Multi-machine cooperative unmanned rolling method and system based on joint compaction quality evaluation

Abstract

The application discloses a multi-machine cooperative unmanned rolling method and system based on joint compaction quality evaluation, relates to the field of civil engineering construction automation, and comprises the following steps: acquiring sensor acquisition point cloud data of an unmanned rolling machine, including IMU data and positioning data; a joint compaction model receives the acquisition point cloud data and outputs parameter control of a group of unmanned rolling machines on a planned path; the joint compaction model comprises the following steps: establishing a joint compaction quality evaluation model, adjusting a multi-machine cooperative control strategy based on the evaluation, and then realizing dynamic adjustment in a predictive control framework; the application quantifies the compaction state of an overlapping area, proposes a three-machine cooperative phase matrix control algorithm, optimizes the vibration energy superposition effect, and realizes dynamic closed-loop regulation and control of rolling parameters.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of civil engineering construction automation, in particular to a multi-machine cooperative unmanned rolling method and system based on joint compaction quality evaluation. BACKGROUND

[0002] In the construction of rock-fill dams, the compaction work of the roller is crucial to the quality of the dam body. Traditional rolling operations rely on manual operation, which has low efficiency, large quality fluctuations, and safety hazards. Although unmanned rolling machines have appeared in existing technologies, there are still problems when multiple machines work together:

[0003] 1. When multiple machines work together, the compaction quality of the overlapping area between the rolling wheels is difficult to guarantee;

[0004] 2. There is a lack of real-time evaluation and dynamic adjustment of the compaction quality of multiple machines working together;

[0005] 3. Existing systems cannot optimize the cooperative control parameters based on real-time collected data (such as point cloud, IMU data).

[0006] Although the existing technology (CN112982278A) proposes a single-machine control system for unmanned rolling machines, it does not solve the problem of vibration interference and compaction uniformity when multiple machines work together. Technology (CN113550248B) involves compaction degree evaluation, but does not establish an overlapping area interaction effect model.

[0007] Therefore, there is a need for a system and method that can evaluate the joint compaction quality in real time and dynamically adjust the cooperative control parameters of multiple machines. SUMMARY

[0008] The present application provides a multi-machine cooperative unmanned rolling method and system based on joint compaction quality evaluation, which solves the problems of existing technologies.

[0009] In a first aspect, the present application provides a multi-machine cooperative unmanned rolling method based on joint compaction quality evaluation, comprising:

[0010] Obtaining sensor-collected point cloud data of unmanned rolling machines, including IMU data, positioning data;

[0011] The joint compaction model receives the collected point cloud data and outputs parameter control of a group of unmanned rolling machines on the planned path. The joint compaction model includes: establishing a joint compaction quality evaluation model, adjusting the multi-machine cooperative control strategy based on the evaluation, and then implementing dynamic adjustment in a predictive control framework;

[0012] Among them, a group of unmanned rolling machines are provided with an overlapping area for same-direction rolling work;

[0013] The multi-machine cooperative control strategy comprises a feedback strategy for an interaction item of point clouds in a joint compaction degree in a joint compaction quality evaluation model output.

[0014] The prediction control framework predicts, by a model, a set of control parameters of the unmanned roller compactor, corrects the model in a cycle prediction according to feedback data after the unmanned roller compactor executes the control parameters, and performs rolling optimization.

[0015] Further, the set of unmanned roller compactors comprises three unmanned roller compactors that travel in the same direction, and a rolling path in sequence is adjusted according to a rolling wheel.

[0016] In the set of unmanned roller compactors, the second roller compactor covers the right half of the rolling wheel of the first roller compactor, and the third roller compactor covers the left half of the rolling wheel of the first roller compactor.

[0017] In another set of unmanned roller compactors, the second roller compactor covers the left half of the rolling wheel of the first roller compactor, and the third roller compactor covers the right half of the rolling wheel of the first roller compactor.

[0018] Further, the joint compaction quality evaluation model is specifically:

[0019] The joint compaction quality evaluation model receives collected data of all roller compactors, establishes a joint compaction quality evaluation model, calculates a joint compaction degree of the overlapping area, and outputs the joint compaction degree.

[0020] For the joint compaction degree calculation, the overlapping area of adjacent roller compactors is counted, and the joint compaction degree is calculated. K joint The calculation is as follows:

[0021]

[0022] Wherein: K 1 , K 2 The compaction degrees of the two roller compactors in the overlapping area are respectively K interaction is an interaction item, which is used to reflect the influence of the joint action of the two roller compactors on the compaction quality, α , β , gamma is a weight coefficient.

[0023] Further, for the interaction item:

[0024]

[0025] Wherein:

[0026] Δ v : The speed difference between the two roller compactors is measured by GNSS in real time.

[0027] Delta A : Amplitude difference, measured by vibration axis strain sensor

[0028] : Vibration phase difference, measured by vibration wheel accelerometer FFT analysis

[0029] delta k : Overlap depth, calculated as point cloud elevation rate of change

[0030] c1, c2, c3 are weight parameters.

[0031] Further, the joint compaction quality evaluation model is established, and a multi-machine cooperative control strategy is adjusted based on the evaluation, and then dynamic adjustment is realized in a predictive control framework, wherein the multi-machine cooperative control strategy comprises:

[0032] A set of phase coordination control parameters of the unmanned roller compactor is set in advance, and the vibration phases of the three unmanned roller compactors are set to a specific relationship through the compaction degree matrix as follows:

[0033] First: 0°, second: 180°, third: 90°.

[0034] Further, for the interaction effect term, the number of compaction passes is also set to compensate for the overlap depth, which is calculated according to the point cloud elevation rate of change delta k , and the number of compaction passes n_overlap is calculated by the following formula:

[0035] n_overlap = n_base + ceil( delta k / 0.02);

[0036] Wherein, n_base represents the preset basic compaction pass, and ceil is the ceiling function.

[0037] Further, for the parameter delta k , the specific calculation includes:

[0038] The current actual compaction degree is calculated by fusing point cloud data, GNSS elevation measurement and CMV value K actual , and the target compaction degree K ref is determined according to the engineering design requirements.

[0039] ​Furthermore, for the predictive control framework, the predictive control framework uses model prediction to continuously optimize a set of control parameters for the unmanned compactor. After the unmanned compactor executes the control parameters, the model is corrected and the prediction is cyclically made based on feedback data. The specific steps include the following:

[0040] The sensor-collected data is loaded, processed through a gridded point cloud, and then fused using Kalman filtering to obtain the actual compaction degree. A compaction degree prediction model is then constructed as follows:

[0041]

[0042] in, t For the current sampling period, u t These are the control variables within the current sampling period. The control variables include velocity, amplitude, and phase, which correspond to... v, A, d t The disturbance is determined by the water content. w Material temperature T Sure;

[0043] A 5-channel network was established using a spatiotemporal convolutional network. K,v,A, ,w Input convolutional layer, output ;

[0044] The objective function for establishing the rolling optimization engine is:

[0045]

[0046] Np Predict the number of time-domain steps. Nc Control the number of time-domain steps. Nc ≤ Np ; In time t + k Predicted compaction degree, : Control increment, λ is the weight matrix, where the control vector u t Includes control parameters for all rolling mills: The superscripts 1, 2, and 3 represent the first, second, and third rolling mills, respectively.

[0047] By optimizing the output layer of the spatiotemporal convolutional network using the objective function, the control parameters of the optimal compaction control sequence are obtained for each optimization. The control quantity at the current moment, i.e., the first control quantity, is applied to the unmanned compactor. At the next sampling moment, optimization is performed again based on the new sampling data until the compaction degree at the current completion moment reaches the target value. When the corresponding area is qualified, the next roller enters a new area, and the optimization step is repeated.

[0048] In a second aspect, the present application provides a multi-machine cooperative unmanned roller compaction system based on joint compaction quality evaluation, which is used to realize the method as described in any of the first aspect, comprising:

[0049] The multi-source sensor cluster arranged on the unmanned roller compaction machine comprises a laser radar, a GNSS positioning module, and an IMU inertial measurement unit.

[0050] The local computing unit deployed at the construction site is used as a node for realizing edge computing, performs real-time filtering on the point cloud data, converts the coordinate system from CGCS2000 to the construction coordinate system, and is used for feature extraction on the overlapping area in a group of unmanned roller compaction machines.

[0051] The cloud decision center based on cloud computing is used to execute the joint compaction quality evaluation model algorithm containing the interaction effect term calculation, execute the vibration phase matrix optimization, i.e., the 0° / 90° / 180° phase configuration, and generate the parameter dynamic regulation and control strategy under the predictive control framework.

[0052] The roller compaction machine terminal is the execution unit of the unmanned roller compaction machine, which responds to the instructions of the cloud decision center.

[0053] The multi-machine cooperative unmanned roller compaction method and system based on joint compaction quality evaluation provided by the present application establishes a joint compaction quality evaluation model containing an interaction effect term, quantifies the compaction state of the overlapping area, proposes a three-machine cooperative phase matrix control algorithm, optimizes the vibration energy superposition effect, constructs a predictive control framework, and realizes dynamic closed-loop regulation and control of the roller compaction parameters. BRIEF DESCRIPTION OF DRAWINGS

[0054] The accompanying drawings, which are included to provide a further understanding of the embodiments of the present application and constitute a part of the present application, do not constitute a limitation to the embodiments of the present application. In the drawings:

[0055] Figure 1 The system and data flow framework in the embodiments of the present application are shown in the accompanying drawings. DETAILED DESCRIPTION

[0056] The exemplary embodiments will be described in detail herein with reference to the accompanying drawings. Unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application.

[0057] The multi-machine cooperative unmanned roller compaction method and system based on joint compaction quality evaluation provided by the present application aims to solve the above technical problems of the prior art.

[0058] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0059] Example 1:

[0060] The system modules in this embodiment are as follows: Figure 1 As shown, it includes the following components: a data acquisition module, a local computing unit deployed at the construction site, a compactor terminal, a cloud-based decision center deployed based on cloud computing, which includes a joint compaction quality evaluation model, a multi-machine collaborative control module, and a communication module. The specific functions of each module are described below in order.

[0061] The data acquisition module consists of a multi-source sensor cluster installed on the unmanned roller compactor, including: LiDAR sensors (Light Detection and Ranging) deployed on each unmanned roller compactor, used to collect point cloud data of the environment around the roller compactor; IMU (Inertial Measurement Unit) used to collect attitude and motion data of the roller compactor; and GNSS positioning module used to provide real-time location information of the roller compactor.

[0062] The local computing unit deployed at the construction site serves as a node for edge computing. It performs real-time filtering of point cloud data and transforms the coordinate system from CGCS2000 to the construction coordinate system. It is also used to extract features from overlapping areas in a group of unmanned compactors, calculate the compaction degree, and generate a compaction degree matrix to be uploaded to the cloud decision center.

[0063] In a cloud-based decision center deployed using cloud computing, the joint compaction quality evaluation model receives data collected from all rollers, establishes a joint compaction quality evaluation model, and outputs the joint compaction degree (including the interaction effect term of the overlapping area). The core of the model is to calculate the joint compaction degree of the overlapping area and consider the interaction effect between adjacent rollers.

[0064] The joint compaction quality evaluation model specifically includes: point cloud data processing: preprocessing the point cloud data collected by each roller (denoising, registration, coordinate transformation) to extract the point cloud of the roller action area.

[0065] Combined compaction degree calculation:

[0066] For the overlapping area of ​​adjacent compactors (such as the first and second, or the first and third), since a group of unmanned compactors may be configured such that the second compactor covers the right half of the first compactor's roller and the third compactor covers the left half of the first compactor's roller, there will inevitably be unequal overlapping areas in each group. Therefore, the combined compaction degree... Kjoint The calculation is as follows:

[0067]

[0068] Wherein: K 1 , K 2 The compaction degree of the two rollers in the overlapping area, K interaction is the interaction effect term, reflecting the influence of the joint action of the two rollers on the compaction quality, and its calculation depends on the relative position of the two rollers, vibration parameters (phase difference, amplitude difference, speed difference) and material properties. α , β , gamma is the weight coefficient, which is obtained by field calibration.

[0069] Interaction effect term:

[0070]

[0071] Specifically:

[0072] Wherein:

[0073] Δ v : The speed difference between the two rollers, measured by GNSS real-time speed;

[0074] Δ A : Amplitude difference, measured by vibration shaft strain sensor;

[0075] : Vibration phase difference (°), analyzed by FFT of vibration wheel accelerometer;

[0076] delta k : Overlapping depth (calculated according to the width of the roller and the overlapping ratio), calculated as the point cloud elevation change rate, calculated by the fusion of point cloud data, GNSS elevation measurement and CMV value to calculate the current actual compaction degree K actual , according to the design requirements of the project to determine the target compaction degree K ref , the calculation gap is delta k = K ref - K actual ; delta k Also used to calculate the rolling variable n_overlap: n_overlap =n_base + ceil( delta k / 0.02); wherein, n_base represents a preset base compaction pass, ceil is a rounding up function, and the ceil function ensures that even a small gap is increased by at least one pass of compaction, avoiding insufficient compaction due to fractional passes;

[0077] c1, c2, c3 are weight parameters;

[0078] By combining the compaction quality evaluation model, the compaction quality of the overlapping area is realized in real time, and the weak link in multi-machine cooperative operation is solved.

[0079] Multi-machine cooperative control module: based on the joint compaction quality evaluation result, generate control parameter adjustment instructions for each roller. The control strategy is dynamically adjusted based on the output of the joint compaction quality evaluation model:

[0080] Compaction degree feedback control: if the joint compaction degree of the overlapping area is lower than the threshold, adjust the control parameters of the related roller to realize dynamic adjustment of the compaction parameters (including vibration phase), optimize the superposition effect of the vibration wave, and improve the compaction uniformity. For example:

[0081] Reduce the speed to improve the compaction effect;

[0082] Increase the amplitude or adjust the vibration frequency;

[0083] Increase the compaction pass (especially for areas with insufficient compaction).

[0084] Phase coordination control: in order to avoid the mutual offset of the vibration wave, the vibration phase of the three rollers is set to a specific relationship:

[0085] First: 0° (reference phase)

[0086] Second: 180° (opposite phase to the first, reduce interference)

[0087] Third: 90° (forms a phase difference with the first and second, promotes constructive superposition);

[0088] The phase control matrix of a group of unmanned rollers is set as:

[0089]

[0090] Where, - means none.

[0091] Path adjustment: dynamically adjust the overlapping width of the compaction path (for example, from 50% to 55%) according to the compaction quality of the overlapping area.

[0092] Communication module: realize data transmission between rollers and between rollers and control center.

[0093] Take three unmanned rollers (R1, R2, R3) as an example, and explain the implementation steps in detail:

[0094] Step 1. Initialization:

[0095] Set the rolling path so that R2 covers the left half of the R1 roller and R3 covers the right half of the R1 roller.

[0096] Set the initial parameters: speed 3 km / h, amplitude 1.2 mm, vibration frequency 30 Hz, R1 phase 0°, R2 phase 180°, R3 phase 90°.

[0097] Step 2. Data collection and transmission:

[0098] Each roller collects LiDAR point cloud data (including overlapping areas), IMU data (attitude, acceleration), and GNSS positioning data in real time.

[0099] Through vehicle-to-vehicle communication (DSRC) and 5G network, the data is transmitted to the joint compaction quality evaluation module.

[0100] Step 3. Joint compaction quality evaluation:

[0101] Preprocess the point cloud data and extract the point cloud of the overlapping area (R1 and R2, R1 and R3).

[0102] Calculate the joint compaction degree of each overlapping area K joint , especially the interaction term K interaction .

[0103] Step 4. Control parameter adjustment:

[0104] If the compaction degree of a certain overlapping area K joint <0.95 (target value), adjust the parameters of the corresponding roller:

[0105] If the R1-R2 overlapping area is insufficient, reduce the speed of R1 and R2 (for example, to 2.8 km / h) and increase the rolling number of R2 by 1.

[0106] If the R1-R3 overlapping area is insufficient, adjust the phase of R3 (for example, from 90° to 100°) to optimize the vibration superposition effect.

[0107] Adjust the command through the communication module and issue it to the corresponding roller.

[0108] Step 5. Repeat steps 2-4:

[0109] After each rolling operation, rescan the point cloud, update the compaction quality evaluation, and continue until all areas meet the requirements.

[0110] Example 2:

[0111] On the basis of Example 1, the calculation method of weight coefficient α , β , gamma is calculated as follows:

[0112] Field calibration steps:

[0113] The material area is selected as the main rock pile area (particle size 40-80 cm) and the cushion area (particle size <8 cm), and the control variables are fixed: moisture content (20±1%), paving thickness (1.0 m), and rolling speed (3 km / h). For data K 1 , K 2 , the water irrigation method is used to test the pit diameter of 2 m, the frequency is 3 points per area, and the accuracy requirement is ±0.01 g / cm³. For data K joint , the nuclear density instrument is used to collect one point every 0.5 m² with depth mode, and the accuracy requirement is ±0.02 g / cm³. For data Δ v , The GNSS speed difference operation is used, with 10 times per second and an accuracy requirement of ±0.05 km / h. For data The vibration wheel accelerometer FFT analysis is used, with 10 times per second and an accuracy requirement of ±1°.

[0114] The equation set is established as follows:

[0115]

[0116] The parameters are inverted by constraint optimization, and the algorithm is as follows:

[0117] from scipy.optimize import least_squares

[0118] # define residual function

[0119] def residual(params, K1, K2, K_int, K_joint):

[0120] alpha, beta, gamma = params

[0121] return K_joint - (alpha*K1 + beta*K2 + gamma*K_int)

[0122] # add physical constraints

[0123] constraints = (

[0124] [0, 0, 0], # lower bounds for alpha, beta, gamma

[0125] [1, 1, 1], # upper bounds for alpha, beta, gamma

[0126] {'type': 'eq', 'fun': lambda x: x[0]+x[1]+x[2]-1} # alpha + beta + gamma = 1 )

[0128] result = least_squares(residual, [0.3,0.3,0.4], bounds=constraints,

[0129] args=(K1_data, K2_data, K_int_data, K_joint_data))

[0130] Example 3: On the basis of example 1, increase the predictive control framework, adjust the multi-machine cooperative control strategy based on evaluation, realize dynamic adjustment in the predictive control framework, the steps are as follows:

[0131] Through model prediction, a group of unmanned roller control parameters are optimized, and the model is corrected according to the feedback data after the unmanned roller executes the control parameters.

[0132] The architecture design is:

[0133] State estimator-prediction model-optimization engine-execution period-sensor-state estimator;

[0134] The specific implementation steps are as follows:

[0135] Obtain the actual compaction degree data to determine whether the target area compaction degree is met, and load the compaction degree prediction model as:

[0136] Wherein, t is the current sampling period, u t is the control amount in the current sampling period, including speed, amplitude and phase, corresponding to v, A, , d t is the disturbance, determined by the water content w and the material temperature T ;

[0137] Through the space-time convolution network, 5 channels K,v,A, ,w Input convolutional layer, output ;

[0138] Establish a rolling optimization engine, where the objective function of the rolling optimization engine is:

[0139]

[0140] in, Np Predict the number of time-domain steps. Nc Control the number of time-domain steps. Nc ≤ Np ; In time t + k Predicted compaction degree, : Control increment, λ is the weight matrix, where the control vector u t Includes control parameters for all rolling mills:

[0141] The superscripts 1, 2, and 3 represent the first, second, and third rolling mills, respectively.

[0142] Execution period:

[0143] By optimizing the output layer of the spatiotemporal convolutional network through the objective function, the control parameters of the optimal control sequence of the rolling mill in each optimization are obtained. The control quantity at the current moment, i.e. the first control quantity, is taken and applied to the unmanned rolling mill, and the rolling mill terminal responds to the instructions of the cloud decision center.

[0144] Then, before re-optimizing based on new sampling data acquired through sensors, the state estimator function is executed until the compaction degree reaches the target value at the current completion point. Once the corresponding area is deemed satisfactory, the next set of compactors enters the new area, and the optimization steps are repeated. In this embodiment, the system operates within a predictive control framework, achieving closed-loop control of "measurement-evaluation-adjustment," significantly improving compaction efficiency and quality.

[0145] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative.

[0146] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0147] In addition, each function module in the embodiments of the present application can be integrated in one processing module, or each module can exist physically, or two or more modules can be integrated in one module. The integrated module can be realized in the form of hardware, or in the form of hardware plus software function module.

[0148] It should be further understood that the terms "comprise", "comprising", or any other variation thereof are intended to cover a non-exclusive inclusion, such that processes, methods, articles, or apparatuses that comprise a list of elements do not include only those elements but can also include other elements not expressly listed or inherent to such processes, methods, articles, or apparatuses. Without further limitation, an element preceded by "comprises a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0149] The above only is the embodiment of the present application, and does not limit the present application. The present application can have various changes and modifications for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of claims of the present application.

[0150] Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed here. The present application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such steps, compositions, ingredients, products, and methods for carrying out results of the application or obtaining similar results. The specification and examples are to be regarded as exemplary only, and the true scope and spirit of the application are indicated by the following claims.

[0151] It should be understood that the present application is not limited to the precise construction that has been described above and illustrated in the accompanying drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the application is limited only by the claims that follow.

Claims

1. A multi-machine cooperative unmanned rolling method based on joint compaction quality evaluation, characterized in that, The method comprises the following steps: Obtaining sensor acquisition point cloud data of unmanned roller compactor, including IMU data and positioning data; The joint compaction model receives the acquisition point cloud data and outputs parameter control of a group of unmanned roller compactors on the planned path, and the joint compaction model comprises: establishing a joint compaction quality evaluation model, adjusting the multi-machine cooperative control strategy based on the evaluation, and then realizing dynamic adjustment in a predictive control framework; Wherein, the group of unmanned roller compactors is provided with an overlapping area for the same direction compaction operation; the group of unmanned roller compactors comprises three unmanned roller compactors traveling in the same direction, and the compaction path in sequence is adjusted according to the compaction wheel; in the group of unmanned roller compactors, the second compactor covers the right half of the compaction wheel of the first compactor, and the third compactor covers the left half of the compaction wheel of the first compactor; in another group of unmanned roller compactors, the second compactor covers the left half of the compaction wheel of the first compactor, and the third compactor covers the right half of the compaction wheel of the first compactor; Wherein, the multi-machine cooperative control strategy comprises a feedback strategy for the interaction effect item of the joint compaction degree in the overlapping area point cloud output by the joint compaction quality evaluation model; Wherein, the predictive control framework predicts through a model, rolls and optimizes the control parameters of the group of unmanned roller compactors, and corrects the model for cyclic prediction according to the feedback data after the unmanned roller compactors execute the control parameters; For the interaction effect item: ； Wherein: Delta v : Speed difference between two rollers, measured in real time using GNSS Δ A : amplitude difference, measured by a vibrating axis strain sensor; : Vibration phase difference, by FFT analysis of vibration wheel accelerometer c1, c2, and c3 are weight parameters. k : overlap depth, calculated as point cloud elevation rate of change; The joint compaction quality evaluation model specifically comprises:

2. The multi-machine cooperative unmanned rolling method based on joint compaction quality evaluation according to claim 1, characterized in that, Receiving acquisition data of all compactors, establishing a joint compaction quality evaluation model, calculating the joint compaction degree of the overlapping area, and outputting the joint compaction degree; The joint compaction quality evaluation model is established based on the evaluation to adjust the multi-machine cooperative control strategy, and then dynamic adjustment is realized in a predictive control framework, wherein the multi-machine cooperative control strategy comprises: wherein, for the joint compactness calculation: the overlapping area of adjacent rollers is counted, the joint compactness K joint is calculated as follows: ； wherein: K 1 , K 2 respectively are the compaction degrees of the two rollers in the overlapping area, K interaction is the interaction effect term, which is used to reflect the influence of the joint action of the two rollers on the compaction quality, α , β , Pre-setting phase coordination control parameters of a group of unmanned roller compactors, setting the vibration phases of the three unmanned roller compactors to a specific relationship through a compaction degree matrix as follows: is the weight coefficient. 3.The method of claim 2, wherein, First: 0°, second: 180°, and third: 90°. Wherein, n_base represents a preset basic compaction pass number, and ceil is a rounding up function. For the predictive control framework, the predictive control framework predicts through a model, rolls and optimizes the control parameters of the group of unmanned roller compactors, and corrects the model for cyclic prediction according to the feedback data after the unmanned roller compactors execute the control parameters, specifically comprising the following steps:

4. The multi-machine cooperative unmanned rolling method based on joint compaction quality evaluation according to claim 3, characterized in that, For the interaction effect term, also included is a set of overlap passes to compensate for the overlapping depth, calculated as a function of the point cloud elevation rate of change Loading sensor acquisition data, processing data through gridded point cloud, fusing the processed data through Kalman filtering, obtaining the actual compaction degree, and constructing a compaction degree prediction model as follows: k The number of overlap passes n overlap is calculated by the following equation: ； v, A, 5. The multi-machine cooperative unmanned rolling method based on joint compaction quality evaluation according to claim 4, characterized in that, For the parameters K, v, A, k the specific calculations include: The current actual compaction degree is calculated by fusing point cloud data, GNSS elevation measurement and CMV value K actual The target compaction degree is determined according to the engineering design requirement K ref The gap is calculated as . 6.The method of claim 5, wherein, The objective function of the rolling optimization engine is established as follows: Np ； wherein, t is the current sampling period, u t is the control quantity in the current sampling period, the control quantity including speed, amplitude and phase, corresponding to Nc , d t is the disturbance quantity, determined by the water content w and the material temperature T ; 5-channel established by a spatio-temporal convolutional network Nc , w input convolutional layer, output ; Np ； The method for realizing the multi-machine cooperative unmanned compaction based on joint compaction quality evaluation according to any one of claims 1-6 comprises: : Predicted number of time steps, A multi-source sensor cluster arranged on the unmanned roller compactor, comprising: a laser radar, a GNSS positioning module, and an IMU inertial measurement unit; : Controlled number of time steps, ​ ≤ ​ ; : Predicted compaction degree at time t + k : Controlled compaction degree at time : Control increment, λ is a weight matrix, where the control vector u t Contains all the control parameters of the roller: Where the superscripts 1, 2, 3 represent the first roller, the second roller, and the third roller, respectively. Through the objective function, the output layer of the space-time convolution network is optimized to obtain the control parameters of the optimal roller control sequence of each optimization, and the control amount at the current time, i.e. the first control amount, is applied to the unmanned roller. At the next sampling time, the optimization is performed again according to the new sampling data until the current completion degree of compaction reaches the target value, i.e. If the corresponding area is qualified, the next group of rollers enters a new area, and the optimization step is repeated.

7. A multi-machine cooperative unmanned rolling system based on joint compaction quality evaluation, characterized in that, ​ ​ The local computing unit deployed at the construction site is a node for realizing edge computing, which performs real-time filtering on the point cloud data, converts the coordinate system from CGCS2000 to the construction coordinate system, and is used for feature extraction on the overlapping area of a group of unmanned roller machines; The cloud decision center deployed based on cloud computing is used for executing the joint compaction quality evaluation model algorithm containing the interaction effect term calculation, is used for executing the vibration phase matrix optimization, i.e. 0° / 90° / 180° phase configuration, and is used for generating the parameter dynamic regulation and control strategy under the prediction control framework; The roller machine terminal is an execution unit of the unmanned roller machine, which responds to the instruction of the cloud decision center.

Images