A full-face rolling construction quality assessment method based on machine learning

Abstract

The present invention discloses a method for evaluating the construction quality of full-warehouse surface rolling based on machine learning. First, the real-time three-dimensional spatial position data of the rolling machinery at the construction site is monitored, and the real-time rolling construction parameter information is obtained by using the spatial position data. Further, a real-time high-precision digital elevation model of the rolling warehouse surface is constructed to calculate the settlement value. Secondly, a rolling experiment is carried out, and the data of the compaction quality indexes under different rolling speeds and rolling passes are collected to obtain the sample data set of this method, and then a construction quality evaluation model for the rolling construction warehouse surface is established. Finally, during the construction process of the rolling warehouse surface, the real-time evaluation of the construction quality of the full-warehouse surface rolling is realized based on the model. The present invention can quickly and accurately estimate the compaction quality indexes of the full-warehouse surface of the rolling construction. Further, through the "double evaluation" of the rolling parameters and the compaction quality indexes, the compaction quality of the warehouse surface is comprehensively evaluated, and the accuracy problem of evaluating the rolling quality of the whole warehouse surface through a small number of samples can be effectively avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of process control management of rolling construction quality in earth-rock filling projects, and particularly relates to a method for evaluating the rolling construction quality of the entire bin surface based on machine learning. Background Art

[0002] In earth-rock filling projects, the compaction quality of filling materials is crucial for the stability and durability of structures. Therefore, quality control during the rolling construction process of earth-rock filling is a key link to ensure the construction quality and safety of structures. According to the current specification requirements, "dual control" should be achieved in the quality control of rolling construction: one is the "process control" during construction, mainly controlling the rolling parameters during construction (including paving thickness, water addition amount, vibration excitation force, rolling speed, number of rolling passes, etc.); the other is the "final parameter control" after the completion of bin surface construction, analyzing whether the dry density or compaction degree of the filling body meets the designed compaction quality standard by sampling test pits. Currently, some scholars have used modern advanced positioning technologies such as GNSS and surveying robots to achieve real-time, continuous, and high-precision automatic monitoring of rolling machinery during construction. After obtaining the three-dimensional coordinate data of the rolling machinery during construction operation, important control parameters of rolling construction have been further calculated through these data, achieving the purpose of monitoring the rolling construction quality. The rolling construction quality monitoring system in application has basically achieved the "process control" of construction, monitoring real-time vibration excitation force status, rolling speed, number of rolling passes, rolling layer thickness, etc., and providing real-time feedback for timely adjustment to guide construction and ensure that the rolling parameters meet the requirements. However, the construction quality of the entire bin surface is still evaluated by discrete test pit additional mass method or sand replacement method to detect samples. This way of substituting points for the whole is difficult to ensure the construction quality of the entire bin surface. Therefore, how to estimate the compaction quality index through the rolling parameters obtained by real-time monitoring and finally achieve the construction quality evaluation of the entire bin surface is a key technical difficulty for the development of the rolling construction quality monitoring system towards intelligence and unmanned operation. Summary of the Invention

[0003] To overcome the deficiencies of the prior art, based on the fact mechanism that "when the rolling layer tends to be in a compaction stable state, the surface settlement value before and after rolling decreases and approaches zero", through correlation analysis, the strong correlation between the compaction settlement value, number of rolling passes, rolling speed and the final compaction quality index is clarified. Further, aiming at the problem that the mathematical model between parameters such as compaction settlement value, number of rolling passes, rolling speed and the compaction quality index is unknown, a machine learning algorithm is used to find the best mapping function between the rolling parameters during construction and the final compaction quality index, and then estimate the compaction quality at any position on the entire bin surface.

[0004] Therefore, the technical solution adopted by the present invention is as follows:

[0005] A full-surface rolling construction quality evaluation method based on machine learning, comprising the following steps:

[0006] Step S1. Monitor the real-time three-dimensional spatial position data of the rolling machinery at the construction site, and calculate the real-time rolling construction parameter information from the three-dimensional spatial position data;

[0007] Step S2. Construct a real-time high-precision digital elevation model of the rolling surface based on an improved algorithm of the radial basis function neural network, and calculate the settlement value by the difference between the digital elevation models before and after rolling;

[0008] Step S3. Conduct a rolling experiment on-site for the earth-rock filling project. While the construction quality monitoring system monitors in real time, arrange a certain number of sampling points to collect the compaction quality index data under different rolling speeds and rolling passes, and obtain the sample data set of this method.

[0009] Step S4. Use machine learning to establish a mathematical model of the compaction quality index, settlement value, rolling passes, and rolling speed, that is, the construction quality evaluation model of the rolling construction surface;

[0010] Step S5. During the rolling surface construction process, realize the real-time evaluation of the construction quality of the entire rolling surface based on the model, analyze the areas with unqualified quality according to the filling construction quality control standards, give feedback in a timely manner, and control the construction quality.

[0011] Further, in the step S1, use the software and hardware of the intelligent system for monitoring the construction quality of filling and rolling to monitor the real-time three-dimensional spatial position data of the rolling machinery, and use the spatial data mining algorithm to calculate the real-time rolling construction parameter information from the spatial position data.

[0012] Further, the construction of the real-time high-precision digital elevation model of the rolling surface in the step S2 includes the following steps:

[0013] Step S21. Introduce the time dimension on the basis of the spatial dimension of the three-dimensional spatial position data. Since the rolling strips overlap partially with each other, some discrete points will be covered by the subsequent rolling strips. The overlapping part will be further compacted with the increase of the rolling passes, and thus the elevation will change. Therefore, the covered points can no longer represent the rolling surface, and new points need to be updated to replace them to represent the rolling surface. Therefore, the elevation point data on the final rolling surface is obtained through a continuous process of judgment and update.

[0014] S22. Based on the elevation data of the rolling surface extracted in the previous step, use the improved radial basis function neural network algorithm to establish a real-time high-precision digital elevation model of the rolling surface;

[0015] S23. Calculate the settlement value by the difference between the digital elevation models before and after rolling. The calculation formula is

[0016] ΔZ(i,j) = Z early (i,j) - Z later (i,j)

[0017] In the formula, i and j correspond to the row and column in the grid data, and Z later (i,j) is the elevation value after rolling, and Z early (i,j) is the elevation value before rolling.

[0018] Furthermore, the specific steps for obtaining sample data in the rolling experiment in step S3 are as follows:

[0019] S31. Conduct an earth-rock filling and rolling experiment before the filling and rolling work. Use the real-time monitoring system to obtain the main rolling parameter information during the construction process, including the number of rolling passes and the rolling speed. At the same time, arrange a certain number of sampling points to collect the compaction quality index data by the additional mass method;

[0020] S32. Select an area for the rolling experiment. During the experiment, collect the compaction quality index data under a certain number of different rolling speeds and numbers of rolling passes, and locate the corresponding rolling speed, number of rolling passes of each sampling data and the settlement value information obtained through step S2 by coordinates to obtain the sample data set of this method.

[0021] Furthermore, the specific steps for establishing a rolling construction face construction quality evaluation model based on machine learning in step S4 are as follows:

[0022] S41. Use the common regression algorithms in machine learning. Take the settlement value, number of rolling passes, and rolling speed as inputs and the compaction quality index as the output, and train the regression model respectively; conduct cross-validation, that is, divide the sample data set multiple times. One part is used as the training set to train the model, and the other part is used as the test set. Calculate in this way multiple times to finally evaluate the model effect;

[0023] S42. Through the training results and cross-validation results, select the regression algorithm with the best comprehensive fitting effect and cross-validation stability as the specific machine learning algorithm for estimating the rolling construction quality of the entire warehouse surface of this project, and use the sample data to obtain the rolling construction quality estimation model for the entire warehouse surface.

[0024] Furthermore, in step S5, based on the data of the rolling experiment, determine the quality control requirements in the filling and rolling construction of this project as follows

[0025]

[0026] In the formula, N is the number of rolling passes, N min is the minimum number of rolling passes, N max is the maximum number of rolling passes, v is the rolling speed, vmin is the minimum rolling speed, v max is the maximum rolling speed, Z is the settlement value, Z lim is the limit value of the settlement value, Q is the compaction quality index, Q min is the minimum compaction quality index; therefore, the final quality assessment criterion for the full - surface rolling construction is

[0027]

[0028] In the formula, S all refers to the total area of the full - surface construction, S(*) refers to the sum of the areas where a certain index on the full - surface meets the requirements, is the qualified rate statistically calculated for the full - surface, and p0 refers to the qualified rate limit specified according to the actual situation.

[0029] Furthermore, during the construction process, the quality assessment of the full - surface rolling construction based on machine learning includes the following steps:

[0030] S51. During the construction process, use the monitoring data of the real - time monitoring system to statistically obtain the rolling parameters information such as the number of rolling passes and rolling speed at any position on the full - surface. At the same time, construct a real - time high - precision model of the rolling surface to calculate the settlement value at any position on the full - surface, and take the settlement value, the number of rolling passes, and the rolling speed as input quantities into the model in step S2 to obtain the estimated value of the compaction quality index at any position on the full - surface;

[0031] S52. Statistically calculate the qualified rates of the rolling parameters and compaction quality indexes on the full - surface according to the quality control requirements in step S5, and evaluate the construction quality of the full - surface according to the quality assessment criterion of the full - surface rolling construction in step S5.

[0032] Furthermore, the improved radial basis function neural network algorithm in step S1 adds a robust algorithm on the basis of the radial basis function neural network algorithm.

[0033] Compared with the existing calculations, the characteristics and beneficial effects of the present invention are:

[0034] The present invention provides a method for quality assessment of full - surface rolling construction based on machine learning. Based on the fact mechanism that "when the rolling layer tends to be in a compaction - stable state, the surface settlement value before and after rolling decreases and approaches zero", it can quickly and accurately estimate the compaction quality index of the full - surface of the rolling construction. Further, through the "double - evaluation" of the rolling parameters and compaction quality indexes, the compaction quality of the surface is comprehensively evaluated, which can effectively avoid the problem of the accuracy of evaluating the overall surface rolling quality through a small number of samples, provides a solution for the rolling monitoring system to achieve "final parameter control", and is a strong support for the development of the filling and rolling construction quality monitoring system towards intelligence and unmanned operation, having important theoretical significance and application value. Brief Description of the Drawings

[0035] Figure 1 is the flowchart of the method of the present invention.

[0036] Figure 2 is the schematic diagram for extracting the elevation data of the rolling surface in the embodiment of the present invention.

[0037] Figure 3 is the example diagram of the real-time digital elevation model of the rolling surface in the embodiment of the present invention.

[0038] Figure 4 is the distribution diagram of the qualified situation of the settlement values of the whole surface in the embodiment of the present invention.

[0039] Figure 5 is the distribution diagram of the qualified situation of the number of rolling passes of the whole surface in the embodiment of the present invention.

[0040] Figure 6 is the distribution diagram of the qualified situation of the average rolling speed of the whole surface in the embodiment of the present invention.

[0041] Figure 7 is the distribution diagram of the qualified situation of the compaction quality index (dry density) of the whole surface in the embodiment of the present invention. Detailed implementation manners

[0042] In order to make the technical problems, technical solutions and beneficial effects to be solved by the embodiments of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0043] It should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the embodiments of the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0044] In the description of the present invention, unless otherwise specified, the term "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0045] Embodiment:

[0046] The method provided by the present invention can implement the process with computer software technology. See Figure 1In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments as follows:

[0047] A full-surface rolling construction quality evaluation method based on machine learning, comprising the following steps:

[0048] Step S1. Monitor the real-time three-dimensional spatial position data of the rolling machinery at the construction site, and calculate the real-time rolling construction parameter information from the three-dimensional spatial position data, including the number of rolling passes, rolling speed, rolling track, etc.;

[0049] Step S2. Construct a real-time high-precision digital elevation model of the rolling surface during construction based on an improved algorithm of a radial basis function neural network, and calculate the settlement value by the difference between the digital elevation models before and after rolling;

[0050] Step S3. Conduct a rolling experiment on the site of the earth-rock filling project. While the construction quality monitoring system monitors in real time, arrange a certain number of sampling points to collect the compaction quality index data under different rolling speeds and numbers of rolling passes to obtain the sample data set of this method,

[0051] Step S4. Use machine learning to establish a mathematical model of the compaction quality index, settlement value, number of rolling passes, and rolling speed, that is, a construction quality evaluation model for the rolling construction surface;

[0052] Step S5. During the rolling surface construction process, realize the real-time evaluation of the construction quality of the entire rolling surface based on the model, analyze the areas with unqualified quality according to the construction quality control standards for filling, and give timely feedback to control the construction quality.

[0053] In step S1 of the present embodiment above, the real-time three-dimensional spatial position data of the rolling machinery is monitored by using the software and hardware of the intelligent system for monitoring the construction quality of filling and rolling, and the real-time rolling construction parameter information is calculated from the spatial position data by using the spatial data mining algorithm.

[0054] The specific method for constructing the real-time high-precision digital elevation model of the rolling surface during construction in step S2 is as follows:

[0055] Step S21. Introduce a time dimension on the basis of the spatial dimension of the three-dimensional spatial position data. Since there is partial overlap between rolling strips, some discrete points will be covered by the subsequent rolling strips, such as Figure 2The points Pi+1, Pi+2, Pi+3 in it. In terms of the compaction mechanism, the overlapping part will be further compacted as the number of rolling passes increases, so the elevation will change. Therefore, the covered points can no longer represent the rolling surface, and new points Pj, Pj+1, Pj+2, Pj+3 need to be updated to replace them to represent the rolling surface. Specifically, at a certain moment, the elevation point data on the rolling surface is Pi, Pi+1, Pi+2, Pi+3, and after construction for a period of time, the elevation point data on the rolling surface becomes Pi, Pj, Pj+1, Pj+2, Pj+3. Therefore, the final elevation point data on the rolling surface is obtained through such a continuous judgment and update process;

[0056] S22. Based on the elevation data of the rolling surface extracted in the previous step, use the improved radial basis function neural network algorithm to establish a real-time high-precision digital elevation model of the rolling surface, as Figure 3 shown. The horizontal and vertical coordinates in the figure are the horizontal and vertical coordinate values in the construction layout coordinate system, and different gray levels in the figure correspond to different elevation values;

[0057] S23. Calculate the settlement value through the difference between the digital elevation models before and after rolling. The calculation formula is

[0058] ΔZ(i,j) = Z early (i,j) - Z later (i,j)

[0059] In the formula, i and j correspond to the rows and columns in the raster data, Z later (i,j) is the elevation value after rolling, and Z early (i,j) is the elevation value before rolling.

[0060] The specific steps for obtaining sample data in the rolling experiment in step S3 are as follows:

[0061] S31. Conduct an earth-rock filling rolling experiment before the filling and rolling work. Use the real-time monitoring system to obtain the main rolling parameter information during the construction process, including the number of rolling passes and the rolling speed. At the same time, arrange a certain number of sampling points to collect the compaction quality index data by the additional mass method. In this embodiment, the dry density is used as the final compaction quality index;

[0062] S32. Select an area for the rolling experiment. During the experiment, collect the dry density data at a certain number of different rolling speeds and rolling passes, and locate the corresponding rolling speed, rolling passes of each sampling data and the settlement value information obtained through step S2 by coordinates, and obtain a sample data in the form of

[0063] {settlement value, number of rolling passes, rolling speed, dry density data}

[0064] Finally, summarize to obtain the data set of this embodiment.

[0065] In step S4, the specific steps for establishing a construction quality evaluation model for the rolling construction surface based on machine learning are as follows:

[0066] S41. Using common regression algorithms in machine learning, with the settlement value, number of rolling passes, and rolling speed as inputs and the dry density as the output, regression model training is carried out respectively. Since the sample data volume for establishing the compaction quality regression model is small, in order to more conveniently find suitable model parameters, cross-validation is required, that is, the sample data set is divided multiple times, with a part used as the training set to train the model and another part used as the test set, and calculations are performed multiple times to finally evaluate the model effect.

[0067] S42. From the training results and cross-validation results, it can be seen that the gradient boosting regression algorithm has the best fitting effect among all models, manifested in being able to explain 99.99% of the variance changes and having the lowest values for each error term. In addition, in the 5 times of cross-validation tests, the results of this algorithm have relatively high stability, which also shows the stable effect of this algorithm in dealing with different data sets. Therefore, the gradient boosting regression model is the optimal model for estimating the rolling construction quality of the entire warehouse surface in local rockfill rolling construction. Through this model, the dry density at any position on the rolling surface can be estimated.

[0068] In step S5, corresponding to the "dual control" in quality control, "dual evaluation" should also be achieved in the rolling construction quality evaluation of the entire warehouse surface: both the qualification of the rolling parameters of the entire warehouse surface and the qualification of the compaction quality standard (dry density or compaction degree) of the entire warehouse surface should be evaluated. Only when the evaluations in both aspects meet the requirements can the surface in the monitoring system be evaluated as qualified. Based on the data of the rolling test, the quality control requirements in the filling and rolling construction of this project are determined as follows

[0069]

[0070] In the formula, N is the number of rolling passes, v is the rolling speed, Z is the settlement value, and Q is the compaction quality index. Therefore, the final evaluation criterion for the rolling construction quality of the entire warehouse surface is

[0071]

[0072] In the formula, S all refers to the total area of the entire warehouse surface construction, S(*) refers to the sum of the areas where a certain index on the entire warehouse surface meets the requirements, is the qualified rate statistically calculated for the entire warehouse surface, and p0 refers to the specified qualified rate limit according to the actual situation.

[0073] During the construction process, the rolling construction quality evaluation of the entire warehouse surface based on machine learning includes the following steps:

[0074] During the construction process, the compaction passes, compaction speed and other compaction parameter information at any position on the entire filling surface are statistically obtained by using the monitoring data of the real-time monitoring system. At the same time, a real-time high-precision model of the compaction filling surface is constructed to calculate the settlement value at any position on the entire filling surface. The settlement value, compaction passes and compaction speed are used as input quantities and brought into the model in step S2 to obtain the estimated value of the compaction quality index at any position on the entire filling surface;

[0075] S52. The qualification rates of the compaction parameters and compaction quality indexes on the entire filling surface are statistically obtained according to the quality control requirements in step S5, and the construction quality of the entire filling surface is evaluated according to the compaction construction quality evaluation criteria of the entire filling surface in step S5. The specific evaluation results are shown in Table 1, and the distribution diagrams of each evaluation result are as Figure 4 , Figure 5 , Figure 6 , Figure 7 shown. The horizontal and vertical coordinates in the figure are the horizontal and vertical coordinate values in the construction layout coordinate system. As can be seen from the table, neither the compaction passes in the compaction parameters nor the final compaction quality index meets the requirements. Therefore, the final evaluation conclusion of this filling surface is unqualified, and it is necessary to re-evaluate after additional compaction in the weak areas.

[0076] Table 1 Evaluation Results of the Construction Quality of the Entire Filling Surface in the Embodiment

[0077]

[0078] The specific embodiments described above are merely illustrative of the spirit of the present invention. Those skilled in the art to which the present invention pertains can make various modifications or supplements to the described specific embodiments or use similar methods for substitution, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A full-face rolling construction quality evaluation method based on machine learning, characterized in that, It includes the following steps: Step S1. Monitor the real-time three-dimensional spatial position data of the rolling machinery at the construction site, and calculate the real-time rolling construction parameter information from the three-dimensional spatial position data; Step S2. Construct a real-time high-precision digital elevation model of the rolling surface based on the improved algorithm of the radial basis function neural network, and calculate the settlement value by the difference between the digital elevation models before and after rolling; it includes the following steps: Step S21. Introduce the time dimension on the basis of the spatial dimension of the three-dimensional spatial position data. Since the rolling strips overlap partially, some discrete points will be covered by the subsequent rolling strips. The overlapping part will be further compacted with the increase of the rolling passes, and thus the elevation will change. Therefore, the covered points can no longer represent the rolling surface, and new points need to be updated to replace them to represent the rolling surface. Therefore, the elevation point data on the final rolling surface is obtained through a continuous process of judgment and update; S22. Based on the elevation data of the rolling surface extracted in the previous step, establish a real-time high-precision digital elevation model of the rolling surface by using the improved radial basis function neural network algorithm; S23. Calculate the settlement value by the difference between the digital elevation models before and after rolling, and the calculation formula is ΔZ(i,j) = Z early (i,j) - Z later (i,j) where i and j correspond to the rows and columns in the raster data, and Z later (i, j) is the elevation value after compaction, and Z early (i, j) is the elevation value before compaction; Step S3. Conduct a rolling experiment at the site of the earth-rock filling project. While monitoring in real time by the construction quality monitoring system, arrange a certain number of sampling points to collect the data of the compaction quality indicators under different rolling speeds and rolling passes, and obtain the sample data set of this method, Step S4. Use machine learning to establish a mathematical model of the compaction quality indicator, the settlement value, the number of rolling passes, and the rolling speed, that is, the construction quality evaluation model of the rolling construction surface; Step S5. During the construction process of the rolling surface, realize the real-time evaluation of the construction quality of the entire rolling surface based on the model, analyze the areas with unqualified quality according to the filling construction quality control standard, and give timely feedback to control the construction quality.

2. The full-warehouse surface rolling construction quality evaluation method based on machine learning according to claim 1, wherein In the step S1, the real-time three-dimensional spatial position data of the rolling machinery is monitored by using the software and hardware of the intelligent system for monitoring the construction quality of filling and rolling, and the real-time rolling construction parameter information is calculated from the spatial position data by using the spatial data mining algorithm.

3. The full-warehouse surface rolling construction quality evaluation method based on machine learning according to claim 1, characterized in that The specific steps for obtaining sample data in the rolling experiment in the step S3 are as follows: S31. Conduct an earth-rock filling rolling experiment before the filling and rolling work, use the real-time monitoring system to obtain the main rolling parameter information during the construction process, including the number of rolling passes and the rolling speed, and arrange a certain number of sampling points to collect the data of the compaction quality indicators by the additional mass method; S32. Select an area for the rolling experiment. During the experiment, collect the data of the compaction quality indicators under a certain number of different rolling speeds and rolling passes, and locate the corresponding rolling speed, number of rolling passes of each sampling data and the settlement value information obtained through the step S2, and obtain the sample data set of this method.

4. The full-warehouse surface rolling construction quality evaluation method based on machine learning according to claim 1, characterized in that The specific steps for establishing the construction quality evaluation model of the rolling construction surface based on machine learning in the step S4 are as follows: S41. Adopt the common regression algorithms in machine learning, use the sinking value, number of rolling passes, and rolling speed as inputs, and the compaction quality index as the output, and train the regression model respectively; conduct cross-validation, that is, divide the sample data set multiple times, use a part as the training set to train the model, and the other part as the test set, and calculate multiple times in this way to finally evaluate the model effect; S42. Through the training results and cross-validation results, select the regression algorithm with the best comprehensive fitting effect and cross-validation stability as the specific machine learning algorithm for estimating the rolling construction quality of the entire warehouse surface in this project, and use the sample data to obtain the rolling construction quality estimation model for the entire warehouse surface.

5. A full-warehouse surface rolling construction quality assessment method based on machine learning according to claim 1, characterized in that In step S5, based on the data of the rolling test, the quality control requirements in the filling and rolling construction of this project are determined as follows where N is the number of rolling passes, N min is the minimum number of rolling passes, N max is the maximum number of rolling passes, v is the rolling speed, v min is the minimum rolling speed, v max is the maximum rolling speed, Z is the settlement value, Z lim is the limit value of the settlement value, Q is the compaction quality index, Q min is the minimum compaction quality index; therefore, the final quality assessment criterion for the rolling construction of the entire warehouse floor is Where, S all refers to the total area of the full warehouse surface construction, and S(*) refers to the sum of the areas that meet the requirements for a certain index on the full warehouse surface. is the qualified rate statistically calculated for the full warehouse surface, and p0 refers to the specified qualified rate limit according to the actual situation.

6. The full-warehouse surface rolling construction quality evaluation method based on machine learning according to claim 5, wherein During the construction process, the rolling construction quality assessment of the entire warehouse surface based on machine learning includes the following steps: S51. During the construction process, use the monitoring data of the real-time monitoring system to statistically obtain the rolling parameter information such as the number of rolling passes and rolling speed at any position on the entire warehouse surface. At the same time, construct a real-time high-precision model of the rolling warehouse surface to calculate the sinking value at any position on the entire warehouse surface, and take the sinking value, number of rolling passes, and rolling speed as input quantities and bring them into the model in step S2 to obtain the estimated value of the compaction quality index at any position on the entire warehouse surface; S52. Statistically calculate the pass rates of the rolling parameters and compaction quality indexes on the entire warehouse surface according to the quality control requirements in step S5, and evaluate the construction quality of the entire warehouse surface according to the rolling construction quality assessment criteria in step S5.

7. A full-face rolling construction quality evaluation method based on machine learning according to claim 1, characterized in that: The improved radial basis function neural network algorithm in step S1 adds a robust algorithm on the basis of the radial basis function neural network algorithm.

Images