Temperature-quality driven road surface unmanned aerial vehicle group operation whole process path optimization method

Abstract

The invention relates to a temperature-quality driven road surface unmanned aerial vehicle group operation whole process path optimization method, which is characterized in that a data set is established through a large amount of asphalt mixture cooling test data, and an asphalt mixture temperature attenuation curve prediction model is trained to predict the attenuation trend of the asphalt mixture temperature along with time, and a temperature attenuation curve is obtained. An intelligent prediction model for predicting the actual temperature of the asphalt mixture by collecting the temperature through a non-contact infrared temperature measurement camera is established, and the intelligent prediction model is used for monitoring the asphalt temperature in real time in the compaction operation. According to the obtained asphalt mixture temperature attenuation curve, an optimal compaction temperature interval and a corresponding optimal compaction time window are extracted, and the length of the operation section is divided on the basis of the optimal compaction time window; when the machine group works, the actual temperature of asphalt at the rolling position is predicted in real time, the speed of the road roller is adjusted, and it is guaranteed that the road roller conducts compaction work in the optimal compaction temperature interval.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned driving of road roller groups, and in particular to a temperature-quality driven method and system for optimizing the path of a road roller group during the entire operation process. Background Art

[0002] In recent years, with the development of science and technology, many technologies have been gradually applied to construction machinery, such as unmanned driving technology and positioning technology. The field of construction machinery is also developing towards intelligence, digitization and unmanned operation. Among them, road rollers are indispensable construction machinery for road compaction. Their realization of high-precision and high-efficiency digital unmanned operations is of great significance to the development of my country's highway industry.

[0003] Asphalt pavement compaction is usually done with a double steel wheel roller. As the asphalt mixture is very sensitive to compaction degree and temperature, over-pressing or under-pressing, overheating or overcooling the asphalt mixture may lead to unsatisfactory compaction results. Therefore, the compaction of the asphalt mixture must be completed within the optimal compaction temperature range of the asphalt mixture. The following problems exist in the current compaction operation: (1) Although existing methods can roughly predict the time period of the optimal compaction temperature range of the asphalt mixture under different environmental conditions, the temperature of the asphalt mixture within the optimal compaction temperature range cannot be obtained. Since the on-site construction environment conditions may change during the operation, only the time period without the specific asphalt mixture temperature cannot guide the operation of the roller. (2) During the compaction process, the length of the operation section is mostly determined by the driver's experience. Longer operation sections are prone to the problem of overcooling the asphalt mixture, resulting in reduced compaction quality, while shorter operation sections will lead to problems such as low compaction efficiency and excessive fuel consumption of the roller.

[0004] Therefore, the present invention proposes to establish a prediction model for the temperature attenuation curve of asphalt mixture and a real-time monitoring model for the actual temperature of the asphalt mixture during the compaction operation of the roller based on the intelligent prediction algorithm. Finally, based on the predicted temperature attenuation curve of the asphalt mixture and the real-time predicted actual temperature of the asphalt mixture, the length of the working section of the unmanned roller fleet is controlled to meet the actual engineering needs and improve the compaction quality of the asphalt pavement. Summary of the invention

[0005] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to propose a temperature-quality driven path optimization method for the whole process of pavement drone group operation. A data set is established through a large amount of asphalt mixture cooling test data and an asphalt mixture temperature decay curve prediction model is trained to predict the decay trend of asphalt mixture temperature over time and obtain a temperature decay curve. An intelligent prediction model is established to predict the actual asphalt mixture temperature by collecting temperature with a non-contact infrared temperature measurement camera, which is used to monitor the asphalt temperature in real time during compaction operation. According to the obtained asphalt mixture temperature decay curve, the optimal compaction temperature range and the corresponding optimal compaction time window are extracted, and the length of the working section is divided based on the optimal compaction time window; when the group is operating, the actual temperature of the asphalt at the rolling position will be predicted in real time, and the roller speed will be adjusted based on the temperature to ensure that the roller performs compaction operations in the optimal compaction temperature range, so as to achieve real-time adjustment under unmanned driving conditions during the compaction process. In addition, the length of the next working section is optimized according to the actual time when the previous roller completes its working section, so as to achieve real-time adjustment of the working section length.

[0006] The technical solution adopted by the present invention to solve the technical problem is:

[0007] In a first aspect, the present invention provides a temperature-quality driven path optimization method for a whole process of a road drone swarm operation, the method comprising the following contents:

[0008] Obtain asphalt mixture temperature data under different environmental conditions, and use intelligent prediction algorithms to establish a prediction model that can predict the asphalt mixture temperature attenuation curve under different environmental conditions;

[0009] The surface temperature and actual temperature data of asphalt mixture radiating outward during compaction under different environmental conditions are obtained, and a real-time monitoring model for the actual temperature of asphalt mixture is established using an intelligent prediction model. The relationship between the surface temperature and the actual temperature of asphalt mixture radiating outward during compaction under different environmental conditions can be obtained, and the actual temperature inside the asphalt mixture can be monitored in real time during compaction operations.

[0010] When the fleet is performing compaction operations, the operation section is divided into a straight rolling area and a lane change area. When compacting in the straight rolling area, the asphalt mixture actual temperature real-time monitoring model is used to predict the actual temperature of the asphalt at the rolling position in real time and record it. Then, according to the current operating environment, the asphalt mixture temperature attenuation curve prediction model is used to predict the asphalt mixture temperature attenuation curve that meets the on-site environmental conditions. The optimal compaction time window T corresponding to the optimal compaction temperature range is determined on the attenuation curve. max ; Determine the length of the next operation section according to the optimal compaction time window. The total operation time T of the fleet in any operation section ranges from: 0.75T max <T<0.9Tmax , at this time T max The temperature decay curve is determined by the temperature decay curve predicted by the actual temperature data of the asphalt mixture in the previous operation section. The temperature decay curve of the asphalt mixture in the first operation section is predicted and generated by the actual temperature data of the asphalt mixture in the pre-operation section. Under the premise that the roller speed V remains unchanged, the operation section length is divided according to L=V(T-t2) / q.

[0011] Where L is the length of the straight rolling area in the working section, that is, the length of the working section, t2 is the total time spent on changing lanes in the same working section; q is the number of straight rolling lanes that a roller is responsible for in the working section;

[0012] When the machine fleet is actually operating, compaction operations are carried out according to the divided operating sections, and the difference between the actual temperature of the asphalt mixture predicted by the real-time monitoring model of the actual temperature of the asphalt mixture and the temperature value of the point predicted on the temperature attenuation curve is compared. The temperature difference threshold is set. If the difference is greater than the temperature difference threshold, the roller is accelerated to ensure that the roller completes the compaction operation within the optimal compaction time window.

[0013] Furthermore, during the compaction operation of the machine group, the path planning of rolling and lane changing is optimized. The straight rolling area includes several straight rolling lanes parallel to the roadside. The number of straight rolling lanes is determined by the road width, the wheel width of the roller, and the overlapping width of adjacent straight rolling lanes. The lane changing area includes several curved segments. The lane changing curve selects a third-order Bezier curve. The lane changing curve ensures that the roller straightens the body and the articulated steering angle is zero after changing lanes, and the entire lane changing path is continuous and smooth, meeting the kinematic and dynamic constraints of the roller. At the same time, straight rolling and curved lane changing are performed alternately to ensure that the roller can roll all the asphalt mixture in the operation section in a relatively short time.

[0014] Before compaction, a pre-operation section is first set for pre-compaction. The optimal compaction speed V of the roller is determined based on experience. The roller speed is fixed to the optimal compaction speed V, and a speed difference threshold is set. Compaction is performed in the pre-operation section. The speed difference between the real-time speed and the optimal compaction speed during lane change is recorded. The total lane change time t2 is also recorded. If the speed difference is greater than the speed difference threshold, the lane change curve of the next operation section needs to be optimized to obtain the optimized lane change curve.

[0015] In each working section, the total working time t1 of the straight rolling area is determined by T-t2. At this time, t2 is the actual total lane changing time of the previous working section. The length of the current working section is determined according to L=v·t1 / q. The lane changing curves of different rolling lanes in the same working section are the same. The optimized t2 of each working section changes and shows a trend of shortening. Correspondingly, the length of the straight rolling area becomes larger, thereby optimizing the length of the working section.

[0016] Furthermore, the third-order Bezier curve is composed of a starting point P0, an end point P3, and two points P1 and P2 in the longitudinal driving direction. The optimization process of the lane-changing curve is as follows: first, the coordinates of P1 and P2 are optimized so that the difference between the maximum curvature radius and the minimum curvature radius of the planned curve is minimized, so that the roller can not only ensure a continuous and smooth lane-changing path during the entire lane-changing process, but also control the maximum value of the relative turning angle between the front and rear bodies of the roller within a certain range during the lane-changing process, thereby ensuring the efficiency of the roller's lane-changing; the optimization function of P1 and P2 is: J(P1, P2) = k max (t a )-k min (t b ),t a ,t b ∈(0,1), t a ,t b Get the maximum curvature radius k for the Bezier curve max With the minimum curvature radius k min The value at time;

[0017] Secondly, when changing lanes, if it is found that after optimizing the control points P1 and P2, the speed difference between the real-time speed of the roller and the optimal compaction speed V of the roller is still greater than the speed difference threshold, it means that the length of the lane changing area is set too small. At this time, the lane changing starting point P0 will be kept unchanged, and the position of the end point P3 will be adjusted to increase the length of the lane changing area. The process of optimizing the coordinates of P1 and P2 and adjusting the position of the end point P3 will be repeated until the speed difference is controlled within the speed difference threshold. The length of the lane changing area increases by 0.5m each time.

[0018] Furthermore, the asphalt mixture temperature attenuation curve prediction model includes an intelligent prediction algorithm and a fitting module. The input of the intelligent prediction algorithm is ambient temperature, ambient humidity and wind speed, and the output is the asphalt mixture temperature in the future. The intelligent prediction algorithm adopts at least one of LSTM, Transformer, BiLSTM, and GRU; the input of the fitting module is the historical asphalt mixture temperature before prediction and the asphalt mixture temperature after prediction, and the output is the attenuation curve.

[0019] Furthermore, the input of the intelligent prediction model is ambient temperature, ambient humidity, actual wind speed, vehicle speed, and the surface temperature radiated outward by the asphalt mixture, and the output is the actual temperature of the asphalt mixture; the intelligent algorithm adopts at least one of the BP neural network, ANN neural network, and RNN neural network.

[0020] In a second aspect, the present invention provides a temperature-quality driven road drone swarm operation whole process path optimization system, comprising a number of rollers, and a control module for assisting automatic driving, a host computer and a decision module, wherein a monocular camera for detecting the curbs on both sides to prevent the roller from driving out of the working area and causing danger, an angle sensor for obtaining the relative turning angle of the vehicle body, a speed sensor for obtaining the driving speed of the roller, a thermometer and hygrometer for collecting ambient temperature and humidity, an infrared temperature measuring camera for collecting the outwardly radiated surface temperature of the asphalt mixture, a wind speed sensor for collecting relative wind speed, and an RTK GNSS for collecting real-time position information of the roller are installed on the roller;

[0021] The host computer is used to view the real-time operation status of the fleet, including the visualization of the compaction of the fleet in the operation area, the compaction parameters of each roller in the fleet, the real-time temperature display of the compacted material, and the path planning based on the optimal compaction temperature range;

[0022] The decision module communicates wirelessly with the host computer and can obtain data from the monocular camera, angle sensor, speed sensor, thermometer, infrared temperature camera, RTK GNSS, and wind speed sensor. It is also loaded with the asphalt mixture temperature attenuation curve prediction model and the asphalt mixture actual temperature real-time monitoring model.

[0023] The asphalt mixture temperature decay curve prediction model is used to predict the asphalt mixture temperature decay curve under different environmental conditions;

[0024] The real-time monitoring model for the actual temperature of the asphalt mixture is used to obtain the relationship between the surface temperature radiated outward by the asphalt mixture during compaction operations under different environmental conditions and the actual temperature.

[0025] Furthermore, the decision module is also loaded with a lane-changing curve optimization algorithm, which is used to optimize the speed difference between the set speed and the actual speed of the roller, the middle two control points and the end point on the third-order Bezier curve, until the speed difference is controlled within the set speed difference threshold range;

[0026] The length of the straight rolling area of ​​the next working section is determined according to the total lane changing time of the previous working section.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. The present invention establishes a prediction model for the temperature decay curve of asphalt mixture, which can predict the temperature decay curve of asphalt mixture under different environmental conditions, and obtain the optimal compaction temperature range and corresponding time window of asphalt mixture. The data source of this curve has two parts. The first part is generated by the temperature data corresponding to each time period collected in real time during the on-site operation of the roller. The second half of the curve will predict the temperature of the asphalt mixture in the next period of time based on the collected temperature data of the asphalt mixture and the environmental factors that affect the temperature drop of the asphalt mixture through the prediction model of the temperature decay curve of the asphalt mixture; the historical temperature + real-time temperature + predicted temperature are fitted to a complete temperature decay curve through curve fitting. This temperature decay curve can view all temperature data of the entire process of asphalt mixture temperature drop. The length of the subsequent construction operation section will be determined by the optimal compaction temperature range and the corresponding optimal compaction time window of the temperature decay curve; and the temperature corresponding to each time point on the curve will be compared with the actual asphalt temperature measured during the operation of the roller, and it is judged whether the roller needs to be accelerated based on the comparison results to guide the compaction operation.

[0029] 2. The present invention establishes a real-time monitoring model for the actual temperature of asphalt mixture during the compaction process of the roller, which takes into account environmental factors such as temperature, humidity, wind speed, and vehicle speed, so as to realize the whole process, continuous, real-time, and contactless monitoring of the actual temperature of the asphalt mixture rolled by the roller instead of directly using the surface temperature radiated by the asphalt. The roller collects the surface temperature of the asphalt mixture to be rolled in front or behind through the infrared temperature measuring camera installed on the frame, obtains the current environmental conditions and roller parameters through the speed sensor, thermometer, infrared temperature measuring camera and wind speed sensor installed on the roller, and predicts the real temperature inside the asphalt mixture according to the intelligent prediction model. At the same time, the operation time and roller speed of the roller will be displayed in real time on the working platform during the operation of the roller, wherein the roller speed is obtained by the speed sensor. When the roller is operating, the real temperature of the currently rolled asphalt will be compared with the real temperature of the asphalt at the same time point on the temperature decay curve in real time, and judges whether the roller needs to be accelerated according to the difference, so as to ensure that the compaction operation of the roller is carried out within the optimal compaction time window. According to the above real-time monitoring model of the actual temperature of asphalt mixture, the roller can record the real temperature data of asphalt mixture in the whole process and whole section in each working section, and use this data in combination with the prediction model of temperature decay curve of asphalt mixture to predict the temperature decay curve that meets the on-site construction conditions. The temperature decay curve of each working section will be generated in real time from the temperature data of asphalt mixture collected in the previous working section. Since the first working section does not have a previous working section, a pre-compaction test will be carried out when the formal compaction operation is carried out. A pre-operation section length is set, and the asphalt mixture temperature data collected in real time in the pre-operation section is used to predict and generate the asphalt mixture temperature decay curve that guides the first working section.

[0030] 3. The present invention establishes the path planning for rolling and lane changing during the compaction operation of the fleet, and optimizes the path. Due to the material properties of asphalt, the compaction of the asphalt mixture needs to be carried out in a relatively short time, so it is necessary to carry out reasonable path planning for the compaction operation of the unmanned roller fleet. When performing path planning, the working section is divided into a straight rolling area and a lane changing area. The straight rolling area includes a number of straight rolling lanes parallel to the roadside. The number of divided straight rolling lanes is determined by the road width, the roller wheel width, and the overlapping width of adjacent straight rolling lanes; the lane changing area includes a number of curve segments, and the lane changing curve selects a third-order Bezier curve. The lane changing curve can ensure that the roller straightens the vehicle body after changing lanes, the articulated steering angle is zero, and the entire lane changing path is continuous and smooth, satisfying the kinematic and dynamic constraints of the roller. At the same time, in order to complete the rolling of asphalt mixture in a shorter time, some improvements have been made to the path planning. The previous practice of changing lanes after reciprocating rolling on any straight rolling road has been improved to alternating straight rolling and curved lane changing, which can ensure that the roller can roll all the asphalt mixture in the working section in a shorter time. At the same time, the third-order Bezier curve is optimized. It is found in the actual operation of the roller that if the relative turning angle of the front and rear bodies of the roller is too large, the roller speed will drop significantly. Therefore, if the relative turning angle of the front and rear bodies is too large during the lane changing process, it will greatly affect the total lane changing time of the roller, affect the construction efficiency, and increase the difficulty of the roller to complete all compaction operations in the optimal compaction temperature range. Therefore, the lane changing curve of the roller, that is, the third-order Bezier curve, is optimized, and the coordinates of the control points are optimized, so that the roller can not only ensure the continuous and smooth lane changing path during the entire lane changing process, but also control the maximum value of the relative turning angle of the front and rear bodies of the roller within a certain range during the lane changing process, so as to ensure the efficiency of the roller lane changing.

[0031] 4. The present invention will accurately determine the asphalt mixture paving length of each working section based on the above two intelligent prediction methods and the compaction parameters of the roller to ensure the quality and efficiency of the compaction operation. The optimal compaction temperature range and the corresponding time window of the asphalt mixture can be obtained according to the temperature decay curve of the asphalt mixture. According to the time window of the roller, the length of each working section is pre-divided. The total working time of the group in any working section is T, and T will be calculated according to the maximum compaction time window T max A certain proportion of is taken, and T = t1 + t2 is set, where t1 is the total operation time of the roller in the straight rolling area, t2 is the total time spent by the roller in changing lanes, the lane changing length and the coordinates of the control points of the lane changing curve are preset, and the initial lane changing curve can be obtained. The roller will use this curve to change lanes when performing the pre-operation section operation. The total lane changing time t2 is recorded and t1 is calculated. According to t1, the length of the operating section is determined. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagram of the asphalt mixture operation path of a fleet of unmanned rollers.

[0033] Figure 2 Schematic diagram of temperature attenuation curve of asphalt mixture.

[0034] Figure 3 Schematic diagram of the roller lane change curve generation.

[0035] Figure 4 A schematic flow chart of the temperature-quality driven path optimization method for the entire process of road drone swarm operations of the present invention. DETAILED DESCRIPTION

[0036] The specific embodiments of the present invention are given below. The specific embodiments are only used to further illustrate the present invention in detail and are not limited to the protection scope of the present application.

[0037] The present invention provides a temperature-quality driven road drone swarm operation full-process path optimization system, which can adjust the operation path of the unmanned roller swarm in real time according to the temperature change of asphalt mixture, and realize unmanned and intelligent high-precision operation.

[0038] The system comprises a roller subsystem, a construction personnel operation subsystem and a paver control subsystem. The roller subsystem comprises a control module, a sensor module, a decision module and a transmission module.

[0039] The roller subsystem: 1. Control module. The control module of the roller refers to the power-assisting device installed in each actuator of the roller. The power-assisting device is directly connected to the decision module and is controlled by the decision module. The actuators used in the roller operation include: gear actuator, steering actuator, throttle actuator, brake actuator, ignition and vibration switch mechanisms.

[0040] The gear actuator uses a stepper motor as an external power-assist device, which relies on the motor torque to drive the gear push rod to rotate to achieve the gear switching function;

[0041] The steering actuator adopts an electric steering wheel to replace the original steering wheel;

[0042] The throttle actuator and brake actuator use power-assisting devices such as electric push rods to drive the throttle push rod and brake pedal to achieve acceleration, deceleration and braking operations;

[0043] The ignition and vibration switch mechanisms use relays to control the switches, including modifications to the ignition switch, vibration switch, and large and small vibrations.

[0044] 2. The sensor module mainly includes: 1) RTK GNSS. It is used to collect the real-time position information of the roller. It consists of a GNSS base station and a GNSS mobile station to realize differential analysis. 2) Monocular camera. The monocular camera is fixed on both sides of the roller cab to detect the curbs on both sides to prevent the roller from driving out of the working area and causing danger. 3) LiDAR. The LiDAR is installed at the front end of the front frame and the rear end of the rear frame respectively to detect surrounding obstacles and transmit obstacle information to the decision module, which makes judgments and takes corresponding actions. 4) Angle sensor. The angle sensor is installed at the hinged connection between the front frame and the rear frame of the roller. It is used to obtain the relative turning angle of the vehicle body and transmit the data to the decision module for analysis and calculation. It is used for path tracking control of the roller. The communication protocol is RS485 protocol. 5) Speed ​​sensor. It is used to obtain the driving speed of the roller.

[0045] 3. The decision module is a roller control board, which integrates the interfaces of the above-mentioned sensor modules and the actuator control equipment. The main control adopts the STM32 single-chip microcomputer. This board is the "brain" of the entire unmanned roller, and all judgments and corresponding decisions are made on this board.

[0046] 4. The transmission module adopts a wireless transmission module, including an antenna. The antenna is placed on the top of the roller cab and communicates with the host computer through a virtual serial port.

[0047] The construction personnel operation subsystem (host computer) is a PC-side page that can be used to view the real-time operation status of the fleet, mainly including the visualization of the compaction of the fleet in the operating area, the compaction parameters of each roller in the fleet, the real-time temperature display of the compacted material, and the path planning based on the optimal compaction temperature range.

[0048] The present invention provides a temperature-quality driven path optimization method for the entire process of a road drone swarm operation, which specifically includes the following steps:

[0049] Step 1: Establish a prediction model for the temperature attenuation curve of asphalt mixture.

[0050] Firstly, environmental conditions that may affect the temperature of asphalt mixture are selected, mainly including ambient temperature, ambient humidity, wind speed, etc. Then, through a large number of experiments, the amplitude and value of the temperature drop of asphalt mixture over time under different environmental conditions are collected; the corresponding asphalt mixture temperature under different ambient temperatures, ambient humidity, and wind speeds are obtained, and the first data set composed of ambient temperature, ambient humidity, wind speed, and asphalt mixture temperature is divided into training data and test data, and 75% of the sample data is used as the training set, and 25% of the samples are used as the verification test set, which are randomly split into the training set and the verification test set;

[0051] In this embodiment, a long short-term memory neural network (LSTM) is used as the basic model of the intelligent prediction algorithm to establish an asphalt mixture temperature decay curve prediction model. The asphalt mixture temperature decay curve prediction model is obtained by:

[0052] (1) Both training data and test data are normalized. The normalization formula is: Among them, max is the maximum value of each influencing factor in the sample data, min is the minimum value of each influencing factor in the sample data, x is the sample data, x * is the normalized sample data;

[0053] (2) The ambient temperature, ambient humidity, and wind speed are used as inputs, and the temperature of the asphalt mixture in the future is used as output. The number of neurons in the input layer and the number of neurons in the output layer of the LSTM model are 3 and 1, respectively. The number of neurons in the hidden layer is not fixed and needs to be determined by trial and error. By substituting the number of neurons from 1 to 50, the error results in each case are compared and the number of neurons in the hidden layer with the best prediction results is selected. The learning rate is set to 0.02 and the number of iterations is 150. In the process of establishing the LSTM model, the LSTM function library provided by Matlab is referenced to W f , W i , W c , Wo and other weight matrices and b f 、b i 、b c 、b o The bias items are randomly initialized, and then the Adam optimization algorithm is used to iteratively update the network to adjust the model weights, so that the network can learn better and faster.

[0054] (3) The model is trained with the training set and the validation test set is used for validation testing to obtain a trained LSTM model. The asphalt mixture temperature measured on site in the previous period is input, and the trained LSTM model is used to predict the output asphalt mixture temperature in the future period. The predicted temperature is then curve-fitted with the input asphalt mixture temperature in the previous period in the fitting module to obtain the temperature attenuation curve of the asphalt mixture.

[0055] The LSTM model and fitting module as a whole constitute the asphalt mixture temperature attenuation curve prediction model, which is used to predict the temperature attenuation trend of asphalt mixture.

[0056] Step 2: Establish a real-time monitoring model for the actual temperature of the asphalt mixture during compaction operation.

[0057] (1) Install a speed sensor, a thermometer and humidity meter, a contact temperature probe, an infrared temperature camera, and a wind speed sensor on the roller. The speed sensor is installed on the steel wheel of the roller to collect the real-time vehicle speed; the thermometer and hygrometer is installed on the roof of the roller body to collect the ambient temperature and humidity of the current construction; the contact temperature probe is installed on the bottom of the roller frame through the guide rail and the motor to collect the actual temperature of the asphalt mixture. The motor can drive the guide rail to move up and down. The contact temperature probe is installed on the guide rail. When temperature measurement is required, the guide rail slides vertically downward to drive the probe deep into the asphalt mixture. During normal driving, the probe is retracted to prevent the probe from being damaged too quickly; the infrared temperature camera is installed at the bottom of the roller frame to collect the outward radiating surface temperature of the asphalt mixture; the wind speed sensor is installed on the roof of the roller body to collect the wind speed. It should be noted that since the wind speed sensor is installed on the roller body, the relative wind speed to the vehicle speed is measured, and the factor that actually affects the temperature of the asphalt mixture is the actual wind speed. Therefore, the actual wind speed is obtained by combining the vehicle speed measured by the speed sensor. Subsequently, multiple sets of data are collected and processed, each set of data including actual wind speed, ambient temperature and humidity, actual temperature of asphalt mixture, and surface temperature of asphalt mixture radiating outward.

[0058] (2) Construct a real-time monitoring model for the actual temperature of asphalt mixture. The first step is to standardize the data. Standardize all the collected data to between 0 and 1, and construct a second data set with the standardized data in groups and divide it into training and test sets. The training set accounts for 75% and the test set accounts for 25%. The second step is to use the BP neural network to construct an actual temperature monitoring model for asphalt mixture. The BP neural network has five input layer nodes, which are ambient temperature, ambient humidity, actual wind speed, vehicle speed, and the surface temperature radiating outward from the asphalt mixture. The output layer node of the BP neural network is one, which is the actual temperature of the asphalt mixture. There is one hidden layer, and the number of hidden layer nodes is determined by the empirical formula In the formula, l is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, and α is a constant between 1 and 10. According to the formula, the number of hidden layer nodes is 3 to 12. In this embodiment, 10 is initially selected to obtain the basic architecture of the BP neural network. The third step is to train the BP neural network. The training set divided by the second data set is input into the BP neural network for training to obtain the actual temperature prediction value of the asphalt mixture; and the output error of each layer of neurons is calculated through the error function for back propagation, and then the weights and thresholds of each layer of neurons are adjusted according to the error function.

[0059] The error function is defined as follows: Assume that the number of learning samples is P, and x 1 ,x 2 ,...x pAfter inputting the learning sample into the BP neural network, the output is The BP neural network algorithm generally uses a square error function as the objective function of the neural network. At this time, the error E of the pth learning sample is p It can be expressed as: in, is the expected output of the pth learning sample, y k is the actual temperature of asphalt mixture corresponding to the learning sample.

[0060] The global error of these P learning samples can be further obtained, and the calculation formula is:

[0061] m is the number of input layer nodes, n is the number of output layer nodes, in this embodiment, n=1, and P is the number of learning samples.

[0062] After the error function is determined, the weights and thresholds of the output layer and the hidden layer can be changed according to the gradient descent method to set the error limit E min , when E<E min If the training is stopped when E≥E min , then continue training until E<E min So far, a trained BP neural network model is obtained, which is the real-time monitoring model for the actual temperature of asphalt mixture.

[0063] (3) The real-time monitoring model of the actual temperature of asphalt mixture is used to establish the relationship between the temperature collected by the infrared temperature camera and the temperature collected by the contact temperature probe, that is, the relationship between the surface temperature radiated outward by the asphalt mixture and the actual temperature is established. The real-time monitoring model of the actual temperature of asphalt mixture is stored in the construction personnel operation subsystem of the unmanned roller fleet and is used to monitor the real-time actual temperature of the asphalt mixture during the compaction process; during the unmanned operation of the roller, no contact temperature probe is set, and the surface temperature radiated outward by the infrared temperature camera is collected and used as the input of the real-time monitoring model of the actual temperature of asphalt mixture, and the actual wind speed, vehicle speed, ambient temperature and ambient humidity are obtained according to the working environment, and then the real-time monitoring model of the actual temperature of asphalt mixture outputs the actual temperature of the asphalt mixture at the temperature measurement point, so as to realize the real-time monitoring of temperature during the compaction operation.

[0064] Step 3: Determine the optimal compaction time window T according to the asphalt mixture temperature decay curve and the optimal compaction temperature range. max , and then carry out path planning and divide the length of the working section.

[0065] Asphalt mixtures are very sensitive to the degree and temperature of compaction. Compacting at temperatures above or below the optimal threshold (i.e. outside the optimal compaction temperature range) may result in failure to achieve the desired layer density / thickness or even premature failure.

[0066] Based on the actual temperature of the monitored asphalt mixture obtained in step 2, it is input into the asphalt mixture temperature decay curve prediction model obtained in step 1 to predict the asphalt mixture temperature decay curve that meets the on-site environmental conditions.

[0067] The optimal compaction temperature threshold is determined based on construction experience, and the range between the upper and lower limits of the optimal compaction temperature threshold is the optimal compaction temperature range;

[0068] Find the optimal compaction time window T corresponding to the optimal compaction temperature range on the asphalt mixture temperature decay curve that meets the on-site environmental conditions. max (See Figure 2 ).

[0069] Since the roller operation time is long, the working environment conditions may change at any time, resulting in different temperature drop rates of the asphalt mixture. Therefore, different working sections may have different asphalt mixture temperature decay curves. In order to obtain the temperature decay curve of the asphalt mixture more accurately, the present invention will collect the actual temperature of the asphalt mixture when the roller is operating in real time, and the temperature decay curve of each working section will be generated in real time by the asphalt mixture temperature data collected in the previous working section. Since the first working section does not have a previous working section, a pre-compaction test will be carried out before the formal compaction operation. A pre-working section length is set, and the asphalt mixture temperature data collected in real time in the pre-working section is used to predict and generate the asphalt mixture temperature decay curve to guide the first working section. The specific implementation is as follows:

[0070] Before the first operation section, a pre-operation section is set up. After the paver unloads the material, the roller performs compaction in the pre-operation section. During operation, the roller obtains the outward radiated surface temperature of the asphalt mixture through an infrared temperature measuring camera, and then uses the asphalt mixture actual temperature real-time monitoring model to obtain the actual temperature of the asphalt mixture in real time, and uploads the data records to the construction personnel operation subsystem.

[0071] Since the compaction operation needs to be carried out within the optimal compaction temperature range of the asphalt mixture, the above-mentioned measured data is incomplete. Therefore, the actual temperature of the asphalt mixture measured by the real-time monitoring model of the actual temperature of the asphalt mixture is transmitted to the asphalt mixture temperature attenuation curve prediction model to predict the temperature of the asphalt mixture in the next period of time. The temperature attenuation curve is completed and the optimal compaction temperature range and time window of the first operation section are determined by this temperature attenuation curve.

[0072] When operating in the first working section, the real-time monitoring model of the actual temperature of the asphalt mixture is used to predict and record the real-time actual temperature data of the asphalt mixture. The temperature of the asphalt mixture in a later period of time is predicted according to the prediction model of the temperature decay curve of the asphalt mixture, which is used to generate a new temperature decay curve. The new temperature decay curve is used to determine the optimal compaction time window for the second working section, thereby guiding the construction of the second working section, and so on, until the end of the operation.

[0073] When performing asphalt mixture compaction operations, the area is first divided into compaction zones for each roller according to the number of rollers. Subsequent groups of machines will perform independent compaction operations on their respective compaction zones.

[0074] After dividing the compaction zones, path planning is performed for each roller in its own compaction zone. Path planning includes path planning for the straight rolling area and the lane change area. Since the compacted material asphalt mixture is temperature sensitive, path planning is performed based on the time window. The total operation time in a work section passing through the straight rolling area is set to t1, the total time passing through the lane change area is set to t2, and the total operation time of a work section is T, T = t1 + t2. It is necessary to ensure that T <T max , T max is the optimal compaction time window of asphalt mixture under the current environmental conditions. If the total time T of a working section exceeds T max , it means that the compaction operation is not completed in the optimal compaction temperature range of the asphalt mixture, which may lead to quality problems in the part that is not compacted in the optimal compaction temperature range. Therefore, it is necessary to ensure that the compaction time period of the entire operation section is completed within the optimal compaction time window. However, T that is too close or too conservative may have some problems. If T is too close to T max , the time loss caused by some unknown factors in the operation of the cluster may cause the final compaction time of the operation section to exceed T max This is undesirable, therefore, T<0.9T is specified max If T is set too conservatively, the straight rolling area of ​​each working section may be too short, there may be too many working sections, and the total compaction time may increase. If the straight section working length is too short, the roller may frequently change lanes, which may reduce construction efficiency and increase fuel consumption. Therefore, in order to save time and reduce fuel consumption of the roller, a minimum straight section working length should be set for each working section to ensure that T>0.75T max ; Therefore, the final setting range of T is: 0.75T max <T<0.9T max .

[0075] For a certain operation section, T max is a certain value, T takes a certain multiple of T maxFor the pre-operation stage, take T = 0.85T max , T=t1+t2, Where L is the length of the straight rolling area in the working section, that is, the length of the working section; V is the roller speed; q is the number of straight rolling lanes that a roller is responsible for in the working section.

[0076] The lane-changing curve is a third-order Bezier curve. The total lane-changing time is affected by the lane-changing length, lane-changing width, and Bezier curve control points. In the pre-operation section, the lane-changing length, lane-changing width, and control point coordinates of the lane-changing area are given first. During the operation of the pre-operation section, it is determined whether the lane-changing curve needs to be optimized, and the total lane-changing time t2 is recorded to calculate t1, and then L is calculated according to the above formula. If the lane-changing curve needs to be optimized, the optimized lane-changing curve is applied to the next operation section, and the total lane-changing time t2 is re-recorded.

[0077] Assuming that the area for path planning is a rectangular area, the rectangular area is divided into a straight rolling area and a lane-changing area through path planning. The straight rolling area includes a number of parallel rolling strips, and the lane-changing area includes a number of lane-changing curves. The lane-changing curve connects two adjacent straight rolling areas (see Figure 1 ). In traditional compaction operations, the roller reciprocates on a rolling strip, and after the strip is completely rolled, it changes lanes and rolls the next strip. However, due to the material properties of asphalt mixtures that are easy to cool, rolling using this method will result in reduced efficiency and consume a lot of time. The unrolled strips behind are easy to cool, which affects the compaction quality. Therefore, two lane change areas will be planned during path planning, located at the starting position and the end position of the operation section, respectively. The starting position is the lane change area 1, and the end position is the lane change area 2. The operation of the roller will be carried out in sequence: straight line operation-lane change-straight line operation, that is, firstly, the straight line rolling of the first strip is carried out, and then the lane change area 2 is entered for lane change, and the straight line rolling is continued after the lane change to the second rolling strip, and the lane change is carried out after the rolling to the lane change area 1, and the straight line rolling is continued after the lane change to the third rolling strip, and so on, until all the rolling strips in the compaction area of ​​the roller are compacted, and the compaction operation of the operation section is ended.

[0078] In detail, in step 3, the lane change area curve selects a Bezier curve. n The n-order Bezier curve B(t) can be expressed as follows:

[0079]

[0080] The lane-changing curve of the present invention adopts a third-order Bezier curve, and the control points are P0, P1, P2, and P3. The formula of the third-order Bezier curve in a two-dimensional plane rectangular coordinate system is:

[0081]

[0082] In the formula, (x i ,y i ), i = 1, 2, 3, 4... are the coordinate information of each selected control point. The third-order Bezier curve requires four control point coordinates. y is the lateral displacement of the lane change process (perpendicular to the straight rolling direction), x is the longitudinal displacement of the lane change process (straight rolling direction); t is the curve parameter.

[0083] A lane-changing curve is generated based on the coordinates of the control points. The initial control point coordinates will select the lane-changing starting point coordinate P0, the lane-changing end point coordinate P3, and the two midpoint coordinates (P1, P2) on the longitudinal driving direction line. The positions of the coordinate points will be adjusted later to obtain a more optimal lane-changing curve.

[0084] Step 4: After the compaction operation of the current operation section is completed, the temperature decay curve of the asphalt mixture under the current environmental conditions is predicted based on the asphalt mixture temperature data recorded and saved in the operation section. Based on the temperature decay curve, the optimal compaction time window of the new operation section is re-determined to optimize the operation length of the new operation section.

[0085] The unmanned operation of the roller fleet is completed through the coordination of various subsystems. The specific steps are as follows: 1. Coordinate acquisition. The operator collects the coordinates of the four corner points of the rectangular compaction area and inputs them into the construction personnel operation subsystem. The construction personnel operation subsystem converts the collected longitude and latitude coordinates into two-dimensional plane rectangular coordinates through the Gauss Kruger orthographic projection calculation method. All subsequent collected coordinates will undergo the projection coordinate conversion. 2. Divide the area. After the coordinates of the four corner points are input into the construction personnel operation subsystem, the rectangular rolling area will appear on the page. Then the compaction partition is divided, and the compaction area is divided into compaction partitions with the same number of rollers in the fleet along the compaction direction, and the length and width of each compaction partition are equal. 3. Determine the length of the working section. The compaction area is divided into several working sections along the direction of the roller. The length of the working section that needs to be compacted is determined by the temperature attenuation curve generated by the real-time temperature data of the asphalt mixture monitored in real time in the previous working section. If it is the first working section compaction operation, the temperature attenuation curve generated by the temperature data of the asphalt mixture in the pre-working section is used to determine the length of the working section. 4. Path planning. All rollers in the first working section are planned for their respective compaction zones. According to the compaction zone width, roller width, overlapping distance of adjacent rolling strips, etc., all compaction zones are divided into several parallel rolling strips, and lane change area 1 and lane change area 2 are set at the beginning and end of the working section. According to the initial lane change parameters and points, the lane change curve equation is obtained. At this time, the construction personnel operation subsystem will obtain the straight line equations of all rolling strips, the curve equations of the lane change curves, and a large number of coordinate points on the straight line and curve. The construction personnel operation subsystem will transmit the equations and points to the decision module through the transmission module. Through the data transmission of the sensor module and the calculation of the decision module, the roller will be controlled to operate according to the given route. 5. Paving machine unloading. After the path planning, the compaction operation begins. First, the paver enters the site to pave the asphalt mixture. The paver is equipped with a paver control subsystem composed of a positioning module and a control module for controlling its own work. After the paver unloads to the length L of each working section, the paver is prompted to stop unloading. The construction personnel operation subsystem can display the movement direction, trajectory, distance of the paver, and the temperature of the asphalt mixture. 6. Compacting operation of the machine group. After the unloading is completed, the machine group starts to enter the site for compacting operations. The machine group performs compaction operations according to the given path of the path planning. At the same time, all parameters of the roller and all parameters of the compacted material will be displayed and saved on the construction personnel operation subsystem interface during operation, which is convenient for subsequent path optimization. The roller parameters mainly include: the roller's travel speed, vibration frequency, relative turning angle of the vehicle body, lateral error, heading error, turning angle error, etc.; the parameters of the compacted material mainly include: the predicted temperature of the material temperature attenuation curve, and the actual temperature of the asphalt mixture predicted by the infrared temperature measuring camera when the roller rolls the asphalt mixture using the real-time monitoring model of the actual temperature of the asphalt mixture. 7. Control of the machine group operation.The operation of the fleet is controlled by the construction personnel operating subsystem. The temperature decay curve can reflect the change of the temperature of the asphalt mixture over time in the current or recent period, but the environmental conditions may change at any time. Therefore, according to the difference between the actual temperature of the asphalt mixture predicted by the real-time monitoring model of the actual temperature of the asphalt mixture and the temperature value at the corresponding time point on the temperature decay curve during the actual operation of the fleet, the change of the current environment is judged, so as to control the operation of the fleet. For example: if the temperature predicted by the real-time monitoring model of the actual temperature of the asphalt mixture is less than the temperature value of the point on the temperature decay curve, and the difference is too large, it means that the change of the current environment has accelerated the cooling of the asphalt mixture, and the roller is accelerated to ensure that the roller completes the compaction operation within the optimal compaction time window. 8. Optimization of path planning. After the fleet completes the operation of the previous operation section, it will adjust and optimize according to the current environmental conditions, the operating parameters of the roller in the previous operation section, and the parameters of the compacted material. According to the current environmental conditions, the temperature decay curve is re-determined, and then the optimal compaction time window for the new operation section is determined.

[0086] The path in the lane-changing area will also be optimized. According to practice, when the roller is turning, the greater the relative turning angle of the body, the slower the roller speed, which leads to an increase in the total lane-changing time. Therefore, if it is found that the speed of the roller when changing lanes is much lower than the speed during normal operation, it means that the lane-changing curve is unreasonable, and optimization is required to ensure that the lane-changing curve is relatively smooth and the curvature of the curve changes little. Continuously adjusting the control points of the Bezier curve can change the curve equation to achieve the desired smoothing effect. Fix the two end points P0 and P3 of the third-order Bezier curve, and change the two middle points of the third-order Bezier curve. Finally, a trajectory that satisfies the kinematic constraints, initial posture state constraints, target posture state constraints, and curvature continuity constraints can be obtained. The optimal trajectory is found by solving the optimal parameters that meet the above conditions.

[0087] The optimization function used is: J(P1,P2)=k max (t a )-k min (t b ),t a ,t b ∈(0,1), t a ,t b The value when the maximum and minimum curvature radii of the Bezier curve are obtained, and k represents the curvature radius.

[0088] The physical meaning of the optimization function is to optimize the parameters P1 and P2 so that the difference between the maximum curvature radius and the minimum curvature radius of the planned curve is minimized, thereby ensuring that the obtained curve trajectory is relatively smooth.

[0089] If it is found that the speed of the roller is still different from the normal driving speed of the roller when changing lanes, it means that the length of the lane changing area is set too small. At this time, the lane changing starting point P0 will be kept unchanged, and the end point P3 position will be adjusted to increase the length of the lane changing area. The above process will be repeated to optimize P1 and P2, and the length of the lane changing area will be increased by 0.5m each time.

[0090] In the present invention, the optimized lane changing curve obtained from the data of the previous working section is used in the next working section. When working according to this optimized lane changing curve, the speed difference between the real-time speed during lane changing and the optimal compaction speed is recorded. At the same time, the current total lane changing time t2 is recorded, and it is repeatedly determined whether curve optimization is needed to determine the total lane changing time t2 of the optimized lane changing curve in the next working section.

[0091] When changing lanes in the pre-operation section, the difference between the set roller speed and the actual speed is calculated. If the difference is greater than 1km / h, it means that the curve needs to be optimized. The lane change curve will be optimized and used in the first operation section. After obtaining the total lane change time t2 of the pre-operation section, the time t1 of the straight rolling area of ​​the first operation section is obtained by T-t2. The optimal compaction speed v of the roller is determined based on experience, and the roller speed is fixed to the optimal compaction speed v. The length of the straight rolling area of ​​the first operation section is set to: L = v·t1 / q. When performing the first working section, the total lane changing time t2 of the first working section is recorded, and the difference between the set roller speed and the actual speed is calculated. If the difference is greater than 1 km / h, the lane changing curve needs to be optimized. When changing lanes in the second working section, the lane changing curve after optimization of the first section is used to change lanes, and the total lane changing time t2 of the second working section is recorded. The total operating time t1 of the straight rolling area of ​​the second working section is recalculated by T-t2. At this time, t2 is the total lane changing time of the first working section. The length of the straight rolling area of ​​all working sections is obtained by analogy.

[0092] If the difference between the set speed and the actual speed of the roller in the current working section is less than 1 km / h, it means that the lane change is smooth and the lane change curve of the next working section does not need to be optimized, and the lane change curve of the current working section is still used.

[0093] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A temperature-quality driven path optimization method for the entire process of road drone swarm operation, characterized in that: The method comprises the following: Obtain asphalt mixture temperature data under different environmental conditions, and use intelligent prediction algorithms to establish a prediction model that can predict the asphalt mixture temperature attenuation curve under different environmental conditions; The surface temperature and actual temperature data of asphalt mixture radiating outward during compaction under different environmental conditions are obtained, and a real-time monitoring model for the actual temperature of asphalt mixture is established using an intelligent prediction model. The relationship between the surface temperature and the actual temperature of asphalt mixture radiating outward during compaction under different environmental conditions can be obtained, and the actual temperature inside the asphalt mixture can be monitored in real time during compaction operations. When the fleet is performing compaction operations, the operation section is divided into a straight rolling area and a lane change area. When compacting in the straight rolling area, the asphalt mixture actual temperature real-time monitoring model is used to predict the actual temperature of the asphalt at the rolling position in real time and record it. Then, according to the current operating environment, the asphalt mixture temperature attenuation curve prediction model is used to predict the asphalt mixture temperature attenuation curve that meets the on-site environmental conditions. The optimal compaction time window T corresponding to the optimal compaction temperature range is determined on the attenuation curve. max ; Determine the length of the next operation section according to the optimal compaction time window. The total operation time T of the fleet in any operation section ranges from: 0.75T max <T<0.9T max , at this time T max The temperature decay curve is determined by the temperature decay curve predicted by the actual temperature data of the asphalt mixture in the previous operation section. The temperature decay curve of the asphalt mixture in the first operation section is predicted and generated by the actual temperature data of the asphalt mixture in the pre-operation section. Under the premise that the roller speed V remains unchanged, the operation section length is divided according to L=V(T-t2) / q. Where L is the length of the straight rolling area in the working section, that is, the length of the working section, t2 is the total time spent on changing lanes in the same working section; q is the number of straight rolling lanes that a roller is responsible for in the working section; When the machine fleet is actually operating, compaction operations are carried out according to the divided operating sections, and the difference between the actual temperature of the asphalt mixture predicted by the real-time monitoring model of the actual temperature of the asphalt mixture and the temperature value of the point predicted on the temperature attenuation curve is compared. The temperature difference threshold is set. If the difference is greater than the temperature difference threshold, the roller is accelerated to ensure that the roller completes the compaction operation within the optimal compaction time window.

2. The method according to claim 1, characterized in that During the compaction operation of the machine group, the path planning of rolling and lane changing is optimized. The straight rolling area includes several straight rolling lanes parallel to the roadside. The number of straight rolling lanes is determined by the road width, the wheel width of the roller, and the overlapping width of adjacent straight rolling lanes. The lane changing area includes several curved segments. The lane changing curve selects a third-order Bezier curve. The lane changing curve ensures that the roller straightens the body and the articulated steering angle is zero after changing lanes. The entire lane changing path is continuous and smooth, meeting the kinematic and dynamic constraints of the roller. At the same time, straight rolling and curved lane changing are performed alternately to ensure that the roller can roll all the asphalt mixture in the operation section in a relatively short time. Before compaction, a pre-operation section is first set for pre-compaction. The optimal compaction speed V of the roller is determined based on experience. The roller speed is fixed to the optimal compaction speed V, and a speed difference threshold is set. Compaction is performed in the pre-operation section. The speed difference between the real-time speed and the optimal compaction speed during lane change is recorded. The total lane change time t2 is also recorded. If the speed difference is greater than the speed difference threshold, the lane change curve of the next operation section needs to be optimized to obtain the optimized lane change curve. In each working section, the total working time t1 of the straight rolling area is determined by T-t2. At this time, t2 is the actual total lane changing time of the previous working section. The length of the current working section is determined according to L=v·t1 / q. The lane changing curves of different rolling lanes in the same working section are the same. The optimized t2 of each working section changes and shows a trend of shortening. Correspondingly, the length of the straight rolling area becomes larger, thereby optimizing the length of the working section.

3. The method according to claim 2, characterized in that The third-order Bezier curve is composed of the starting point P0, the end point P3 and two points P1 and P2 in the longitudinal driving direction. The optimization process of the lane-changing curve is as follows: first, the coordinates of P1 and P2 are optimized to minimize the difference between the maximum curvature radius and the minimum curvature radius of the planned curve, so that the roller can not only ensure the continuous and smooth lane-changing path during the entire lane-changing process, but also control the maximum value of the relative turning angle between the front and rear bodies of the roller within a certain range during the lane-changing process to ensure the efficiency of the roller's lane-changing; the optimization function of P1 and P2 is: J(P1, P2) = k max (t a )-k min (t b ),t a ,t b ∈(0,1), t a ,t b Get the maximum curvature radius k for the Bezier curve max With the minimum curvature radius k min The value at time; Secondly, when changing lanes, if it is found that after optimizing the control points P1 and P2, the speed difference between the real-time speed of the roller and the optimal compaction speed V of the roller is still greater than the speed difference threshold, it means that the length of the lane changing area is set too small. At this time, the lane changing starting point P0 will be kept unchanged, and the position of the end point P3 will be adjusted to increase the length of the lane changing area. The process of optimizing the coordinates of P1 and P2 and adjusting the position of the end point P3 will be repeated until the speed difference is controlled within the speed difference threshold. The length of the lane changing area increases by 0.5m each time.

4. The method according to claim 1, characterized in that: The asphalt mixture temperature attenuation curve prediction model includes an intelligent prediction algorithm and a fitting module. The input of the intelligent prediction algorithm is ambient temperature, ambient humidity and wind speed, and the output is the asphalt mixture temperature in the future. The intelligent prediction algorithm adopts at least one of LSTM, Transformer, BiLSTM, and GRU; the input of the fitting module is the historical asphalt mixture temperature before prediction and the asphalt mixture temperature after prediction, and the output is the attenuation curve.

5. The method according to claim 1, characterized in that The input of the intelligent prediction model is ambient temperature, ambient humidity, actual wind speed, vehicle speed, and the surface temperature radiating outward from the asphalt mixture, and the output is the actual temperature of the asphalt mixture; the intelligent algorithm adopts at least one of the BP neural network, the ANN neural network, and the RNN neural network.

6. A temperature-quality driven road drone swarm operation full process path optimization system, including a number of rollers, and a control module for assisting automatic driving, a host computer and a decision module, characterized in that: The roller is equipped with a monocular camera for detecting the curbs on both sides to prevent the roller from running out of the working area and causing danger, an angle sensor for obtaining the relative turning angle of the vehicle body, a speed sensor for obtaining the driving speed of the roller, a thermometer and hygrometer for collecting ambient temperature and humidity, an infrared temperature measuring camera for collecting the surface temperature of the asphalt mixture radiating outward, a wind speed sensor for collecting relative wind speed, and an RTK GNSS for collecting the real-time position information of the roller; The host computer is used to view the real-time operation status of the fleet, including the visualization of the compaction of the fleet in the operation area, the compaction parameters of each roller in the fleet, the real-time temperature display of the compacted material, and the path planning based on the optimal compaction temperature range; The decision module communicates wirelessly with the host computer and can obtain data from the monocular camera, angle sensor, speed sensor, thermometer, infrared temperature camera, RTK GNSS, and wind speed sensor. It is also loaded with the asphalt mixture temperature attenuation curve prediction model and the asphalt mixture actual temperature real-time monitoring model. The asphalt mixture temperature decay curve prediction model is used to predict the asphalt mixture temperature decay curve under different environmental conditions; The real-time monitoring model for the actual temperature of the asphalt mixture is used to obtain the relationship between the surface temperature radiated outward by the asphalt mixture during compaction operations under different environmental conditions and the actual temperature.

7. The system according to claim 6, characterized in that The decision module is also loaded with a lane-changing curve optimization algorithm, which is used to optimize the speed difference between the set speed and the actual speed of the roller, and optimize the middle two control points and the end point on the third-order Bezier curve until the speed difference is controlled within the set speed difference threshold range; The length of the straight rolling area of ​​the next working section is determined according to the total lane changing time of the previous working section.

Images