Earth pressure balance shield improved muck classification evaluation method and system

Abstract

The invention relates to the technical field of earth pressure balance shield construction, in particular to an earth pressure balance shield improved muck classification evaluation method and system. According to the method, unsupervised clustering analysis is carried out by comprehensively considering multiple target judgment indexes related to muck improvement, and an improved muck classification system comprehensively considering the influence of multiple factors is established. The index interval value corresponding to each improved muck category is determined through probability density analysis, the improved muck with different physical and mechanical properties can be effectively distinguished, a clear quantitative reference is provided for subsequent muck improvement effect classification prediction based on improved muck classification, and the accuracy of soil pressure balance shield improved muck classification evaluation is improved. And finally, constructing a classification prediction model by using a limit gradient lifting algorithm, predicting the improved muck category to which the improved muck obtained by the muck improvement scheme to be detected belongs, and realizing accurate classification evaluation of the muck improvement effect.

Description

Technical Field

[0001] This invention relates to the field of earth pressure balance shield tunneling technology, and in particular to a method and system for classifying and evaluating improved slag and soil in earth pressure balance shield tunneling. Background Technology

[0002] In shield tunneling projects, due to problems such as uneven particle distribution, high looseness, and poor self-stability of the strata, water, foaming agents, bentonite slurry, dispersants, and polymers are usually used to improve the shield excavated soil at the cutterhead face, soil chamber, and screw conveyor location. This is to give the discharged excavated soil good fluidity, suitable plasticity, low shear strength and cohesiveness, low permeability, and a certain degree of compressibility, thereby stabilizing the water and soil pressure at the shield face and ensuring smooth discharge of the excavated soil.

[0003] Regarding the physical and mechanical properties of modified slag soil after the addition of the aforementioned modifiers, scholars both domestically and internationally have conducted systematic research using various methods, including indoor experiments, field experiments, theoretical analysis, and numerical simulations. In terms of macroscopic mechanical properties, scholars have used indoor experimental methods such as atmospheric and pressurized rotary shear tests, direct shear tests, and permeability tests to determine the fluidity, shear strength, and permeability of the modified slag soil. They have also analyzed the influence of modifiers such as foaming agents and bentonite slurry on the rheological and plastic properties of the slag soil.

[0004] Although existing research has made some progress, most studies focus on a single type of geological formation encountered in a specific tunnel boring machine (TBM) project. Furthermore, most rely on empirical methods for soil improvement during TBM construction, often resulting in a trial-and-error phase of around 100 rings after the initial TBM launch, thus reducing construction efficiency. Simultaneously, these problems lead to a single standard for judging the effectiveness of soil improvement, requiring independent analysis of different soil properties and failing to comprehensively consider the influence of multiple factors, thus hindering accurate and efficient evaluation of soil improvement results. Summary of the Invention

[0005] The present invention aims to provide a method and system for classifying and evaluating the improved slag from earth pressure balance shield tunneling machines, so as to comprehensively consider the influence of various factors related to the slag from earth pressure balance shield tunneling machines to evaluate the effect of slag improvement and improve the accuracy of the evaluation of the effect of slag improvement.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for classifying and evaluating improved excavated soil from earth pressure balance shield tunneling machines, comprising the following steps: Based on the results of the pre-set soil improvement effect analysis test, several target judgment indicators were obtained; Obtain improved waste soil sample data, and perform unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories; Based on the improved waste soil sample data, obtain the category data corresponding to each of the improved waste soil categories; For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, thereby obtaining the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category. Based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, a classification prediction model is constructed according to the extreme gradient improvement algorithm. Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested; The target improvement parameter value is input into the classification prediction model, and then the predicted waste soil category corresponding to the waste soil improvement scheme to be tested is predicted among several improved waste soil categories, and the predicted waste soil category is used as the improved waste soil classification evaluation result.

[0007] The aforementioned method for classifying and evaluating modified excavated soil from earth pressure balance (EPB) shield tunneling projects establishes a comprehensive classification system for modified excavated soil by comprehensively considering multiple target judgment indicators related to excavated soil modification through unsupervised cluster analysis. Probability density analysis is used to determine the index intervals corresponding to each modified excavated soil category, effectively distinguishing modified excavated soil with different physical and mechanical properties. This provides a clear quantitative benchmark for subsequent classification and prediction of excavated soil modification effects based on the modified soil classification, improving the accuracy of the evaluation of modified excavated soil from EPB shield tunneling projects. Finally, a classification prediction model is constructed using the limit gradient lifting algorithm to predict the modified excavated soil category obtained from the tested modification scheme, achieving accurate classification and evaluation of the excavated soil modification effect.

[0008] This invention overcomes the shortcomings of existing technologies, such as the single judgment standard and the difficulty in comprehensively evaluating multiple factors. It significantly improves the accuracy and reliability of the classification and evaluation of modified soil in earth pressure balance shield tunneling, reduces trial and error in modification after shield launch, improves shield construction efficiency and reduces engineering risks, and provides effective technical support for soil modification decisions in earth pressure balance shield tunneling.

[0009] Furthermore, based on the results of the pre-set analysis test on the effect of soil improvement, several target judgment indicators are obtained, including: Based on the results of the pre-set soil improvement effect analysis test, several candidate judgment indicators were obtained. Using Spearman correlation analysis, for any candidate judgment index, the Spearman correlation between each of the remaining candidate judgment indices and the candidate judgment index is obtained. Several target judgment indicators are selected from several candidate judgment indicators, such that the Spearman correlation between the several target judgment indicators is less than a preset correlation threshold.

[0010] In this implementation, Spearman correlation analysis is used to screen candidate judgment indicators, effectively eliminating redundant indicators with high information overlap and multicollinearity. This ensures that the selected target judgment indicators can characterize the key physical and mechanical properties of the improved slag from different and independent dimensions, thus ensuring the relative independence between the final target judgment indicators.

[0011] The above implementation method not only simplifies the input data dimensions for subsequent unsupervised clustering analysis and classification prediction model construction, reducing the training complexity of the classification prediction model and thus improving its training efficiency, but also avoids the problem of low efficiency in evaluating the improvement effect of waste soil due to redundant indicator information. Furthermore, it avoids the overfitting problem of the classification prediction model caused by high correlation between indicators, improving the generalization ability of the subsequent classification prediction model and the accuracy and reliability of the final classification evaluation results.

[0012] Furthermore, the process involves acquiring improved waste soil sample data and performing unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories, including: In the improved slag soil sample data, several data samples are determined as several first center points, and a corresponding data cluster is matched for each first center point; Based on several first centroids, several rounds of cluster analysis are performed until all first centroids stop changing after any round of cluster analysis. Several improved waste soil categories are obtained based on several data clusters; Each round of cluster analysis includes: For any data sample, calculate the distance between each first center point and the data sample, and then, based on the distance between each first center point and the data sample, assign the data sample to a data cluster that meets a preset distance condition; For any data cluster, a second center point is obtained based on all the data samples in the data cluster, and then the first center point corresponding to the data cluster is updated based on the second center point.

[0013] In this implementation, by employing the aforementioned iteratively optimized unsupervised clustering analysis method, the inherent structural features and natural groupings of improved waste soil samples can be automatically discovered from multi-dimensional improved waste soil sample data without relying on pre-defined category labels. This method iteratively calculates the distance between samples and centroids, reassigns cluster affiliations, and updates centroids until they stabilize. This ensures that the properties of samples within the final improved waste soil categories are highly similar, automatically grouping waste soil samples with similar physical and mechanical properties into one category, forming several improved waste soil categories with significantly different improvement effects, thus ensuring the objectivity and accuracy of the classification results. Ultimately, this unsupervised clustering analysis provides an objective and reliable data foundation for subsequently establishing quantitative evaluation intervals that integrate multiple target judgment indicators for each category. This enhances the representational ability and discrimination accuracy of subsequent classification prediction models, further improving the accuracy and reliability of the predicted waste soil category corresponding to the tested waste soil improvement scheme, and using the predicted waste soil category as the evaluation result for improved waste soil classification.

[0014] Further, for any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data to obtain the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category, including: For any of the aforementioned improved waste soil categories: Obtain the indicator dataset corresponding to each target judgment indicator from the category data; Based on the indicator dataset corresponding to each of the target judgment indicators, obtain the probability density function corresponding to each of the target judgment indicators; Based on the probability density function corresponding to each of the target judgment indicators, obtain the cumulative distribution function corresponding to each of the target judgment indicators; Based on the cumulative distribution function corresponding to each target judgment indicator, the index interval value corresponding to each target judgment indicator is obtained, which serves as several index interval values ​​corresponding to the improved slag category.

[0015] In this implementation, the specific numerical ranges of each category of improved waste soil are analyzed based on the target judgment indicators selected after Spearman correlation analysis. Using analysis methods based on probability density functions and cumulative distribution functions, the distribution patterns of the target judgment indicator data in each category of improved waste soil are statistically analyzed, and finally, statistically significant quantitative intervals for each target judgment indicator data are determined.

[0016] This implementation method reflects the distribution pattern of target judgment index values ​​for each category of improved waste soil data through probabilistic statistical means. Compared with the method of directly using the maximum and minimum values ​​of data samples to determine the numerical range, this method can effectively eliminate the interference of abnormal data points, thereby ensuring that the obtained index range values ​​can more accurately cover the typical characteristics of this type of improved waste soil. This provides a clear and reliable range of classification and discrimination parameters for subsequent classification and prediction models, significantly improving the interpretability and accuracy of the model when making category predictions, and thus ensuring the accuracy and engineering practicality of the final comprehensive classification evaluation results of waste soil improvement effects.

[0017] Furthermore, the step of constructing a classification prediction model based on several improved waste soil categories and several index interval values ​​corresponding to each improved waste soil category, according to the extreme gradient boosting algorithm, includes: Based on the preset historical waste soil improvement schemes, several candidate improvement parameters are obtained; Based on the improved slag sample data, several decision trees are generated using several candidate improvement parameters as input features and several improved slag categories as output categories. A random forest model is constructed based on several of the aforementioned decision trees; Obtain the Gini index of each candidate improved parameter in the random forest model; Among the candidate improvement parameters, select several target improvement parameters that correspond to the Gini index and satisfy a preset Gini index threshold. According to the extreme gradient boosting algorithm, a classification prediction model is trained using several target improvement parameters as input features and several improved slag soil categories as prediction categories.

[0018] In this implementation, for the various parameters involved in the waste soil improvement scheme, the Gini index of each candidate improvement parameter is first calculated using a random forest model and feature screening is performed. This objectively and quantitatively evaluates the correlation between each parameter and the target judgment indicators of each improved waste soil category, identifies the core parameter set with the highest correlation to the target judgment indicators of each improved waste soil category, eliminates redundant or irrelevant feature parameters, significantly reduces the dimensionality and noise of the subsequent classification prediction model input, and simplifies the model structure.

[0019] Based on this, the classification prediction model is trained using the extreme gradient boosting algorithm, based on the selected target improvement parameters. This modeling strategy of first selecting features and then training the model not only optimizes the training and computational efficiency of the model but also improves the accuracy and stability of the final classification prediction model. This allows it to be reliably applied to quickly and accurately predict the improved waste soil category for unknown waste soil improvement schemes, and to accurately and efficiently evaluate the improvement effect based on the waste soil category.

[0020] A second aspect of the present invention provides a classification and evaluation system for improved excavated soil from earth pressure balance shield tunneling machines, comprising: The judgment index acquisition module is used to acquire several target judgment indicators based on the preset test results of the soil improvement effect analysis. The improved waste soil classification module is used to acquire improved waste soil sample data and perform unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories. The improved waste soil category numerical analysis module is used to perform the following steps: Based on the improved waste soil sample data, obtain the category data corresponding to each of the improved waste soil categories; For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, thereby obtaining the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category. The classification prediction model construction module is used to construct a classification prediction model based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, according to the extreme gradient improvement algorithm. The waste soil improvement scheme classification and evaluation module is used to perform the following steps: Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested; The target improvement parameter value is input into the classification prediction model, and then the predicted waste soil category corresponding to the waste soil improvement scheme to be tested is predicted among several improved waste soil categories, and the predicted waste soil category is used as the improved waste soil classification evaluation result.

[0021] Furthermore, based on the results of the pre-set analysis test on the effect of soil improvement, several target judgment indicators are obtained, including: Based on the results of the pre-set soil improvement effect analysis test, several candidate judgment indicators were obtained. Using Spearman correlation analysis, for any candidate judgment index, the Spearman correlation between each of the remaining candidate judgment indices and the candidate judgment index is obtained. Several target judgment indicators are selected from several candidate judgment indicators, such that the Spearman correlation between the several target judgment indicators is less than a preset correlation threshold.

[0022] Furthermore, the process involves acquiring improved waste soil sample data and performing unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories, including: In the improved slag soil sample data, several data samples are determined as several first center points, and a corresponding data cluster is matched for each first center point; Based on several first centroids, several rounds of cluster analysis are performed until all first centroids stop changing after any round of cluster analysis. Several improved waste soil categories are obtained based on several data clusters; Each round of cluster analysis includes: For any data sample, calculate the distance between each first center point and the data sample, and then, based on the distance between each first center point and the data sample, assign the data sample to a data cluster that meets a preset distance condition; For any data cluster, a second center point is obtained based on all the data samples in the data cluster, and then the first center point corresponding to the data cluster is updated based on the second center point.

[0023] Furthermore, for any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data to obtain the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category, including: For any of the aforementioned improved waste soil categories: Obtain the indicator dataset corresponding to each target judgment indicator from the category data; Based on the indicator dataset corresponding to each of the target judgment indicators, obtain the probability density function corresponding to each of the target judgment indicators; Based on the probability density function corresponding to each of the target judgment indicators, obtain the cumulative distribution function corresponding to each of the target judgment indicators; Based on the cumulative distribution function corresponding to each target judgment indicator, the index interval value corresponding to each target judgment indicator is obtained, which serves as several index interval values ​​corresponding to the improved slag category.

[0024] Furthermore, the step of constructing a classification prediction model based on several improved waste soil categories and several index interval values ​​corresponding to each improved waste soil category, according to the extreme gradient boosting algorithm, includes: Based on the preset historical waste soil improvement schemes, several candidate improvement parameters are obtained; Based on the improved slag sample data, several decision trees are generated using several candidate improvement parameters as input features and several improved slag categories as output categories. A random forest model is constructed based on several of the aforementioned decision trees; Obtain the Gini index of each candidate improved parameter in the random forest model; Among the candidate improvement parameters, select several target improvement parameters that correspond to the Gini index and satisfy a preset Gini index threshold. According to the extreme gradient boosting algorithm, a classification prediction model is trained using several target improvement parameters as input features and several improved slag soil categories as prediction categories. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating a method for classifying and evaluating improved slag from earth pressure balance shield tunneling, as provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the results of a Spearman correlation analysis provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a clustering analysis result provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the probability density analysis results of yield shear strength provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of a collapse probability density analysis result provided by an embodiment of the present invention; Figure 6 This is a schematic diagram of a compression probability density analysis result provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of a random forest algorithm provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of the structure of an earth pressure balance shield tunneling improved slag classification and evaluation system provided in an embodiment of the present invention. Detailed Implementation

[0026] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that the following detailed descriptions are exemplary and intended to provide further detailed explanation of the invention. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used herein in the specification is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings are used to distinguish different objects, not to describe a particular order.

[0027] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0028] Before describing this application in detail with reference to the accompanying drawings and embodiments, the terms and application scenarios involved in this application will first be explained.

[0029] The present invention aims to provide a method and system for classifying and evaluating the improved slag from earth pressure balance shield tunneling machines, so as to comprehensively consider the influence of various factors related to the slag from earth pressure balance shield tunneling machines to evaluate the effect of slag improvement and improve the accuracy of the evaluation of the effect of slag improvement.

[0030] Please refer to Figure 1 To achieve the above objectives, the first embodiment of the present invention provides a method for classifying and evaluating improved excavated soil from earth pressure balance shield tunneling, comprising the following steps: S101. Based on the results of the pre-set analysis test on the effect of soil improvement, obtain several target judgment indicators; S102. Obtain improved slag sample data and perform unsupervised cluster analysis on the improved slag sample data to obtain several improved slag categories. S103. Based on the improved slag sample data, obtain the category data corresponding to each of the improved slag categories; S104. For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, and then the index interval value corresponding to each of the target judgment indicators is obtained as several index interval values ​​corresponding to the improved waste soil category. S105. Based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, a classification prediction model is constructed according to the extreme gradient boosting algorithm. S106. Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested. S107. Input the target improvement parameter value into the classification prediction model, and then predict the predicted waste soil category corresponding to the waste soil improvement scheme in several improved waste soil categories, and use the predicted waste soil category as the improved waste soil classification evaluation result.

[0031] The aforementioned method for classifying and evaluating modified excavated soil from earth pressure balance (EPB) shield tunneling projects establishes a comprehensive classification system for modified excavated soil by comprehensively considering multiple target judgment indicators related to excavated soil modification through unsupervised cluster analysis. Probability density analysis is used to determine the index intervals corresponding to each modified excavated soil category, effectively distinguishing modified excavated soil with different physical and mechanical properties. This provides a clear quantitative benchmark for subsequent classification and prediction of excavated soil modification effects based on the modified soil classification, improving the accuracy of the evaluation of modified excavated soil from EPB shield tunneling projects. Finally, a classification prediction model is constructed using the limit gradient lifting algorithm to predict the modified excavated soil category obtained from the tested modification scheme, achieving accurate classification and evaluation of the excavated soil modification effect.

[0032] This invention overcomes the shortcomings of existing technologies, such as the single judgment standard and the difficulty in comprehensively evaluating multiple factors. It significantly improves the accuracy and reliability of the classification and evaluation of modified soil in earth pressure balance shield tunneling, reduces trial and error in modification after shield launch, improves shield construction efficiency and reduces engineering risks, and provides effective technical support for soil modification decisions in earth pressure balance shield tunneling.

[0033] Furthermore, in step S101, based on the results of the pre-set analysis test on the effect of soil improvement, several target judgment indicators are obtained, including: Based on the results of the pre-set soil improvement effect analysis test, several candidate judgment indicators were obtained. Using Spearman correlation analysis, for any candidate judgment index, the Spearman correlation between each of the remaining candidate judgment indices and the candidate judgment index is obtained. Several target judgment indicators are selected from several candidate judgment indicators, such that the Spearman correlation between the several target judgment indicators is less than a preset correlation threshold.

[0034] In a specific embodiment, step S101 first obtains multiple candidate judgment indicators such as consistency coefficient, yield shear strength, compression ratio, and slump based on the test results and analysis results of the improved slag soil.

[0035] Due to the limitations of different testing methods for various parameters, it is impossible to quickly and accurately measure all indicators in actual construction. To explore the mutual influence among the indicators for judging the improvement effect, the properties of the improved soil with high correlation are merged and screened to achieve rapid judgment of the soil improvement effect.

[0036] To overcome the limitations of quickly and accurately measuring all indicators in actual construction, Spearman correlation analysis was used to calculate the correlation coefficients between candidate indicators, thereby uncovering the mutual influence between indicators and screening out parameters with low correlation that can independently characterize the key properties of improved soil. The Spearman correlation calculation method is shown in the following formula: ; Where n refers to the number of candidate judgment indicator data, d i This represents the difference in rank between the observed values ​​of the two candidate indicators. The calculated coefficient ρ ranges from [-1, 1]. The closer its absolute value is to 1, the stronger the monotonic correlation between the two variables; the closer its absolute value is to 0, the weaker the monotonic correlation.

[0037] Correlation analysis results as follows Figure 2 As shown, the larger the absolute value of the correlation coefficient, the higher the correlation between the two. Therefore, it can be simplified and the parameter with a lower degree of correlation can be selected as the indicator for soil improvement.

[0038] The analysis results show a significant correlation between the consistency coefficient and yield shear strength of the improved slag soil. Considering the greater convenience and universality of the yield shear strength test method, it was chosen as one of the target evaluation indicators. Slump showed no significant correlation with other indicators and was therefore directly selected as one of the target evaluation indicators. The compressibility of the improved slag soil under different vertical loads showed a significant correlation, but no significant correlation with other parameters. Considering that the soil chamber pressure in earth pressure balance shield tunneling is typically less than 2 bar, the compressibility at a vertical load of 200 kPa was selected as one of the target evaluation indicators to ensure good compressive strength of the improved slag soil.

[0039] Ultimately, after screening, three target indicators were determined to be used to evaluate the effect of improved slag soil: yield shear strength, slump, and compression ratio under a vertical load of 200 kPa.

[0040] In this embodiment, the Spearman correlation analysis method is used to screen candidate judgment indicators, effectively eliminating redundant indicators with high information overlap and multicollinearity. This ensures that the selected target judgment indicators can characterize the key physical and mechanical properties of the improved slag from different and independent dimensions, thus ensuring the relative independence between the final target judgment indicators.

[0041] The above embodiments not only simplify the input data dimensions for subsequent unsupervised clustering analysis and classification prediction model construction, reducing the training complexity of the classification prediction model and thus improving its training efficiency, but also avoid the problem of low efficiency in evaluating the improvement effect of waste soil due to redundant indicator information. Furthermore, they avoid the overfitting problem of the classification prediction model caused by high correlation between indicators, improving the generalization ability of the subsequent classification prediction model and the accuracy and reliability of the final classification evaluation results.

[0042] Further, step S102 involves acquiring improved waste soil sample data and performing unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories, including: In the improved slag soil sample data, several data samples are determined as several first center points, and a corresponding data cluster is matched for each first center point; Based on several first centroids, several rounds of cluster analysis are performed until all first centroids stop changing after any round of cluster analysis. Several improved waste soil categories are obtained based on several data clusters; Each round of cluster analysis includes: For any data sample, calculate the distance between each first center point and the data sample, and then, based on the distance between each first center point and the data sample, assign the data sample to a data cluster that meets a preset distance condition; For any data cluster, a second center point is obtained based on all the data samples in the data cluster, and then the first center point corresponding to the data cluster is updated based on the second center point.

[0043] Please refer to Figure 3In one specific embodiment, step S102 employs the K-means clustering algorithm to perform unsupervised clustering analysis on the improved slag sample data. Specifically, firstly, three data samples are randomly selected from all the improved slag sample data as initial first centroids, and a corresponding data cluster is matched for each first centroid. Then, the iterative process of clustering analysis begins. In each iteration, for each improved slag sample data, its distance to each first centroid is calculated, and the sample data is assigned to the data cluster corresponding to the nearest first centroid according to the principle of minimum distance. After all sample data have been assigned to the corresponding data clusters, for each data cluster, its centroid is recalculated based on all sample data within the cluster to obtain a new second centroid, and the original first centroid of the corresponding data cluster is updated with this second centroid. Next, it is determined whether the updated first centroid has changed from the original first centroid. If it has changed, the above iterative process of sample allocation and centroid update is repeated until all first centroids no longer change after iteration, at which point the clustering analysis converges. Finally, based on the three data clusters formed after convergence, three corresponding improved waste soil categories were obtained.

[0044] In this embodiment, by employing the aforementioned iteratively optimized unsupervised clustering analysis method, the inherent structural features and natural groupings of improved waste soil samples can be automatically discovered from multi-dimensional improved waste soil sample data without relying on pre-defined category labels. This method iteratively calculates the distance between samples and centroids, reassigns cluster affiliations, and updates centroids until they stabilize. This ensures that the properties of samples within the final improved waste soil categories are highly similar, automatically grouping waste soil samples with similar physical and mechanical properties into one category, forming several improved waste soil categories with significantly different improvement effects, thus ensuring the objectivity and accuracy of the classification results. Ultimately, this unsupervised clustering analysis provides an objective and reliable data foundation for subsequently establishing quantitative evaluation intervals that integrate multiple target judgment indicators for each category. This enhances the representational ability and discrimination accuracy of subsequent classification prediction models, further improving the accuracy and reliability of the predicted waste soil category corresponding to the tested waste soil improvement scheme, and using the predicted waste soil category as the evaluation result for improved waste soil classification.

[0045] Further, in step S104, for any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data to obtain the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category, including: For any of the aforementioned improved waste soil categories: Obtain the indicator dataset corresponding to each target judgment indicator from the category data; Based on the indicator dataset corresponding to each of the target judgment indicators, obtain the probability density function corresponding to each of the target judgment indicators; Based on the probability density function corresponding to each of the target judgment indicators, obtain the cumulative distribution function corresponding to each of the target judgment indicators; Based on the cumulative distribution function corresponding to each target judgment indicator, the index interval value corresponding to each target judgment indicator is obtained, which serves as several index interval values ​​corresponding to the improved slag category.

[0046] In one specific embodiment, probability density analysis is performed on the target judgment indicators of the three categories of improved slag soil obtained by K-means clustering, namely yield shear strength, slump, and compressibility, and the calculation method is shown in the following formula: ; Here, f(t) represents the probability density function of the target judgment indicator, used to describe the probability density distribution of this continuous random variable at different values. t is the integration variable. The integral symbol ∫ represents integration from negative infinity to a specific value x, and the result F(x) is the cumulative distribution function. The physical meaning of F(x) is the probability that the value of the target judgment indicator is less than or equal to a specific threshold x. Therefore, x here represents a preset or query threshold for the indicator.

[0047] By fitting the sample data of each target judgment indicator in each category of improved waste soil, its corresponding probability density function can be obtained. This function intuitively reflects the distribution pattern of the indicator data, such as whether it follows a normal distribution and the central tendency of the data. To obtain accurate indicator interval values, the cumulative distribution function needs to be further utilized. By integrating the probability density function to obtain the cumulative distribution function, the range of values ​​corresponding to the indicator data under any given probability threshold can be determined. For example, by setting probability thresholds of 5% and 95%, the corresponding indicator values ​​can be found on the cumulative distribution function, thereby determining a confidence interval containing 90% of the sample data.

[0048] Please refer to Figure 4 , Figure 5 and Figure 6The analysis results show that the yield shear strength of the improved slag soil in Category 3 is significantly higher than that in Categories 1 and 2, indicating that the shear strength improvement effect of the improved slag soil represented by Category 3 is not good. Regarding slump, the improved slump soil in Category 1 exhibits the highest slump and probability density, while the improved soil in Category 2 has the lowest slump and probability density. For compressibility, the distribution patterns of improved slag soil in Categories 1 and 3 are similar, but the probability density of improved slag soil in Category 1 in the range of compressibility greater than 3.8% / bar is higher than that in Category 3, indicating that improved slag soil in Category 1 has better compressive strength. In contrast, the compressibility of improved slag soil in Category 2 is significantly lower than that in Categories 1 and 3, and its distribution is uneven, lacking a clear normal distribution pattern.

[0049] Based on the probability density and box plot distribution patterns of yield shear strength, slump, and compression ratio of different types of improved slag soil, the improved slag soil is divided into three categories: 1, 2, and 3. The distribution range and characteristics of the evaluation indicators for each type of improved soil are shown in Table 1 below.

[0050] Table 1 Classification of Improved Slag and Soil Among them, Category 1 modified excavated soil is the best. This category of modified excavated soil has good fluidity, low shear strength, and sufficient compressive strength, which can balance the water and soil pressure at the tunnel face while ensuring smooth discharge of modified excavated soil. Category 2 modified excavated soil has a lower compressibility and slump than Category 1 modified excavated soil, which is prone to problems such as mud cake formation and poor soil discharge during shield tunneling. Category 3 modified excavated soil has a lower yield shear strength and slump than Category 1 modified excavated soil, indicating that Category 3 modified excavated soil is prone to causing engineering problems such as shield cutterhead wear and poor soil discharge.

[0051] In this embodiment, the specific numerical ranges of each category of improved waste soil are analyzed based on the target judgment indicators selected after Spearman correlation analysis. Using analysis methods based on probability density functions and cumulative distribution functions, the distribution patterns of the target judgment indicator data in each category of improved waste soil are statistically analyzed, and finally, statistically significant quantitative intervals for each target judgment indicator data are determined.

[0052] This embodiment uses probabilistic statistical methods to reflect the distribution pattern of target judgment index values ​​for each category of improved waste soil data. Compared with the method of directly using the maximum and minimum values ​​of data samples to determine the numerical range, this method can effectively eliminate the interference of abnormal data points, thereby ensuring that the obtained index range values ​​can more accurately cover the typical characteristics of this type of improved waste soil. This provides a clear and reliable range of classification and discrimination parameters for the subsequent classification and prediction model, significantly improving the interpretability and accuracy of the model when making category predictions, and thus ensuring the accuracy and engineering practicality of the final comprehensive classification evaluation results of waste soil improvement effects.

[0053] Further, step S105, which involves constructing a classification prediction model based on several improved waste soil categories and several index interval values ​​corresponding to each improved waste soil category, using the extreme gradient boosting algorithm, includes: Based on the preset historical waste soil improvement schemes, several candidate improvement parameters are obtained; Based on the improved slag sample data, several decision trees are generated using several candidate improvement parameters as input features and several improved slag categories as output categories. A random forest model is constructed based on several of the aforementioned decision trees; Obtain the Gini index of each candidate improved parameter in the random forest model; Among the candidate improvement parameters, select several target improvement parameters that correspond to the Gini index and satisfy a preset Gini index threshold. According to the extreme gradient boosting algorithm, a classification prediction model is trained using several target improvement parameters as input features and several improved slag soil categories as prediction categories.

[0054] Please refer to Figure 7 In one specific embodiment, the particle size d of the untreated sand is first obtained based on a preset historical slag improvement scheme. 10 Unimproved sand particle size d 30 And unimproved sand particle size d 60 Obtain bentonite slurry parameters, including bentonite slurry concentration, dynamic shear force, plastic viscosity, and bentonite slurry injection ratio; obtain foaming agent parameters, including foam concentration and foam injection ratio, as well as clay content, etc., as candidate improvement parameters.

[0055] Subsequently, using these candidate improvement parameters as input features and the determined improved waste soil category as the output category, a random forest model is constructed to analyze the feature importance of each parameter. In this random forest model, the Gini index of each candidate improvement parameter is calculated to quantify its influence on the classification results; parameters with larger Gini indices are considered more important. Based on the feature importance analysis results, key parameters whose Gini indices meet a preset threshold are selected as target improvement parameters.

[0056] It's important to note that the Gini index is a metric used in decision tree algorithms to measure the impurity or uncertainty of a dataset. A lower Gini index indicates that samples in the dataset tend to belong to the same category, meaning higher purity; conversely, a higher Gini index indicates a more mixed distribution of sample categories, meaning higher impurity. In constructing a decision tree, the core objective of the algorithm is to select an optimal feature at each node for splitting, minimizing the weighted Gini index of the resulting child nodes, thereby maximizing the reduction of impurity. In the random forest model, this principle is used to evaluate the feature importance of each candidate improvement parameter. Specifically, the importance of a parameter is determined by calculating the average reduction in Gini index it brings when used as a splitting feature across all decision trees in the forest. Therefore, a candidate improvement parameter with a larger Gini index reduction indicates a greater contribution to distinguishing different improved waste soil categories, meaning higher feature importance.

[0057] In this embodiment, the final selected parameters are the bentonite slurry injection ratio, bentonite slurry concentration, and unmodified sand particle size d. 10 The clay content and foam injection ratio were selected as target improvement parameters. Finally, the Extreme Gradient Boosting (XGBoost) algorithm was used, with the selected target improvement parameters as input features and the improved soil type as the prediction target, to train the classification prediction model, thereby constructing a final model that can accurately predict the soil improvement effect.

[0058] In this embodiment, for the various parameters involved in the slag improvement scheme, the Gini index of each candidate improvement parameter is first calculated using a random forest model and feature screening is performed. This objectively and quantitatively evaluates the correlation between each parameter and the target judgment indicators of each improved slag category, identifies the core parameter set with the highest correlation to the target judgment indicators of each improved slag category, eliminates redundant or irrelevant feature parameters, significantly reduces the dimensionality and noise of the subsequent classification prediction model input, and simplifies the model structure.

[0059] Based on this, the classification prediction model is trained using the extreme gradient boosting algorithm, based on the selected target improvement parameters. This modeling strategy of first selecting features and then training the model not only optimizes the training and computational efficiency of the model but also improves the accuracy and stability of the final classification prediction model. This allows it to be reliably applied to quickly and accurately predict the improved waste soil category for unknown waste soil improvement schemes, and to accurately and efficiently evaluate the improvement effect based on the waste soil category.

[0060] Please refer to Figure 8 The second embodiment of the present invention provides a classification and evaluation system for improved slag from earth pressure balance shield tunneling, comprising: The judgment index acquisition module 100 is used to acquire several target judgment indicators based on the preset test results of the soil improvement effect analysis. The improved waste soil classification module 200 is used to acquire improved waste soil sample data and perform unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories. The improved waste soil category numerical analysis module 300 is used to perform the following steps: Based on the improved waste soil sample data, obtain the category data corresponding to each of the improved waste soil categories; For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, thereby obtaining the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category. The classification prediction model construction module 400 is used to construct a classification prediction model based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, according to the extreme gradient improvement algorithm. The waste soil improvement scheme classification and evaluation module 500 is used to perform the following steps: Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested; The target improvement parameter value is input into the classification prediction model, and then the predicted waste soil category corresponding to the waste soil improvement scheme to be tested is predicted among several improved waste soil categories, and the predicted waste soil category is used as the improved waste soil classification evaluation result.

[0061] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0062] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; however, any combination of these technical features that does not contradict each other should be considered within the scope of this specification.

[0063] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the concept of this application, and these improvements and substitutions should also be considered within the scope of protection of this invention. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A method for classifying and evaluating improved excavated soil from earth pressure balance shield tunneling machines, characterized in that, include: Based on the results of the pre-set soil improvement effect analysis test, several target judgment indicators were obtained; Obtain improved waste soil sample data, and perform unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories; Based on the improved waste soil sample data, obtain the category data corresponding to each of the improved waste soil categories; For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, thereby obtaining the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category. Based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, a classification prediction model is constructed according to the extreme gradient boosting algorithm. Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested; The target improvement parameter value is input into the classification prediction model, and then the predicted waste soil category corresponding to the waste soil improvement scheme to be tested is predicted among several improved waste soil categories, and the predicted waste soil category is used as the improved waste soil classification evaluation result.

2. The method for classifying and evaluating improved slag from earth pressure balance shield tunneling as described in claim 1, characterized in that, Based on the results of the pre-set analysis test on the effect of soil improvement, several target judgment indicators are obtained, including: Based on the results of the pre-set soil improvement effect analysis test, several candidate judgment indicators were obtained. Using Spearman correlation analysis, for any candidate judgment index, the Spearman correlation between each of the remaining candidate judgment indices and the candidate judgment index is obtained. Several target judgment indicators are selected from several candidate judgment indicators, such that the Spearman correlation between the several target judgment indicators is less than a preset correlation threshold.

3. The method for classifying and evaluating improved excavated soil from earth pressure balance shield tunneling machines according to claim 1, characterized in that, The process involves acquiring improved waste soil sample data and performing unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories, including: In the improved slag soil sample data, several data samples are determined as several first center points, and a corresponding data cluster is matched for each first center point; Based on several first centroids, several rounds of cluster analysis are performed until all first centroids stop changing after any round of cluster analysis. Several improved waste soil categories are obtained based on several data clusters; Each round of cluster analysis includes: For any data sample, calculate the distance between each first center point and the data sample, and then, based on the distance between each first center point and the data sample, assign the data sample to a data cluster that meets a preset distance condition; For any data cluster, a second center point is obtained based on all the data samples in the data cluster, and then the first center point corresponding to the data cluster is updated based on the second center point.

4. The method for classifying and evaluating improved slag from earth pressure balance shield tunneling machines according to claim 1, characterized in that, For any of the improved waste soil categories, probability density analysis is performed on each target judgment indicator based on the corresponding category data to obtain the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category, including: For any of the aforementioned improved waste soil categories: Obtain the indicator dataset corresponding to each target judgment indicator from the category data; Based on the indicator dataset corresponding to each of the target judgment indicators, obtain the probability density function corresponding to each of the target judgment indicators; Based on the probability density function corresponding to each target judgment indicator, obtain the cumulative distribution function corresponding to each target judgment indicator; Based on the cumulative distribution function corresponding to each target judgment indicator, the index interval value corresponding to each target judgment indicator is obtained, which serves as several index interval values ​​corresponding to the improved slag category.

5. The method for classifying and evaluating improved slag from earth pressure balance shield tunneling machines according to claim 1, characterized in that, The classification prediction model is constructed based on several improved waste soil categories and several index interval values ​​corresponding to each improved waste soil category, using the extreme gradient boosting algorithm, including: Based on the preset historical waste soil improvement schemes, several candidate improvement parameters are obtained; Based on the improved slag sample data, several decision trees are generated using several candidate improvement parameters as input features and several improved slag categories as output categories. A random forest model is constructed based on several of the aforementioned decision trees; Obtain the Gini index of each candidate improved parameter in the random forest model; Among the candidate improvement parameters, select several target improvement parameters that correspond to the Gini index and satisfy a preset Gini index threshold. According to the extreme gradient boosting algorithm, a classification prediction model is trained using several target improvement parameters as input features and several improved slag soil categories as prediction categories.

6. A classification and evaluation system for improved excavated soil from earth pressure balance shield tunneling machines, characterized in that, include: The judgment index acquisition module is used to acquire several target judgment indicators based on the preset test results of the soil improvement effect analysis. The improved waste soil classification module is used to acquire improved waste soil sample data and perform unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories. The improved waste soil category numerical analysis module is used to perform the following steps: Based on the improved waste soil sample data, obtain the category data corresponding to each of the improved waste soil categories; For any of the improved waste soil categories, probability density analysis is performed on each of the target judgment indicators based on the corresponding category data, thereby obtaining the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category. The classification prediction model construction module is used to construct a classification prediction model based on several improved soil categories and several index interval values ​​corresponding to each improved soil category, according to the extreme gradient improvement algorithm. The waste soil improvement scheme classification and evaluation module is used to perform the following steps: Determine the soil improvement scheme to be tested, and obtain the target improvement parameter values ​​of the soil improvement scheme to be tested; The target improvement parameter value is input into the classification prediction model, and then the predicted waste soil category corresponding to the waste soil improvement scheme to be tested is predicted among several improved waste soil categories, and the predicted waste soil category is used as the improved waste soil classification evaluation result.

7. The earth pressure balance shield tunneling spoil classification and evaluation system according to claim 6, characterized in that, Based on the results of the pre-set analysis test on the effect of soil improvement, several target judgment indicators are obtained, including: Based on the results of the pre-set soil improvement effect analysis test, several candidate judgment indicators were obtained. Using Spearman correlation analysis, for any candidate judgment index, the Spearman correlation between each of the remaining candidate judgment indices and the candidate judgment index is obtained. Several target judgment indicators are selected from several candidate judgment indicators, such that the Spearman correlation between the several target judgment indicators is less than a preset correlation threshold.

8. The earth pressure balance shield tunneling spoil classification and evaluation system according to claim 6, characterized in that, The process involves acquiring improved waste soil sample data and performing unsupervised cluster analysis on the improved waste soil sample data to obtain several improved waste soil categories, including: In the improved slag soil sample data, several data samples are determined as several first center points, and a corresponding data cluster is matched for each first center point; Based on several first centroids, several rounds of cluster analysis are performed until all first centroids stop changing after any round of cluster analysis. Several improved waste soil categories are obtained based on several data clusters; Each round of cluster analysis includes: For any data sample, calculate the distance between each first center point and the data sample, and then, based on the distance between each first center point and the data sample, assign the data sample to a data cluster that meets a preset distance condition; For any data cluster, a second center point is obtained based on all the data samples in the data cluster, and then the first center point corresponding to the data cluster is updated based on the second center point.

9. The earth pressure balance shield tunneling spoil classification and evaluation system according to claim 6, characterized in that, For any of the improved waste soil categories, probability density analysis is performed on each target judgment indicator based on the corresponding category data to obtain the index interval value corresponding to each target judgment indicator, which serves as several index interval values ​​corresponding to the improved waste soil category, including: For any of the aforementioned improved waste soil categories: Obtain the indicator dataset corresponding to each target judgment indicator from the category data; Based on the indicator dataset corresponding to each of the target judgment indicators, obtain the probability density function corresponding to each of the target judgment indicators; Based on the probability density function corresponding to each target judgment indicator, obtain the cumulative distribution function corresponding to each target judgment indicator; Based on the cumulative distribution function corresponding to each target judgment indicator, the index interval value corresponding to each target judgment indicator is obtained, which serves as several index interval values ​​corresponding to the improved slag category.

10. A classification and evaluation system for improved slag from earth pressure balance shield tunneling machines according to claim 6, characterized in that, The classification prediction model is constructed based on several improved waste soil categories and several index interval values ​​corresponding to each improved waste soil category, using the extreme gradient boosting algorithm, including: Based on the preset historical waste soil improvement schemes, several candidate improvement parameters are obtained; Based on the improved slag sample data, several decision trees are generated using several candidate improvement parameters as input features and several improved slag categories as output categories. A random forest model is constructed based on several of the aforementioned decision trees; Obtain the Gini index of each candidate improved parameter in the random forest model; Among the candidate improvement parameters, select several target improvement parameters that correspond to the Gini index and satisfy a preset Gini index threshold. According to the extreme gradient boosting algorithm, a classification prediction model is trained using several target improvement parameters as input features and several improved slag soil categories as prediction categories.

Images