engineering vehicle by multi-sensor data fusion based on a segmented processing result, and establish an engineering vehicle trajectory prediction model; and a cooperative control module configured to predict and plan a driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model, and perform cooperative control on a system of the engineering vehicle by using a vehicle-mounted controller to complete automatic tracking control.

[0020] In a third aspect, an embodiment of the present application provides a computer device comprising a memory and a processor, and the memory stores a computer program, wherein the computer program instructions are executed by the processor to implement the steps of the automatic tracking control method for the engineering vehicle according to the first aspect of the present application.

[0021] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program instructions are executed by a processor to implement the steps of the automatic tracking control method for the engineering vehicle according to the first aspect of the present application.

[0022] The present application has the following beneficial effects: the running feature model is constructed by using a deep learning algorithm to realize feature extraction, the multi-sensor data fusion technology is combined to improve the perception accuracy of the posture of the engineering vehicle and the road condition, the segmented processing strategy is used to reduce the data processing complexity and optimize the calculation efficiency, the trajectory prediction model is used to realize accurate prediction of the future motion state of the vehicle, and finally the cooperative control method is used to ensure the optimized cooperation of the steering, acceleration, braking and other actuators. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings. Among them:

[0024] Figure 1 Flow chart of the automatic tracking control method for the engineering vehicle in embodiment 1. DETAILED DESCRIPTION

[0025] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0026] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0027] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is independent of or mutually exclusive of other embodiments.

[0028] Embodiment 1

[0029] Reference Figure 1 For the first embodiment of the present application, the embodiment provides an automatic tracking control method for an engineering vehicle, comprising,

[0030] S1: collecting real-time data of the engineering vehicle, and constructing an engineering vehicle running feature model by using a deep learning algorithm.

[0031] Specifically, real-time data obtained by a camera array, a GPS positioning module, a wheel speed sensor and an inertial measurement unit (IMU) carried by the engineering vehicle are collected.

[0032] It should be noted that the real-time data includes driving trajectory data, vehicle state data and surrounding environment image data. The camera array is arranged in a surround view mode, and 360-degree environment image data of the engineering vehicle is generated by a multi-view image fusion algorithm; the GPS positioning module and the wheel speed sensor cooperate to measure the geographic coordinates, driving speed and driving trajectory data of the engineering vehicle; the inertial measurement unit (IMU) records the attitude angle, acceleration and angular velocity of the vehicle state data of the engineering vehicle.

[0033] Further, the real-time data is preprocessed, and the preprocessed real-time data is input into a pre-trained deep convolutional neural network to extract running feature information in the running process of the engineering vehicle; an engineering vehicle running feature model is constructed according to the running feature information, wherein the engineering vehicle running feature model uses a graph neural network (GNN) structure to represent the running state of the engineering vehicle and the interaction relationship with the environment, and optimizes the model parameters through a contrast learning method.

[0034] S2: inputting the real-time data into the engineering vehicle running feature model, outputting a dynamic behavior index, and performing segmented processing on the driving path of the engineering vehicle.

[0035] Specifically, the segmented processing method is to input the real-time data into the engineering vehicle running feature model, wherein the engineering vehicle running feature model includes a distance prediction module, a posture evaluation module, and an environment perception module.

[0036] It should be noted that the distance prediction module calculates a safety distance parameter of the engineering vehicle and the surrounding environmental obstacles; the posture evaluation module analyzes pitch angle parameters and roll angle parameters of the engineering vehicle; and the environment perception module identifies road surface type parameters and road condition complexity.

[0037] Further, the safety distance parameter, the pitch angle parameter, the roll angle parameter, the road surface type parameter, and the road condition complexity are combined to form a dynamic behavior index, and the specific formula is as follows:

[0038]

[0039] wherein DBI is the output value of the dynamic behavior index, d s is the safety distance parameter, d max is a preset maximum safety distance threshold, θ p is the pitch angle parameter, θ r is the roll angle parameter, θ th is an angle threshold, R is the road surface type parameter, C is the road condition complexity, S is the environmental state characteristic value, and α, β, and γ are adjustment coefficients and satisfy α+β+γ=1.

[0040] Further, according to the dynamic behavior index, the driving path of the engineering vehicle is divided, when the road condition complexity is greater than a first preset threshold, the corresponding road section is divided into a complex road section, and the tracking control parameters are dynamically adjusted, the multi-sensor data fusion calculation is started to obtain the attitude angle data and the road surface condition data; when the road condition complexity is less than or equal to the first preset threshold, the corresponding road section is divided into a standard road section, and the preset tracking control parameters are used for regular tracking control.

[0041] It should be noted that the first preset threshold is a boundary value set based on the distribution statistics of the road condition complexity in the historical running data of the engineering vehicle.

[0042] S3: Based on the segmentation processing result, the attitude angle and the road surface condition of the engineering vehicle are calculated through multi-sensor data fusion, and an engineering vehicle trajectory prediction model is established.

[0043] Specifically, the construction method of the engineering vehicle trajectory prediction model is that, based on the results of the segmentation processing, the Kalman filtering algorithm is used to perform fusion operation on the real-time data to obtain three-dimensional attitude angle data of the engineering vehicle; at the same time, the road image data obtained by the camera array is input into the pre-trained semantic segmentation network to extract road texture features, slope features and wetness degree features, and form road surface condition data.

[0044] Further, the three-dimensional attitude angle data and the road surface condition data are input into a long short-term memory network to establish an engineering vehicle trajectory prediction model.

[0045] S4: Based on the engineering vehicle trajectory prediction model, the driving trajectory of the engineering vehicle is predicted and planned, and the vehicle-mounted controller is used to cooperatively control the steering system, the braking system and the power system of the engineering vehicle to complete automatic tracking control.

[0046] Specifically, the prediction trajectory data output by the engineering vehicle trajectory prediction model is used to calculate the optimal driving trajectory of the engineering vehicle by using the recursive least squares algorithm; and the steering angle command, the acceleration command and the braking command are generated according to the optimal driving trajectory.

[0047] Further, the specific formula of the optimal driving trajectory is as follows:

[0048]

[0049] wherein, T opt is the optimal driving trajectory value, x i is the current position coordinate, x p (i) is the prediction trajectory coordinate, n is the prediction time domain length, λ is the forgetting factor, v is the current vehicle speed, μ is the smoothing factor, δ is the trajectory curvature, and σ is the curvature standard deviation threshold.

[0050] It should be noted that the engineering vehicle system includes the engineering vehicle steering system, the braking system and the power system.

[0051] Further, the steering angle command is sent to the electric steering system through the vehicle-mounted controller to perform steering adjustment, and the acceleration command and the braking command are sent to the power system and the braking system respectively to perform power output adjustment and braking force adjustment.

[0052] Specifically, real-time response data of the execution of the engineering vehicle system is collected at the same time, and deviation analysis is performed on the real-time response data and the optimal driving trajectory.

[0053] Further, when the deviation is less than or equal to a second preset threshold, the current control instruction is kept unchanged, and the automatic tracking control is continued to be executed; wherein the optimal driving trajectory value T opt Comparison with the trajectory threshold value: if the optimal driving trajectory value T opt ≤1, it indicates that the running trajectory of the engineering vehicle completely meets the expectation, and the current control strategy is continued to be executed; if 1 < the optimal driving trajectory value T opt ≤3, the trajectory deviation is in an acceptable range, the system keeps the current control parameter and strengthens the monitoring.

[0054] Further, when the deviation is greater than the second preset threshold, the steering angle instruction, the acceleration instruction and the braking instruction are regenerated according to the deviation until the automatic tracking control of the engineering vehicle is completed. The steering angle instruction is dynamically adjusted according to the lateral deviation, and a pre-look point tracking algorithm is used to calculate the expected steering angle; the acceleration instruction is optimized by using an engineering vehicle trajectory prediction model prediction control method based on the longitudinal deviation and the speed error; the braking instruction is dynamically adjusted by analyzing the deviation between the current speed and the expected speed and combining the road condition.

[0055] It should be noted that the second preset threshold is based on the optimal driving trajectory value T opt The index is an evaluation result considering the position deviation, the speed influence and the trajectory smoothness.

[0056] Further, the embodiment also provides an automatic tracking control system for an engineering vehicle, comprising: a collection module, configured to collect real-time data of the engineering vehicle and construct an engineering vehicle running feature model by using a deep learning algorithm; a segmented processing module, configured to input the real-time data into the engineering vehicle running feature model, output a dynamic behavior index, and perform segmented processing on a driving path of the engineering vehicle; an establishment module, configured to calculate a posture angle and a road surface condition of the engineering vehicle by multi-sensor data fusion based on a segmented processing result, and establish an engineering vehicle trajectory prediction model; and a cooperative control module, configured to predict and plan a driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model, and perform cooperative control on a system of the engineering vehicle by using a vehicle-mounted controller to complete automatic tracking control.

[0057] The embodiment also provides a computer device suitable for the automatic tracking control method for the engineering vehicle, comprising a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions to realize the automatic tracking control method for the engineering vehicle proposed in the above embodiment.

[0058] The computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless mode can be achieved by WIFI, operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0059] To sum up, the application realizes feature extraction by constructing a running feature model through a deep learning algorithm, improves the perception accuracy of the attitude and road conditions of the engineering vehicle by combining a multi-sensor data fusion technology, reduces the data processing complexity and optimizes the calculation efficiency by adopting a segmented processing strategy, realizes accurate prediction of the future motion state of the vehicle based on a trajectory prediction model, and finally ensures the optimized cooperation of the steering, accelerating, braking and other actuators through a cooperative control method.

[0060] Embodiment 2

[0061] Referring to Table 1, the second embodiment of the application provides an automatic tracking control method for an engineering vehicle. In order to verify the beneficial effects of the application, economic benefit calculation and simulation experiments are used for scientific demonstration.

[0062] Specifically, a certain type of mine dump truck is selected as the test object. The vehicle is equipped with an Nvidia Xavier NX edge computing platform, an 8-way high-definition camera array, an RTK-GPS positioning module, a 6-axis IMU sensor and a distributed wheel speed sensor network. The test site includes two road conditions, standard roads and complex terrain, and the complex terrain includes 15°-30° slope area, muddy road section and sharp turn area. The data set of the vehicle running under different working conditions is collected, including 100 hours of driving trajectory data, vehicle state data and environmental image data. A deep convolutional neural network based on a ResNet-50 backbone network is constructed using the PyTorch deep learning framework, and the model is pre-trained through a transfer learning method to extract key running features of the engineering vehicle. On this basis, a GNN structure containing 6 layers of graph convolution layers is designed to model the interaction between the vehicle state and the environment. The model parameters are optimized by a contrastive learning method to obtain the running feature model of the engineering vehicle.

[0063] Further, after inputting real-time data into the running feature model, the distance prediction module calculates the safety distance based on the laser radar point cloud data in real time, the attitude evaluation module analyzes the vehicle body attitude using IMU data, and the environment perception module identifies the road conditions through the semantic segmentation network. According to the output dynamic behavior index, the system adaptively divides the road section type. For complex road sections, an improved Kalman filtering algorithm is used to fuse multi-source sensor data to achieve centimeter-level positioning accuracy; at the same time, the road image is input into the semantic segmentation network based on the DeepLab v3+ architecture to extract the road surface features. The fused attitude data and road condition data are input into the bidirectional LSTM network to build a trajectory prediction model.

[0064] Further, based on the trajectory data output by the prediction model, an improved recursive least squares algorithm is used to calculate the optimal driving trajectory, and a model predictive control (MPC) method is used to generate control instructions. The vehicle-mounted controller sends the control instructions to the electric steering system, power system and braking system respectively for execution through the CAN bus. At the same time, the system collects execution data in real time and performs deviation analysis, and when the deviation exceeds the second preset threshold, the online optimization update of the control instructions is triggered.

[0065] Specifically, as shown in Table 1, under standard road conditions, the trajectory tracking error of the method of the present application is 0.25m, which is reduced by 40.5% compared with 0.42m of the traditional method; the attitude estimation accuracy is improved to 0.6°, which is improved by 50% compared with 1.2° of the traditional method; the system response time is reduced from 145ms to 85ms, and the processing efficiency is improved by 41.4%. In terms of environmental perception, the road condition recognition accuracy reaches 94.2%, which is improved by 7.7 percentage points compared with the traditional method; the control stability reaches 95.8%, which is improved by 6.6 percentage points compared with 89.2% of the traditional method. At the same time, the energy consumption index is reduced to 2.3kWh / km, which saves 14.8% of energy consumption compared with the traditional method. It is particularly worth noting that the safety distance accuracy is improved from 0.85m to 0.45m, which is improved by 47.1%; the dynamic response time is shortened from 1.2s to 0.6s, and the reaction speed is doubled.

[0066] Table 1, comparison table of the method of the present application and the traditional method

[0067]

[0068] Further, in complex terrain conditions, the advantages of the method of the application are more prominent. The trajectory tracking error is reduced from 1.65m to 0.68m, with an improvement of 58.8%; the attitude estimation accuracy is improved from 2.8° to 1.1°, with an accuracy of 60.7%; the system response time is reduced from 175ms to 95ms, with a real-time performance of 45.7%; the road condition recognition accuracy is improved from 73.5% to 91.8%, with an increase of 18.3 percentage points, indicating that the method of the application has stronger environmental perception ability in complex environment; the control stability is improved from 76.8% to 93.5%, with an increase of 16.7 percentage points, ensuring the smooth running of the vehicle in complex terrain. The energy consumption index is reduced from 3.8kWh / km to 2.9kWh / km, with an energy saving effect of 23.7%. In terms of safety performance, the safety distance accuracy is improved from 1.55m to 0.72m, with an increase of 53.5%; the dynamic response time is shortened from 1.8s to 0.8s, with an increase of 55.6%.

[0069] It should be noted that the above examples are only used to illustrate the technical solutions of the application and are not limiting. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the application, which should be covered in the scope of the claims of the application.

Claims

1. An automatic tracking control method for an engineering vehicle, characterized by: The application relates to an automatic tracking control method for an engineering vehicle. Real-time data of the engineering vehicle is collected, and an engineering vehicle operation feature model is constructed by using a deep learning algorithm; The real-time data is input into the engineering vehicle operation feature model, a dynamic behavior index is output, and the driving path of the engineering vehicle is processed in segments; Based on the segment processing result, the attitude angle and the road surface condition of the engineering vehicle are calculated through multi-sensor data fusion, and an engineering vehicle trajectory prediction model is established; The driving trajectory of the engineering vehicle is predicted and planned according to the engineering vehicle trajectory prediction model, and the engineering vehicle system is cooperatively controlled through a vehicle-mounted controller to complete automatic tracking control; The engineering vehicle system comprises an engineering vehicle steering system, a braking system and a power system; The optimal driving trajectory of the engineering vehicle is calculated by using a recursive least square algorithm according to the predicted trajectory data output by the engineering vehicle trajectory prediction model; The steering angle instruction, the acceleration instruction and the braking instruction are generated according to the optimal driving trajectory; The steering angle instruction is sent to the electric steering system through the vehicle-mounted controller to execute steering adjustment, and the acceleration instruction and the braking instruction are sent to the power system and the braking system respectively to execute power output adjustment and braking force adjustment; Meanwhile, real-time response data of the execution of the engineering vehicle system is collected, and deviation analysis is performed on the real-time response data and the optimal driving trajectory; When the deviation is less than or equal to a second preset threshold, the current control instruction is kept unchanged, and the automatic tracking control is continued to be executed; When the deviation is greater than the second preset threshold, the steering angle instruction, the acceleration instruction and the braking instruction are regenerated according to the deviation until the automatic tracking control of the engineering vehicle is completed; The specific formula of the optimal driving trajectory is as follows: ； wherein, is an optimal trajectory value, is a current position coordinate, is a predicted trajectory coordinate, is a predicted horizon length, is a forgetting factor, is a current vehicle speed, is a smoothing factor, is a trajectory curvature, is a curvature standard deviation threshold.

2. The automatic tracking control method for a construction vehicle according to claim 1, characterized by: The construction method of the engineering vehicle trajectory prediction model is as follows, Based on the segment processing result, Kalman filtering algorithm is used to perform fusion operation on the real-time data to obtain three-dimensional attitude angle data of the engineering vehicle; Meanwhile, the road surface image data acquired by the camera array is input into a pre-trained semantic segmentation network to extract road surface texture features, slope features and wet slip degree features, thereby forming road surface condition data; The three-dimensional attitude angle data and the road surface condition data are input into a long short-term memory network to establish the engineering vehicle trajectory prediction model.

3. The method of automatic tracking control for an engineered vehicle of claim 2, wherein: The segment processing method is as follows, The real-time data is input into the engineering vehicle operation feature model, wherein the engineering vehicle operation feature model comprises a distance prediction module, an attitude evaluation module and an environment perception module; the distance prediction module is used for calculating safety distance parameters of the engineering vehicle and surrounding environmental obstacles; the attitude evaluation module is used for analyzing pitch angle parameters and roll angle parameters of the engineering vehicle; and the environment perception module is used for identifying road type parameters and road condition complexity; The safety distance parameters, the pitch angle parameters, the roll angle parameters, the road type parameters and the road condition complexity are combined to form a dynamic behavior index; According to the dynamic behavior index, the driving path of the engineering vehicle is divided, when the road condition complexity is greater than a first preset threshold, the corresponding road section is divided into a complex road section, and tracking control parameters are dynamically adjusted, and multi-sensor data fusion calculation is started to obtain attitude angle data and road surface condition data; When the road condition complexity is less than or equal to the first preset threshold, the corresponding road section is divided into a standard road section, and a preset tracking control parameter is used for regular tracking control.

4. The method of automatic tracking control for an engineered vehicle of claim 3, wherein: The specific formula of the dynamic behavior index is as follows: ； wherein, is a dynamic behavior indicator output value, is a safety distance parameter, is a preset maximum safety distance threshold value, is a pitch angle parameter, is a roll angle parameter, is an angle threshold value, is a road surface type parameter, is a road condition complexity, is an environmental state characteristic value, , and are adjustment coefficients.

5. The method of automatic tracking control for an engineered vehicle of claim 4, wherein: The real-time data includes driving trajectory data, vehicle state data, and surrounding environment image data; and the method for establishing the engineering vehicle running feature model is, Real-time data obtained by a camera array, a GPS positioning module, a wheel speed sensor, and an inertial measurement unit (IMU) carried by the engineering vehicle are collected. The real-time data is preprocessed, and the preprocessed real-time data is input into a pre-trained deep convolutional neural network to extract running feature information in the running process of the engineering vehicle. An engineering vehicle running feature model is constructed according to the running feature information, wherein the engineering vehicle running feature model includes an engineering vehicle running state and an environmental interaction relationship represented by a graph neural network (GNN) structure, and model parameters are optimized by a contrast learning method.

6. An automatic tracking control system for a construction machine based on the automatic tracking control method for a construction machine according to any one of claims 1 to 5, characterized by: The method comprises the following steps: A collection module is configured to collect real-time data of the engineering vehicle and construct an engineering vehicle running feature model by using a deep learning algorithm. A segmentation processing module is configured to input the real-time data into the engineering vehicle running feature model, output a dynamic behavior index, and perform segmentation processing on a driving path of the engineering vehicle. An establishment module is configured to calculate a posture angle and a road surface condition of the engineering vehicle by multi-sensor data fusion based on a segmentation processing result, and establish an engineering vehicle trajectory prediction model. A cooperative control module is configured to predict and plan a driving trajectory of the engineering vehicle according to the engineering vehicle trajectory prediction model, and perform cooperative control on a system of the engineering vehicle by using a vehicle-mounted controller to complete automatic tracking control.

7. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that: The processor executes the computer program to realize the steps of the automatic tracking control method for the engineering vehicle according to any one of claims 1-5.

8. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the automatic tracking control method for the engineering vehicle according to any one of claims 1-5.

Images